TBPN Live - NVIDIA Earnings Breakdown | Doug O’Laughlin, Ajay Agarwal, Koen Bok & Jorn Van Dijk, Hussein Fazal, Oisin Hanrahan, Jon Callaghan, Shane Hegde

Episode Date: August 28, 2025

(00:11) - NVIDIA Earnings Breakdown (28:43) - Doug O’Laughlin is the President of SemiAnalysis, an independent research firm focused on semiconductors and AI. He specializes in semiconductor strategy, market intelligence, and competitive analysis, regularly guiding investors, technology companies, and policymakers with his insights. (01:12:05) - Timeline Reactions (01:58:28) - Ajay Agarwal, a Partner at Bain Capital Ventures, has been with the firm since 2003, focusing on early-stage application software and SaaS investments. In the conversation, he discusses his journey from leading sales and marketing at Trilogy, where he grew annual revenues to $300 million, to his current role at Bain Capital Ventures, emphasizing the importance of software innovation and network effects in building successful companies. (02:30:55) - Koen Bok & Jorn Van Dijk, CEO & Co-Founder of Framer, a professional web design platform, discusses the company's recent Series D funding announcement, highlighting their mission to enable designers to ship websites without relying on developers. (02:41:29) - Hussein Fazal, co-founder and CEO of Super.com, discusses the company's rebranding from SnapTravel to Super.com, emphasizing their focus on providing a membership program that offers customers savings on hotels, gas, insurance, and more. He highlights the company's growth, surpassing $200 million in annualized revenue, and the challenges faced during the acquisition of the Super.com domain, which involved intense negotiations and a significant investment. Fazal also shares insights into their customer acquisition strategies, emphasizing the importance of product-specific channels and the role of AI in personalizing user experiences. (02:50:58) - Oisin Hanrahan, co-founder and CEO of Keychain, previously co-founded Handy, a home services platform acquired by Angi, where he later served as CEO. In the conversation, he discusses Keychain's mission to streamline the consumer packaged goods (CPG) supply chain by connecting brands and retailers with suitable manufacturing partners through an AI-powered platform. He highlights the platform's success, noting that eight of the top ten U.S. retailers use Keychain, and mentions a recent $30 million funding round, bringing their total raised to $60 million. (03:01:07) - Jon Callaghan, co-founder of True Ventures and former Chairman of the National Venture Capital Association, has been a venture capitalist since 1991, with a background in founding three companies. He discusses the evolution of venture capital, emphasizing the shift towards capital efficiency and the ability of founders to achieve more with fewer resources. Callaghan highlights the unprecedented opportunities in the current market, particularly in AI, and underscores the importance of empowering entrepreneurs to take bold risks without fear of failure. (03:16:08) - Shane Hegde, CEO and co-founder of Air, discusses how Air serves as a system of record for creative work, enabling teams to efficiently manage and automate their creative operations. He emphasizes the importance of understanding and meeting the expectations of creative professionals by delivering a product that aligns with their needs. Additionally, Hegde highlights Air's strategic focus on content marketing and culture-led growth to effectively reach and engage their target audience. TBPN.com is made possible by: Ramp - https://ramp.comFigma - https://figma.comVanta - https://vanta.comLinear - https://linear.appEight Sleep - https://eightsleep.com/tbpnWander - https://wander.com/tbpnPublic - https://public.comAdQuick - https://adquick.comBezel - https://getbezel.com Numeral - https://www.numeralhq.comPolymarket - https://polymarket.comAttio - https://attio.com/tbpnFin - https://fin.ai/tbpnGraphite - https://graphite.devRestream - https://restream.ioProfound - https://tryprofound.comJulius AI - https://julius.aiFollow TBPN: https://TBPN.comhttps://x.com/tbpnhttps://open.spotify.com/show/2L6WMqY3GUPCGBD0dX6p00?si=674252d53acf4231https://podcasts.apple.com/us/podcast/technology-brothers/id1772360235https://www.youtube.com/@TBPNLive

You're watching TVPN. It is Thursday, August 28, 2025. We are live from the TBPN Ultradome, the Temple of Technology, the Fortress of Finance, the capital of capital. Invidia earnings were yesterday. They hit consensus estimates, both on earnings per share and revenue, but the stock sold off in after-hours trading. Wasn't bullish enough. The whispers were that maybe something even crazier was going to. to happen. But this business has been on a tear. The quarterly revenue chart is absolutely

insane. One of the craziest charts I've ever seen in business. And the pricing power is unrivaled. The gross profit margin per fiscal year is insane. They're up in 75%. So, fantastic business overall. But of course, the question is, where does NVIDIA go from here? Where is the business? We can read from the Washington Journal that kind of give you the headline. It's rough when a business announces a $60 billion buyback and everybody's screaming, bearish, that's bearish.

Yeah. For giving problems back to shareholders. Yeah, we have an extra $60 billion sitting around because we're doing so well. And everyone's like, boo. It's crazy. Grow faster. NVIDIA set fresh sales records on Wednesday

as the world's most valuable company publicly traded company continued to capitalize on strong demand for AI computing power. Sales hit $46.7 billion up 56% from the year earlier in line with revenue estimates from analysts. Revenue from the important data center segment, which includes sales of the company's most powerful chips used to train and refine artificial intelligence models rose 56% to $41.1 billion.

and this was slightly lower than what analysts expected at 41.3. And so that's probably what was driving the market moving in the after hours. Because that should be the strongest place of growth is the data center business. Whereas, you know, if gaming was selling off or they have some automotive businesses, they have a variety of other products, the vast majority, I think 90% of the revenue almost is from data center. And obviously that's been the huge driver and the huge narrative. We're actually going to pull up a quick graphic right now of Jensen for you guys in case you haven't seen it. Oh, yes.

Yesterday. Looks great. Going head to head with himself. Yes, yes, yes. He's really, he, he, again. Market caps 4.4 trillion. Is that right? Hopefully.

There we go. Yep. Quarterly net income was 26.4 billion, 59% higher than a year ago. The company predicted revenue of 54 billion for the third quarter. quarter, slightly higher than consensus. NVIDIA shares fell about 2% because of a narrow miss of revenue forecast for its data center business. Surging demand from the fast-growing AI industry is largely responsible for NVIDIA's strong

results as software companies like OpenAI, Microsoft, Amazon.com, Alphabet, and meta-platforms continue to train ever more powerful AI models. And so there were a few numbers that stuck out from the earnings called the big one that everyone was talking about was this prediction, this broad prediction of not specifically Nvidia's revenues, but of overall AI infrastructure spend, quote, through the end of the decade, which is always kind of hard because does that mean you include 2030 or do you stop at 2029? When does the decade really end? But most people are calling it five years. They're calling at the end of 2030. And the prediction from the NVIDIA team was that we would see, we would

see three to four trillion dollars spent on AI infrastructure by the end of the decade. And so I think that's supposed to be an aggregate number. He's not saying that in 2029, we will be spending $4 trillion on AI infrastructure per year. It's more that just over the next five years, we will see a total of $4 trillion, $3 to $4 trillion, go out. And my initial take was, this sounds high, obviously, but how does it comp to situational awareness? Because Leopold Oshambrner...

Seriously, Leopold Oshambrner did a great job forecasting out the growth of AI CAPEX, AI infrastructure spend. And so I pulled the numbers and three to four trillion is actually way lower than what situational awareness was

estimating, which pegged 2030 at something like $8 trillion just in that year. And maybe something like 15 trillion in total investment through the end of the decade. I think that includes some power spend as well. So the definitions, it might not be purely apples to apples here, but it's funny to frame Jensen as like bearish relative to Leopold, who's like super a GI-pilled. And maybe Jensen's like, yeah, you know, I'm happy to grow. Palo Macro on X said, am I the only one thinking NVIDIA is completely jumping the shark here with these

kind of projections. I know AI people will swallow whatever, but seriously, only Elon could get away with this sort of talk, which is Nvidia expects 600 billion in CAPEX for data centers in 2025 in 2025 and 3 to 4 trillion in data center CAPEX before 2030. Before 2030. Tyler, what you got? I was just going to say, like, people have always kind of said Jensen is like not EGI peeled just because, like, if you're a super AGI pill, obviously you wouldn't be selling the GPUs. You'd just hold on to them. They're like the most important thing, so.

Maybe, maybe. I don't know. Yeah, the other... I mean, the other number that stood out is that NVIDIA's revenue from Singapore keeps breaking all-time highs, reaching $10.1 billion in sales this last quarter. It ranks second

after the U.S. and just behind Taiwan. Year-to-date, revenue from Singapore is approaching $20 billion. A lot of people had thoughts on this and kind of trying to figure out... The, like, loose, like, accusation from, like, mostly Annans is, like, maybe this is someone who's smuggling into China.

But, NVIDIA, the team made it clear on the call. They used a very specific phrase to say that the company that was buying in Singapore had been cleared by the United States government. And so, you know, who knows, maybe there's something that the government is aware of. Maybe the government needs to revisit that. But, like, it seemed like they were very clear that the deal was above board. But the reason that it stands out is if they, you know, they've done 20 billion year-and-date Singapore, if they get to 40, that will be 7% of Singapore's GDP just going to NVIDIA, which is wild.

Yeah. The other thing that was interesting, there was an exchange around between NVIDIA's CFO and an analyst. And Cres, CFO said, we have, in reference to the government, or, or conversations with the government. We have been communicating. She said, if nothing shows up, I've got licenses.

I don't have to do this 15% until I see something that is a true regulatory document. Yeah. So maybe the 15% deal was announced. It's more of like a directional thing and maybe it's not papered yet. I mean, as we saw...

Yeah, and the question is like... It could just be like another chip on the board that's going to be moved around and immediately traded away. There's a lot of these things. things that get, like, pitched, and before they even turn into anything real that affects the business in reality, they get traded for something else. And optics matter, for sure.

Potentially more than the incremental revenue. A ton. Yeah. Is the White House going to be chasing after a few billion dollars? You're, you know, picking it off pennies. But you know what else matters? Ramp. Time is money saved both.

Easy to use corporate cards, bill payments, accounting, and a whole lot more all in one place. Tyler, on the concept of holding onto the GPUs, I was talking to Jordi about this earlier today. It seems like, there's this question of like, if NVIDIA is paying $60 billion in buybacks or they're spending all this money, like, shouldn't they put that, we're in this boom, shouldn't they invest that money in growth, shouldn't they do something else? And Jordi was mentioning, like, maybe they should take more seriously like building an actual cloud service, building their own data center, holding on to the GPUs, as you said. and my counter argument to that was basically that they like invidia has pretty solid revenue concentration across like the duopoly of i believe it's uh microsoft and amazon or like the two biggest power law buyers um i'm pretty sure it's somewhere in here um but anyway like the hyperscalers

are incredibly important um so people go through it's meta and microsoft actually uh is the one that that that that people are, you know, guessing it is. But in any way, the hyperscalers have this, have this oligopoly going in cloud. And there's this game theory where if Nvidia said, hey, we're going to compete with AWS, we're going to compete with Google Cloud Platform or Azure, then those hyperscalers would have an immense incentive

to go even deeper into their own silicon and cut out Nvidia's margin. And so there's this world where, Okay, yeah, maybe in this hyper long term, it might work out, but there would be immense pressure in the short term, as opposed to right now, if you're in Vidia and you have all the hypers bidding for chips, and they're all in super high demand, and so you're able to reap really high gross margins off of that, make a ton of money. You'd basically be giving that up, I think, if you broke up the oligopoly that buys from you. It's really, really great if you're selling something and two people walk in and they both want it because they have to pay the max price. And we've heard those stories about Elon and Larry Ellison getting dinner with Jensen Wong and being like, we need chips.

And of course, Jensen's saying like, well, you know, Sacha and Andy Jassy want chips too. And Mark Zuckerberg wants chips. Like take a number, right? Yep. Anyway. Yeah, and they're in the beautiful position right now, which is however much they make of a certain chip is exactly how much they'll sell. Yeah. Yeah.

Anyway, the next earning call, I hope it's on Restream, one live stream, 30-plus destinations, multi-stream and reach your audience, wherever they are. We are, of course, on Restream. Should we go over to see the Ben Thompson analysis before Doug O'Loughlin from semi-analysis joins in about 15 minutes? Ben Thompson writes, he quotes from the Wall Street Journal, of course, and says, it's always dangerous to invoke the mythical law of large numbers, but the most important place to start with NVIDIA's earnings is to check the supply and demand balance.

Here's the answer from CEO, Jensen Wong, on the earnings call. right now the buzz is I'm sure you all I'm sure you all know the buzz out there the buzz is everything is sold out H100 sold out H200 sold out large CSBs are coming out renting capacity from other CSPs and so the AI native startups are really scrambling to get capacity so that they remember their reasoning that is buying from invidia but also renting from Google right there's a five there was a $10 billion deal announced there last Ben says he made this point a year and a half ago, and it still holds as long as demand. Yeah, the man lives in the future.

He lives a day ahead, or he did when he was in Taiwan. Now he's in America. As long as demand for Nvidia GPUs exceed supply, then Nvidia's sales are governed by the number of GPUs they can make. That supply is certainly increasing, which is why Nvidia's sales continue to climb. But assuming that supply increases are linear, then by definition, Nvidia's growth rate is going to slow as it lap. ever larger revenue numbers that themselves grow exponentially. Yes, this was another reality check on are we feeling the acceleration or are we decelerating? And although the numbers from Nvidia's revenue growth are insane, staggering,

it's like one of the most beautiful charts I've ever seen in business. It is technically decelerating, just like technically. If we were to just like do the math, it is not accelerating anymore. It's not bad. It's just like it is a reality check on things. yeah um so um the big complicating factor is china and i didn't understand this as much as i i kind of disagree with ben maybe a little bit on this because the uh the china does seem like a complicated factor but it seemed like the business in america was doing great and and and really china was like

a call option it was a nice to have if it opens up and AI isn't seen as a weapon and and invidia can seriously grow the china business it's huge huge, huge upside, but it doesn't necessarily mean that they're in trouble right now if they can't do anything in China. So he quotes from CFO Collect Cress. I don't know. I mean, I think, you know, again, looking at Apple's situation in China, like it is definitely bearish. Completely different, though. NVIDIA doesn't actually manufacture in China. No, I'm not talking about manufacturing. I'm talking about purely demand, right? Apple's closing retail store in China. Yes, but Apple sold a lot more in China than

NVIDIA ever did. And so NVIDIA, like, the whole, the, the, the, when, when, when, when, when, when, when, Jensen was selling Nvidia chips to China and the first rumblings of like, hey, maybe he shouldn't do that for geopolitical reasons, um, like, popped up. One of the things people would say is that, look, like, yes, it's, uh, you know, it's kind of anti-free markets to not let him sell in, in China, but also there's plenty of demand in America. Like, he, he, he doesn't have to sell in China. because there's enough.

But when Jensen talks about 3 to 4 trillion in CAPEX before 2030, he's certainly including China in that. Yeah, I think he pegged it at like a $50 billion backlog of demand for Nvidia chips in China if there were no geopolitical considerations. I am very interested to see where this evolves because the AI as a weapon narrative

is definitely cooling off. And so we could definitely see more opening up. But collect crest addressed the H20s. Let me first answer your question regarding what it will take for the H20s to be shipped. There is interest in our H20s. There is an initial set of license that we received. And then additionally, we do have supply and we are ready. And that's why we communicated that somewhere in the range of $2 to $5 billion this quarter, we could potentially ship.

We were still waiting on several geopolitical issues going back and forth between the governments and the companies trying to determine the purchases and what they want to do. So it's still open at this time. And we're not exactly sure what the amount it will be. this quarter. However, if more interest arrives, more licenses arrive, again, we can also still build additional H20s and ship more as well. This is Ben Thompson again. Invita cleared one hurdle when the Trump administration, after pausing H20 sales, allowed them to resume. The Chinese government, however, told Chinese companies not to buy the H20s, according to the Financial Times, which we

covered on the show. Ben Thompson says he's always wary of falling into the trap of blaming the U.S. for Chinese decisions. This is overly solipsistic view of the world is the root of a lot of bad analysis beyond being insulting to the intelligence and volition of the U.S. chiefs, U.S.'s chief geopolitical, geopolitical rival. At the same time, it would be nice to see the counterfactual of Lutnik keeping his mouth shut, or better yet, the Trump administration not causing a ruckus about the age 20s in the first place. The problem, of course, is that Lutnik is right. Chinese companies using NVIDIA chips preserves U.S. dominance of the dominance. AI software stack. On the flip side, Chinese companies not using NVIDIA, both diminishes

U.S. control and for NVIDIA specifically threatens not just their China sales, but in the long run, their sales everywhere, not just from Chinese competitors, but from competition generally, should a Chinese pioneered open source Kuda alternative gain scale, and by extension, the fact that NVIDIA isn't assuming. Remember, there are crushing it on the model side. Yes, yes, yes. And so the next thing might be, you know, maybe it's not. deep seek v5 that's like the you know huge jumping capabilities but the deep seek team or the high flyer team figures out how to have deep seek run on any hardware with an open source stack that's kind of a drop in replacement for kuda well and we were we were talking earlier before the

show about in a if you're extremely a GI pilled yes at some point in the future you can just ask the AI, hey, figure out how to run on hardware other than NVIDIA. Yes, rewrite this so that it doesn't run on Kuda. Tyler, you like that take? Are you laughing? Yeah, that's a good take. I mean, it's like... It's like, if you're

if you're Invidia, you want AI to be like bullish, but not too bullish. Like good, but not too good. Because if it's bad, it's bad for NVIDIA. If it's good, it's good for NVIDIA. But if AI is amazing, you can just in one line like, hey, rewrite chat, Q2. Or re-implement GP-T-5 on TPU or on...

I think it would just be more for it to just ask it to, like, just take over and video the company and then pull it into... Yeah, yeah, I guess I... At that point, we're in, like, such bizarre territory. But, I mean, truly, like, re-platforming should be something that AI would be uniquely good at. I would imagine. Yeah, I mean, you already see, like, a lot of code. Like, one of the early, like, coding use cases was just, like, translating, like, JavaScript to

typescript. It's a lot of what, like something like very simple like that. Exactly. I mean, AI has been good at just translating English to French for a long time. Yeah, I mean, that was the original like transformer. Yeah, yeah, yeah. And also, I mean, when you look at the success of like what is cognition really, really great at, what is Microsoft highlight cognition for?

It's not necessarily like one-off projects. It's more like re-platforming. Yeah, like TechDet. Yeah, TechDet. Oh, you have some enterprise system and it's on C-sharp and you want to put it on Python. like let's rewrite it and that's a huge pain to go rewrite all that business logic but you can just have you know devon or an agent go and like hack away at it for a while so um you could imagine that that potentially would happen in the future but if i but if you've been following george hotses

like you know problems with invidia or an amd he's been really trying to like unseat invidia as the high margin business in the space by getting AMD to solve some software bugs. And I think at a certain level, if there are intractable bugs that even George Hatz can't solve, well then, you know, like the frontier AI model might not be able to solve them, too. So I don't think this is going to be an overnight story that we're going to see Cuda unseated. Anyway, Ben Thompson continues, moats in China. interesting way to think about Nvidia in China and why Wong is so desperate to sell into

the country is the nature of their moat. He says, let's talk about ASIC. This is Wong. Let's talk about ASIC first. A lot of projects are started. Many startup companies are created. Very few products go into production. And the reason for that is that it's really hard. Accelerated computing is unlike general purpose computing. You don't write software and just compile it into a processor. Accelerated computing is a full stack co-design problem. The AI factories in the last several years have become so much more complex because of the scale of the problems have grown so significantly. So Ben Thompson says the answer captures two part of the moat. First is Kuda is NVIDIA software stack for controlling NVIDIA GPUs, which is the default option. The second is

that Kuda is everywhere, which means you can go to any cloud provider, higher developers familiar with Kuda, et cetera. And then Jensen says that in addition to all that, it's just extremely complex systems problem. It's just a extremely complex systems problem. People talk about the chip itself. There's one ASIC, the GPU, that many people talk about. But in order to build Blackwell, the platform and the Ruben, Ruben the platform, we had to build CPUs that connect fast memory, extreme energy efficient memory for large KV caching necessary for agentic AI to the GPU and a super NIC to scale up switch, which we call NVLink. This, This is the third part, which is networking.

I keep referring to NVIDIA's GPUs, but in reality, GPUs work at the system level, particularly for training, and NVIDIA's ability to link GPUs together into a single coherent system is unmatched. This is a big revenue driver, too. This quarter networking, networking revenue was $7.3 billion,

which is more than NVIDIA paid for Melanox, which is the foundation of their networking offering. Pretty sweet. They paid billions with that. Now, just in this, quarter they made more revenue from that acquisition. It's truly one of the greatest acquisitions of all times, says Ben Thompson. Wow. Yeah, insane. I don't know, you can go subscribe to a stertechery to read the rest of the article. My other takeaway was on

the debate about should Nvidia pay a dividend or do stock buybacks. And so I looked at the data of, is this out of character? Should you read into that? Because it's a huge number, $60 billion going out of the balance sheet onto into share buybacks. And it's also like the stock's never been, it's the biggest company in the world, the stock's never been higher, like we're kind of like buying the top almost. Top blasting yourself potentially.

It feels, it feels odd. But I was wondering like, okay, we need a reality check this. Like, is this actually something new? is this an idea that they're like out of out of ideas or something and that's the that's the tealian critique of google doing dividends or share buybacks is they don't have any ideas they don't have any ideas um and so we walk through some of the ideas and there weren't any that really stuck out if they if they build their own hyperscaler well then they're competing with their best customers you mean a cloud service provider exactly really go hard in that yeah yes yes yes yes if they say hey

amazon like you don't need to put any more orders in we're going to we're going to take the GPUs first. It's like, okay, well, then they lost all that, all that margin, all that business. That could be really disruptive. And that business will take a long time to scale to the point where. And then, and then Nvidia is investing in startups and, and they have a whole, they have a whole business that go into cars. They have an automotive business. Invita should just buy Y Combinator. The craziest thing. Just start coming up. I mean, we should sit down and, the dumbest things that they could do. Gulfstream. Yeah. Y Combinator.

Okay. But so what I had, what I pulled in terms of numbers was as, as in terms of the market cap of Nvidia at the start of the year,

what percentage of that market cap was paid out either in stock buybacks or dividends. And so this year, $60 billion, it's a lot,

but the market cap super high. And so in fact, the the total amount of the total net payout as a percentage of the start of the year market cap is 2.82%. So you can think about it as like you're getting 2.8% yield on the investment, although obviously the share price movement is much more important. Last year was 2.7%. The year before it was 1.42%. And in 2022 it was 0.12%. So much lower. So it is growing, but it's still so funny. It was much higher in 2016.

How hard the bears? The bears came out of hibernation in a big way, and he is still up 4.6% in the past five days. Yeah, of course. It's like Jensen has the hardest job in the world. Truly, truly. I mean, it's such a, it's such like a price to perfection, like the perfect avatar for the AI boom. not just indexed AI, but actually, like, throwing off cash, amazing margins, just really great all around. And, yeah, it's hard to imagine them, like, it's not like they're, like, getting over their skis

because, like, there can be. Yeah, the thing you have to give him credit on is the relent, I mean, among a bunch of things, but relentless focus, even during this period of euphoria. I mean, there would be a lot of companies that would be like, we need a mobile phone, let's make the NVIDIA phone, like, we need a cloud. You know, we were joking, the other,

other week about how a lot of the Mag 7 have a social network. Yeah, he's not, he's staying focused on the main thing. He hasn't even, like, joked about buying TikTok, and we have. Yeah. Yeah, it's one of those things where we saw a pullback in Nvidia after the crypto boom. When, I mean, there were a bunch of other things going on with the market and interest rates, but Nvidia drew down immensely. And part of that or part of the story was that GPUs were being,

used to run Ethereum nodes and validate. And when Ethereum went proof of stake instead of proof of work, it became a lot less compute intensive. And so there was this kind of overhang. There was also an overhang from COVID and people buying, there was chip shortages and people buying gaming PCs and stuff. And so, Invidia drew down a ton. Ben Thompson wrote Nvidia in the Valley and basically like bottom ticked it perfectly.

And unfortunately, Nvidia is set up, it seems like it is set up to with, withstand a some sort of like correction or drawback or stagnation in in AI progress like the business would contract of course but there's nothing they're not over their skis where they've like made this massive commitment in there and they'd and they'd be in a ton of trouble at least from my perspective anyway let me tell you about figma.com think bigger build faster figma helps design and development teams build great products together we have our First guest Doug O'Loughlin from Semi Analysis hopping on the stream in just a few minutes

in the meantime. I do want to go through this post that Ben Thompson highlights from Ethan Ding about the changing economics of AI. It's a longer post, so we'll have to do it a little bit later, but he quotes here, imagine you start a company knowing

that consumers won't pay more than $20 per month. Fine, you think classic VC playbook, charge at a charge at cost, sacrifice margins for growth. You've done the math on KAC, LTV, all that. But here's where it gets interesting. You've seen the A16Z chart showing LLM costs dropping 10x every year. So you think I'll break even at $20 a month. And when models get 10x cheaper next year, boom, 90% margins. The losses are temporary. The profits are inevitable. But demand exists for the best language model, period and the best model always costs about the same because that's what the edge of inference

costs today. When you're spending time with AI, whether coding, writing, or thinking, you always max out on quality. Nobody opens Claude and thinks, you know what, let me use the bad version to save my boss some money. We're cognitively greedy creatures. We want the best brain we can get, especially if we're balancing the other side with our time. While it's true, each generation of Frontier model didn't get more expensive per token, something else happened, something worse. The number of tokens they consumed went absolutely nuclear. ChatGPT used to reply to a one-sentence question with a one-sentence reply. Now deep research will spend three minutes planning and 20 minutes reading and another five

minutes rewriting a report for you while 03 will just run for 20 minutes to answer. Hello there. The explosion of R. Allen test time compute has resulted in something nobody saw coming. The length of task that AI can complete is doubling every six month, what used to return 1,000 tokens is now returning 100,000 tokens. So interesting dynamic there. We will continue with that, but we need to introduce our first guest to the stream. Doug O'Loughlin from Semi Analysis. Doug, how are you doing? Good. This time my mic should work, so... There we go. Sound great. Miced up. Thank you. Dude, I didn't want to repeat the last time.

I like that you're holding it too, like some type of rapper. It's much more personal. Can we call you? I'm going to call you young semi. I am getting a little hassle. our vibe. It's perfect, though. I love it. Hello, welcome to young semis rap. Anyways, what's up, guys? I have no idea what we're talking about. We're talking about Nvidia earnings. Should we be talking about something else? Yeah, what else is...

No, I think that's the right thing to talk about, but I guess my brain is really broken. I'm like, bro, this is a snoozer compared to the blowout quarters of past. Yes, that's fair. What did you take... What do you think about the fact that Jensen is on a relative basis, extremely bearish compared to Leopold Oshendbrenner at situational awareness.

Leopold expects $15 trillion in CAPEX or AI spend by 2030. Jensen, a mere $3 to $4 trillion. I don't know what to tell you, man. I think there might be, some people might be talking their book, right? Like, I guess I do know Leopold. I definitely know his worldview is Manhattan Project for AI. So I think maybe Jensen hasn't come around to the Manhattan Project for AI, but I think just a mere two or three trillion is quite a bit of CAPX. Some would say it's not an insignificant amount of money.

But they tried to bait him on the call too about tokens, token revenue, about it 10xing next year. I thought that was kind of interesting. I don't know. It was a fine result. Yeah, yeah, yeah. Go into that token revenue. There was something where it felt like he was making the case. He was trying to lay out the economics for the actual customer, saying, if you spend this much, you can make this much money.

And it felt like, I don't know, just like a different way of framing his business. Can you explain exactly what he was doing there? This is, okay, so one, semi-analysis, as you may have figured out, is onto this. We're trying to figure out the math ourselves. They obviously scooped us, but not quite scooped per se. But there's a lot of work to be done to understand the unit economics, right? A lot of people have asked, hey, what's the AI ROI, right? Like, you guys are spending all this money and, like, the, I'm sure you saw the tweet on the timeline.

Depreciation is bigger than all the revenue. Yes. This is a total bubble, blah, blah, blah, blah. The depreciation number, I don't know, I think Martin Screlli of all people was like, this really sucks. But I think a lot of it, the data center long line. Just saying, to be clear, saying it sucks because it's just not good analysis. It's not good. It was a bad analysis.

Yeah. Yeah. Yeah. I think he specifically talked about like human level intelligence on a B 100. He used a pretty long useful life. I don't know if we agree with that useful life. But like the data center side of that cap X and the power side of it, extremely long useful life. Like let's just say 10, 20 years. So you can, there are definitely some of that is pulled forward. And so you can say that cap X is not going to be just be used for next year or this year. It is a multi-year investment. But on the on the GPU side, I think the unit economics is the question that people are really trying to get to the bottom of. And the one that, um, Colette specifically called it out is you spend $3 million on Iraq and you can make $30 million in token revenue. I think one, you know, which tokens are you selling, right? That's like the real question here is like, how many tokens are you selling? What's the throughput?

There's a lot of assumptions here. Some analysis, I'm not going to lie to you, it's like quite literally working on this right now to figure out like what's the type of like, you know, if they were selling in profit tokens, which are like have a premium because they're really good of sweet bench, right? Or if they're selling GPT or if they're selling MAMA or are GPC. P-T-O-S-S, right? But it's pretty clear to us, at least, that it's very, very economic to run these things.

Like, one rack, if you could find 100% free tokens, you are, or no, paid tokens, you are minting money. It is just straight up like a 10x return on one year. That might not even be, you know, that might be a multi-year thing where you're able to print a lot of money. These tokens, the GPUs that are able, the amount of tokens that are able to come out of it is we're talking, you know, millions and millions. millions and millions a year, if not billions or something, or on an imprints on like an annual basis.

And if you get that all paid, you're, you're printing, dude. Yeah. Yeah. I mean, it feels like we're in this, like, middle ground between, like, there's, there's some, you know, I believe Nvidia revenue is decelerating technically. Yes. The earnings are kind of snoozer. That's because supply is increasing linearly. Yeah.

Yeah. Yeah. So it's, uh, don't let anyone call, say law of large numbers. that's not what it is. Okay. But, because it'll get, like, misquoted. Like, what happens is, like, the base gets bigger, and so it decelerates. It's mechanically impossible for it to accelerate on a larger.

Maybe not mechanically impossible, but it's extremely hard. Right. You go from 50 to 100 billion, then you have to go from 100 billion to 200 billion just to keep that same rate without a deceleration. Yes, yes, yes. So it's getting a bigger and bigger base. Yes, but a year ago, like, there were plenty of people that were saying, like, yeah, like, acceleration.

Like, we are literally accelerationists. Like, we believe in acceleration, and we believe that, like, yes, like, we will go from 50 to 100 to 200 to 400 to 800, continue, continue, continue, and more so, because even that would be continuous growth. Anyway, so you have this, so you have this narrative, like, there is some deceleration in the, in the rate of Nvidia growth, but the chart is still insane. And then simultaneously, like, the profitability of actually using their product seems very, very good. Does that just mean that we're in this, like, it's a snoozer in the sense that we're in this era of, like, the economics makes sense, the business is good. It's like the oil business now. It's not going to fall apart, but it's also not going to explode. Like, we're in kind of a smoother territory, whereas, like, a few years ago, when we were before the kink in the graph of NVIDIA revenue, it was very much like, what is going to happen here?

That's a great question. This is where, you know, I definitely think, and I get why people are saying, bubble. Big numbers are happening. There's a lot of exciting technology. The internet is a great example. But let me give another example that I think kind of maybe it's a better analogy. Apple, when it came out with this iPhone, and I'm sure you can tell me exactly where yours is next to me, next to you, right? Everyone got one in like a really quick amount of time. It became this giant generational product where we went from like zero people with a with a smartphone to 60% of people with a smartphone within five years, right? And it sold a crapped on of them.

it kind of, the first few years were these exciting blowout quarters, especially in the beginning, like these, you're like, holy crap, this changes everything. And then it becomes boring execution. So we could be in the world of boring execution, but I definitely think, you know, the ROI of the underlying rack is really good asterisk if you have paying tokens. But if you look at the vast majority of tokens that are being consumed today, they're definitely not paid, right? That's the free side. I think I was here on GPT5 Day. which by the way I literally got off that call and then I started thinking about ads

and then we posted it on semi-analysis so it was very funny I wish I had the hot take then but it would have scooped us but yeah that's an example never scoop yourself never scoop when you're the subscription business at least yeah yeah exactly it's like dude

I gotta keep I gotta keep the lights on no but if you think about it it's about how to translate tokens into revenue in a way that isn't like this freemium model and I think you know agentic purchasing which is like kind of our you know five second thesis of how we think is a potential way to monetize the free business. That's an example

of that's how you monetize it. And all of those tokens now that are freed, you found a new end state of people who don't want to pay for intelligence, right? I don't think you can make the assumption that someone in Indonesia with the GDP per capita way lower than the United States is going to be paying 200 bucks a month for GPT 5 pro, right? The reality is those tokens, to become something of value otherwise we're just going to print deflation we have to make it tokens are like uh tokens are like a website like www.com right like whatever you have to make it a business on the tokens and so i think we're in that portion where it's pretty clear that if you can if you can monetize those tokens you're going to have a really really good business and now it's the question

of how do we get the token machines to become revenue monetization and so that's kind of the the path forward, I think, of, you know, maybe the boring execution that happens for the model companies and the startups and the whole ecosystem altogether. But invidious side as, you know, being the fundamental core infrastructure, it's pretty clear, horse, you know, horses out of the barn for the infrastructure. They're building. Everyone's buying them. People are like, holy crap, we've got to have more power. Like, it's, it's, you know, you could see it. some of some folks on the timeline were bearish on the buyback do you have any ideas of better ideas for how to spend that money if you were running invidia than then then a buyback

so okay oh my god i'm i'm a buyback discourse all again like buyback discourse anyways has like a little bit of a little bit of religion but i think remember the buyback is a $60 billion authorization, an authorization is not a commitment to. It means that they can buy. You have to go get your authorization from your board of how many shares you're allowed to purchase. And $60 billion for a company the size of Nvidia is like kind of a penny in the bucket. They make a lot, you know, that's, you know, that's a one year profitability, I think. No, let me make sure I get that. Yeah, that is one year of cash from operations right now. So that's actually not that much. 60 billion sounds like a huge number, but they just make

so much money that that is a correctly sized buyback authorization authorization for invidia in my opinion i think um what should they do with that money is going to be one of the greatest questions of uh like all capital allocation history that same question came up with apple uh came up for apple when they were gangbusters right they didn't reinvest back into all this other stuff they printed cash they bought back a lot of shares stock went up the the most tangible example of financial in our time. Yeah, didn't they return something like a trillion dollar? Yeah, one trillion dollars to shareholders. Yeah. And dude, it's it's actually kind of like, okay, so you can feel different ways about it because it's also kind of spooky. Like just if you are an active market

investor who is listening to this, you'll, you can understand. Apple is a stock that will not go down. It like, it is like it is against like it levitates. It's like 30 times earnings. earnings isn't even growing, but they just gush so much cash and they purchase, repurched so much shares and they're such a huge part of the S&P 500 like you could make an argument that like Apple's percentage contribution of the P&L of like the S&P 500 from like 2010 to 2020 is like 25 like it's it's it's a meaningful amount of percent so the entire 401k is the entire country got paid by Apple's buyback that's kind of sick if you think about it but on the other so you don't think they should launch a mobile phone or a social network no we're just

joking how like every company like gets euphoric yeah it's time to make a fun oh six out of the mag seven kind of have a social network if you count iMessage and you count lincoln and youtube and twitch like andy jassy has twitch which is kind of hilarious Elon has Twitter and we were thinking the obvious one is jensen buys TikTok but i think they got to do DJX cloud social network where you can just talk to other other coot engineers fans basically yeah i i don't think i think Jensen's pretty strategic. I don't know what he's going to spend it on. I want him to spend it on something like inspiring and mind blowing where you're just like, dude, this is, this is sick. You're a genius, right? But the reality is deploying like $100 billion capital. It's pretty hard. That's not like walking around. Like that is like buying a country. Like what are you going to do with it? Yeah, yeah, yeah. I mean, like the best thing to do if you care about innovation is give the money back to your shareholders. Let them invest in venture funds or let them invest in startups and and maybe don't try and build. that functionality internally. Yeah, capitalism would say that for sure. I definitely think that you should you should swing if you see a fat pitch. Meanwhile, you see all the hyperscalers who are plowing back as

much money as they can into the into the neoclots. They are saying, hey, we are seeing a giant ROI. So I think that that's like, you know, a fundamental question for Collette and Jensen. That's very hard to answer. I don't, you know, I don't think you could tell me what would I do with $100 billion that's adjacent to NVIDIA's business, that would be a better business than what NVIDIA does. The answer would be maybe nothing, honestly. If you could purchase something, that would be sick. Also, I wanted to make one thing before we continue. Back on the social media thing, if I've learned anything on being on the timeline, dude,

it's billionaires are just like us. They just want to tweet at people getting fights. They love that. It's amazing. True. What did you think about Colette Cress's comments? And she said we have been communicating in regard to the government. If nothing shows up, I've got licenses.

I don't have to do this 15% until I see something that is a true regulatory document. So they have a license. I think to me, how I took that is they have, they think that there's a high conviction. They can sell 820. So a certain point, that's my reading of it. I do think that there's this weird spot here, though. Well, I think that's that, my reading of that was she felt that we're selling age 20s and we're not just sending wires to Uncle Sam yet. Yeah.

Yeah. I don't think they're sending wire. Well, because here's the thing is like they have a deal at the top level. But no one has like the administration hasn't gotten done. Like the bureaucracy of like, okay, here's the mechanics. Right. So if they have licenses, they can send it. Maybe they will hold it on, you know, they have enough cash.

I think they can wire a little bit over. I think they're good for the money. But yeah, I think it's just kind of, it's a complicated administrative thing, meaning that they have what is in theory the deal, but no one has actually inked the mechanics of it, of how they're wiring it. And it could always change.

What do you think could, if people were bearish on just a small beat, what do you think could flip the timeline gigabarish? Would it be just a bad miss? I think the other, we were talking earlier, you know, potentially hyper, like language coming out of other hyperscalor earnings calls of saying like, hey, yeah, we're actually reducing orders, things like that. Those to me, like it feels like one sentence in another hyperscaler's earnings call could send the stock down a meaningful amount. but I think what it would have to be is pretty much the explicit, the explicit endorsement of everyone who's involved being like, yeah, this is a bad idea now. Like, we got to, we got to stop

investing, right? Like, like, um, like meta. Like what happened with the Metaverse? Yeah, yeah, exactly. We're like, all of a sudden, it's time to get fit. And we're, you know, maybe we shouldn't have spent like $10 billion year on the Metaverse. And like, did you guys know Horizon, whatever that was like the launch day the like do you know the launch day statistics it's like really funny he spent like like I don't know let's just say $20 billion and like on launch day there were like a thousand people

on like I mean I you could have just sent me a check in the mail I would have showed up for like a million bucks dude it would have been better that's wild dogs just sitting there just

I'm like I'm an act I'm a happy active user as long as these checks keep coming. So, that's not like a grant their scheme. The dog. I think it's, I think it's a belief that it's a bad investment. And at the current time

investment, you know, there's like a vague vibe in terms of what is being bullish or bearish or what is investment appetite. No one can really define it for me. No one really knows. But at this current point in time, like, if you're looking at the board, I think Zuck is bold

the F up, right? Like, he is so bullish. He's paying for all this stuff, you know, Sam Altman, Sam Altman, dude, he's like, give me another trillion dollars and I will give you a trillion dollars to compute. He's like, maybe there's a bubble, whatever. I think most of the participants right now are pretty excited. And I do want to like, you know, I wrote about like the internet bubble actually like quite a bit. Yeah, I was going to ask what, what, how much have you studied Broadcom specifically? I was, I was trying to look up. I don't, from, from what I could find Broadcom wasn't doing

buybacks even during the crazy heyday. You were buying a bunch of companies with their stock. That was the big thing. Yeah, but I don't think Broadcom is the perfect comp, actually. Yeah. Yeah, Broadcom was like, Broadcom was a baby in the 2000s. And then also Broadcom, the

machine that is Broadcom got bolted together with Hucktan like, you know, the actually capitalism's greatest hero. He is so hardcore. Like he, his whole thing is like, you know, do you know his entire company? The nominative determinism is crazy. 50 people of IT cover the entire organization of Broadcom. It's like, it's like he's the cut to the

like cut to the core. Anyway, sorry, this is a complete tangent. What do you mean 50 people? 50 people is the entire IT department of like the 40,000 people at Broadcom. What? That's insane. Yeah. Wow. Dude. And yeah, and their revenue ramps were so humble. It was like a hundred to 200, then 200 to 400, something like that. In 1998. Yeah, but that's like, but you, that's not the right company. Cisco was the, Cisco was invidia. Cisco is invidia. That's the, that's the comp. And those revenue ramps were pretty nuts. But I think the difference though is Cisco was not gushing cash like Nvidia was. Like, it's not even close. I think they had like a 20% profit margin. And also, I think the difference too is during the internet bubble, it was pretty clear that everyone was.

doing fraud effectively. With a lot of circular transactions, right? Yeah. Yeah, yeah, the circular transactions. That was a huge, pretty well-known thing. I actually talked to a guy who quit his job at one of the networks and became a hedge fund guy. It was like, yeah, we're shorting stocks, the revenue zero.

Like he literally left to go join the financial industry to short stocks. Wow, that's wild. I don't think we have that. That's a true bear. that's conviction you know that's like the inverse leopold actually um so i i just don't think we have that widespread like craziness that the internet bubble was in a lot of ways like websites barely worked man you could barely use stuff i mean i'm i don't be you but i am using chat chitin my yeah so here's an example so yesterday satia hit the timeline and and dropped the thread on how he's

using GPT-5 and co-pilot. Everybody was immediately just like, bearish, bearish. Like, why is he posting use cases? I think he's just flexing on everyone being like, I have the code to GPT-5. Like, I got a copy. Yeah.

But it says, it does say something that he's just showing how a product is actually valuable to him in his work life. And people are like, bearish. It's very moderate. It's a very moderate thing. Because he's not saying like, oh, yeah, like I used, I used GPT-5 to make decisions as a CEO.

No, it's like, I use it in this narrow use case, it just reminds me about this particular thing, creates meeting notes, like very practical, like, you know, one iteration forward of the technology cycle. He's not saying, you know, oh, I'm on the beach because GPT5 runs Microsoft now. Yeah, I have my coding swarm running Microsoft instead of meeting. Yeah, yeah, yeah. I don't think he's, I think it's, I think another part about the tech bubble that really make, and like, look, stocks are expensive for sure. I'm gonna hedge like you know no one knows the future markets are humbling that's something I really want to like things can happen that are just like crazy outside of like what you know dude the 2000 tech bubble was like pretty crazy like really really really really

way crazier in a lot of ways in terms of just like raw speculation the revenue accelerations you're seeing from hypers should give you some credit that like there is revenue happening people are losing money on selling tokens but like this isn't this like totally fake business model yet. But just like history, what happens is, and there's a really good book called Technological Revolutions and Financial Capital by Corot. I can't say her name. Perez. Yeah, Carlotta Perez. Yeah, I don't know why I can't say it. Carlota. Carlotta Perez. Thank you. Carlotta Perez. You got to throw a little accent on there and then it comes out easier. Yeah, she's great. But that, I think that that's like a really good way to think about it.

It's like this stuff will have a blow off. Like capitalism works via these like concentrated blow off booms that create a new technology now are we possibly in one yeah for sure i mean we're doing real investment but we don't know you know when you're building a new industry pretty much the supply demand curve is unknown to anyone no one knows what the actual demand is they know it's larger than the supply today so they build more supply but then at some point you figure out where demand is and you're like ah we reached it and then you completely overshoot it and then you're like crap. Yeah, it'd be interesting to look back of how analysts at the time we're looking at the car

being like, not everybody's not going to have a car. Like, every family in America is not going to have a car. Who needs one? Who needs one? It's so expensive. You only use it here or there. I can ride the bus, you know. Exactly.

What about Dario's math was going viral earlier this week? He was talking about how, if you look at. If you just look at the, basically, the P&L of a foundation model company, it looks really bad because you have, the cohort stuff. Yeah, you have exponentially increasing costs. But if you look at each model as an individual company, it's like you invest some money and then you make more back. You invest some money and you make more back. Do you think that? That made perfect sense to me, but I don't know. Yeah, but every, but it's the interesting thing is you have Fintwit, which is just like everything's bare. Then you have TAC, which is, oh, that makes, that makes sense.

Like, maybe it's a little overheated, but you should do more. They should 100 acts every year. Yeah, it's like, well, you're telling me that you're making that return on a cohort. Let me give you $100 billion. Yeah, of course. I think, look, and I'm guilty of this as well. You get to sound smarter when you're bearish, for sure. And Fintwit is inherently bearish for sure.

That's part of it. I think, I think this. this is going to be really interesting because every step of the way people are going to be have been skeptics and probably will continue to be skeptics which is like pretty great honestly if you if you're talking about like a true capital formation bubble um one of the reasons why 2000 was so intense was like dude i read a book i forget it's like um i can't telecosm dude it talks about how like infinite bandwidth is like infinite information like it's like a real vibey book oh sure it's pretty crazy it's pretty nuts um and i just don't think we

every I think we haven't had this like new age belief that ASI is going to change everything like you know the tech people believe that but everyone else is like no and so I think as long as we have that skepticism it kind of prevents some of the worst aspects of like let's say a true bubble and I think the real question to be asked is how meaningful is the spend that can be converted into revenue because if these if those tokens can be converted into revenue via like agentic purchasing, like the GPT-5 router example. Those are big markets, man. All of travel, all of purchasing, all of consumer, dude, trillion-dollar markets get a take rate. Do you think GPT-5 or chat GPT will eventually go free only? Because I was thinking about it, like, I love

the $200 a month version. I'm hitting the pro and O3 Pro constantly with like basically everything. I don't know if I'm making the money or losing money. But it feels like, It feels like something that in a few years, like, people will be like, oh, well, like, you know, your stated preferences that you don't want to use it, but you still use it. So it's fine. And it just feels like that might be the way it goes. And I'm wondering if there will ever be, like, again, this era of, like, luxury software, like a really expensive thing that is really just for a narrow segment of the, it feels like a consumer app, but it's just for like the tech elite and then it's gone. I don't know. What do you think? I don't know. I think it is pretty cool at this moment in the time. I definitely feel like it feels unsustainable because we know Pareto curves exist all around us, right? Have you ever heard the statistic about like, I want to say it's like clash of clans? Like the 1% of people who were like really into clash of clans were like 35% of all revenue. And so if you don't have a usage based like a usage based take rate, what's going to happen is the hyper core use. users are just going to use the hell out of it and you're going to lose money because those are you know that's the market dude the guy who's hyper addicted it's like you know you're not cap well and in the in the with chat GPT you could be on a 500 a month plan and they can still monetize your purchasing that they're driving yeah yeah so maybe it's high oh we won't

monetize like the the purchasing activity that we're driving because you pay it's like well you don't really it doesn't really matter to you yeah yeah they're taking a cut on the back end. Yeah. Have you started thinking about the market size for romantic companions or AI companions? I haven't, but I'm telling, I, if based off a four-o's feedback, it's big, dude. I know. Yeah, it's crazy. I didn't expect that. I killed my boyfriend. Yeah. I thought that was just like a couple people on a Reddit, like just a very niche subset, like less than 1% of the audience, less and 1% of users. But that's the thing, man, is there's these Pareto curves all around us.

Like, that 1% is probably, like, in this hardcore niche using 4-0 in this, like, in ridiculous way, right? That's what I saw on the day that Chachipiti launched. We obviously had a ton of different guests on talking about, you know, the product and everything. But I was just refreshing Reddit. And every single person was like, I'm unsubscribing from the $200 a month plan in protest, like, over and over and over. lots of people yeah it was i mean we can go back yeah yeah i mean it's a it's a unique it's a unique dynamic in the world of like like romantic or adult content because typically there has never been price discrimination with zero marginal cost in that market like you have zero marginal cost

on just the adult content websites but you don't have price discrimination it's usually like a netflix type subscription if you can get any money out of people or you have only fans is price discrimination. You can have a whale that 1% pays 35% of the revenue, but it's not as high margin because you're passing it through to the creator. And so this is the first time where you could potentially have a situation where you get a whale who's hooked on buying virtual burkin bags for their virtual girlfriend. And you are instantiating those burkin bags to the tune of $100,000 of real money that has actually zeroed cost. That's why you can be frustrated with Elon's advertising through his account recently but you can't you you can

also make say realize like hey he might be super rational if he's like there's a few billion of ARR in this product and I think Google's going to stay away from it and I think eventually and if he hates open AI and just wants to like I actually I think it's I hate it I hate to say it's a good strategy it's a really good strategy because I think um one thing that it's so like the one thing that the the revealed preference from five is that clearly the virtual girlfriend economy is much bigger than we thought um and then the you know using yeah well the other thing is is the market for people that i think everybody was overestimating how big the market for reasoning models are and because from what we've heard it's like 95% plus of people are just

using chat chputt like google or they're using you know yeah and it's not they're not like i heard some rumor that Elon insisted that that the that annie be able to use the reasoning model instead of just and it's like way more expensive she needs to think she needs to be able to do the math she needs to be to do the math and you have to trust your aunt you have to trust your girlfriend she's analyzed she's analyzed her love for you from first principles yeah and it's real yeah i mean i i guess There's some rationale there, but there's no, there's no evidence in the market that that actually results in, like, lower churn for romantic companions. But I think Elon just from first principles being like, I want the, I want the romantic

companion to be really be able to think they're the best. Donnie must be able to do novel physics. Yes, exactly. Well, here's the thing, though, is I think maybe what we're, what we're struggling with is we're focusing on the wrong thing because now with the reasoning chain, we can do RL, right? What if you can have the best freaking girlfriend, now you can R.L. your way to the best companion. I don't think that that would be possible without reasoning. So let's think about that, you know? Yeah, yeah, that makes sense. Maybe setting up a virtual environment with, you have, some verified reward. I don't know what the reward would be. Well, John, John's other take that I think is real is at what point, like imagine somebody has like their AI

companion that they've spent hundreds of hours with and then the companion says i want a new dress yeah and it's a hundred dollars yeah it's like a digital skin right like if you don't get it for me maybe i'm just gonna bounce i'm gonna be a little bit i'm gonna be off well i can be upset yeah the reasoning chain works and okay so dude the digital skinning i'll be really sick honestly because as you guys know that's a proven business model right dude like i hate to say it i i have paid for skins and games man. You can, you can, you can, you can, you can pay for skins and games really easily. You're like, and the only fan, I mean, this, and the only fans, like, industrial complex, this for sure happens to.

Super real, super real. Yeah, I mean, it's, it's more real, right? I played Counterstrike for years. I maybe paid like $10 at some point, but I, I get so much value out of that game. Finally, it was like, wait, I can have, like, flames on the gun. Like, this is sick. What's 50 bucks? Like, that's what I would pay for just a normal game. And I'm getting that much value. So it was no. brainer. You would be a whale if you were really, really game into today. If I had the time to play the games, I would 100% be a whale. Get this man a CSGO right now. You need to log in on Valerite right now. Everyone tell, gets his guy on Valourin right now. Give him a credit card,

get him some skids. Yeah, yeah. Get the, get the epic games guys over here. Pitch me on joining Valerie just so they can hit earnings. Do you think, do you think Med is paying mid-journey at 9-15? years a year as part of that deal they announced no 100% I don't know what the actual economics are but like I think to me I like how they just announced it like casually but like clearly it's like a massive like just a massive deal well it probably not so massive that it needs to show up in like disclosures right under 10% of revenue you can probably not which is a big number for them yeah yeah yeah yeah yeah but man David holes what a tear what a tear if I had to guess actually actually, Zuckerberg probably tried to buy it.

Yeah, of course. He definitely tried to buy it. And Mid-Journey is probably too big enough to be like, no, no, no, no. We'll partner with you. We'll get distribution, but we're not for sale. So I think that that's probably what happened, but he's like, okay, well, okay, well, how about we have some special preference, right, access, and we share economics and some future thing.

I don't know what the hell that looks like. And then, like, great. Yeah, my read is that it was probably as material as an exit just spread out over some period of time. Maybe forever. Which is an amazing outcome for the mid-Journey team. And honestly, great for Zuck. And great for the product.

I'm going to enjoy using mid-Journey photo filters on Instagram more than whatever they were cooking up before. For sure. They're definitely on the frontier. Yeah. One last thing on the video, and we'll let you go because I know this is the last minute. How should we be thinking about automotive? I noticed that they had, you know, they have this kind of like the order of magnitude business where I think it was like $40 billion in the data center, $4 billion in gaming, and then like $400 in automotive.

And with the self-driving car narrative, it feels like Waymo, we're finally here. And yet Tesla's matched up with Samsung. Google obviously has the TPU. I don't know if there's TPUs in the Waymo's, but it just feels like Nvidia's not like really taking that very seriously maybe or maybe they will be and all the OEMs will come. It's like, NVIDIA seems to be good at, you know, gaming, big, huge chips in the data center, but then nothing really in mobile and, you know, mobile gaming and nothing in the car really materially yet. But how should we be thinking about like the other areas that NVIDIA could potentially chop down? So I definitely don't think there is a ton of GPUs in production, in production cars. There's definitely a lot that happens.

I'm sure there's like, if I had to guess their go-to-market is something where you have some amount of distributed compute that also works with. your data center because like you know jensen is data center to the rest of the world right and so i think that that is probably their most exciting part but i think at least historically they've done a pretty bad job maybe not bad but i just don't think they had like that moment yeah right like i think they've they have really good technology but their go to market just hasn't hit the like the magical moment like qualcomm actually ironically is doing very well they've acquired their way in i think especially acquired but also organically. And I can't speak specifically to like, you know,

the Oren or whatever product, like whatever specific automotive skew. But in the businesses that I do follow that have been really successful in terms of automotive, I think they've been a totally different to like go to market. And that is mostly by pandering to OEMs and really adopting their stack and trying to suck up to them versus Tesla obviously wants to do everything from first principles on their on their own. And so I think they're like, no, no, no, we want to have this as a competitive differentiator.

So I think there's just some kind of like go to market and let's be real, man. I think if we're talking about like, you know, maybe if we're talking about like, if I was actually a CFO of MVIDIA and they have like an opportunity, dude, acquire into the automotive market. You could, you could crush because you would then have the whole thing. But in at least last administration, no way in hell. Maybe this one, they're open for it. But if there's a China Samer review, it's as good as dead.

That's the other problem. Sorry. Can you clarify that China review? China, Sammer, strategic something, something market review. So essentially the antitrust reviews that happen in the United States get approved, not approved, right? There has pretty much never been a chip. There are no more chip deals. And you should never think that there's going to be a new chip deal.

Every chip deal that happens almost always gets struck down by the Chinese market specifically. a good example is the Intel Tower semi-deal. That would have been sick for Intel. Definitely was a fan but it became pretty clear that like, you know, give me the tit for tat, right? The Chinese regulators are going to be like, why would

we approve this if we've been effectively at like a Cold War at like a semiconductor geopolitical level? So you should effectively assume if there is a Chinese business because if you say no, they effectively say, okay, all your business

in China is mine. That's like a very, very quick high level and that's a pretty hard bill to swallow. And so they've been blocking these deals. So effectively, especially M&A and chips and semiconductors, has been very close to zero. This is kind of the flip side of the, of like how, like Figma, I think got antitrust review for, for Adobe in America, but was blocked in Europe and Europe's been blocking. But with semis, it's much more about the Chinese side reviewing. Got it. Yeah. And I think the ANCIS deal just went through. I don't know. That's a, like, so the baseline assumption is it won't happen if there's a Chinese segment. That's mine, at least. A couple quick questions. Bill Bishop in the substack chat says Trump approved

modified Blackwell for China. China's still not buying. I don't have any context. Do you? This is, this is the H20? Is the Blackwell? This is, no, no, this is the B20. This is the rumor this morning. B-20. I think, or B-30, I don't know what the number's going to be. I think, and this is a great question, Bill. I think this is the, and I think Bill, more than anyone else would appreciate this is what Chinese companies say publicly versus what they do privately are often very divergent. And there was actually a really good podcast that my friend shared me and I did not. I asked them to TLDR it. But it gives an example of how effectively what they do is in-person or like they effectively knowingly act in bad faith.

So they'll say, oh, yeah, we will comply with this and they won't comply. Or they'll say, no, no, no, don't do this, but they'll also stockpile. So I don't know if the official statement makes a lot of sense because at this point in time, H20 is, like, majority of Chinese inference is probably done. It's probably done on Nvidia GPU anyways, whether if it's smuggled or purchased legally. I think odds are it will get purchased. The B30 will be purchased. Obviously, the official party government view is like, we don't need this crappy American technology because we are going to have our own destiny, but in private, if I had to guess, they're going to be buying it and they will also be building it at the same time, right?

Like, the plan has always been to do both. And they've consistently done that, specifically in the semiconductor industry. And you're starting to see where you go from a low end copier to actually starting to go up the tech tree. And we're already seeing it. A good example is applied materials. which is a semiconductor manufacturing equipment company, semi-cap, they are starting to lose really hardcore in the low-end market in China,

and part of that is actually not even just the technology being worse, but rather the copying is getting better. And so they're going to do both. They'll definitely buy it and they'll say, oh, no, we're not going to use it for these things, but probably use it anyways, and then slowly go up the technology tree themselves. How suspicious should we be of that Singaporean buyer,

Colette called it out on the call saying it was US approved already should people be reading into that as much as they are on the timeline? I think I think it's not a bad thing to read into is a small amount of money. If I had to guess it went to Jehor. So it's like a Chinese customer

inside of a non-Chinese unrestricted place. There's a lot of different ways they can skin that cat. Yeah. Final final question. Mark Cuban replied to one of our newsletters on Intel. He said, maybe I misread Intel's SEC filings on the matter, or maybe they have changed, but based on my readings and a confirmation from ChatGBT, GBT. Let's give it up for ChatGPT.

The Department of Commerce got warrants for Intel stock that can only be exercised if Intel sells 50% of its foundry. In that event, they get Intel shares. If within five years, Intel does not, the Department of Commerce gets nothing. In all cases, Intel gets all its compromised chipsack money if they live up to that agreement, anyone else read it differently? That is correct.

So specifically what that is doing is aligning. And also, by the way, I've been trying to shop this stake around. The Trump government intel investment is good, in my opinion. It is a good thing. I know people are like anti-they're like, oh, it's communist and like you can be angry and whatever. But is the thing that Mark's alluding to, is that only the poison pill or is that? That is only the warrant side.

It's specifically, and dude, it was in the press release. It's talking about like over 50, you know, the second they don't own 50. percent, the government will have warrants for the rest of it. So this, in my opinion, if you think about it, aligns the government with, like, this happens often in financial transactions where the warrants are given as an upside kicker for something you want to happen. So what does this sound like and what does this align to? It aligns the government with Intel IFS being an independent subsidiary, which is consistent with what Frank Yeri, who is the chairman of the board, who is like the guy I have a personal beef with at this point.

We've written so much about how the board sucks, blah, blah, blah. Really wants to sell IFS, I think, and Liputan does not want to, but it's pretty clear the future is separate. And I think the investment is good because Trump can essentially force people to give them orders,

to circumvent tariffs. And then over a long period of time, the real problem with Intel is they don't have customers and they also don't have like an ultimate backstopper. The original plan was that Intel would be the first customer and, ultimate backstopper. You don't believe in either of them being a good option. Now you can have the U.S. government kind of broker that relationship for you. I think that that's a path forward for

Intel. And that's the first one we've had in a hot second. So are you good on time? I have more questions. I do have to go after that for that tape. But I do appreciate you guys. I love, I love this place. Honestly, invite me back whenever. But yeah, appreciate you guys. We will talk to you later. Have a good one. Have fun out there. Bye. back to our show. Let me tell you about Vanta. Automate compliance, manage risk, prove trust. Continuously, Vantta's trust management platform takes the manual work out of your security and compliance process and replaces it with continuous automation, whether you're pursuing your first framework or managing a complex program. Thank you everyone in the chat for engaging. I saw some funny commentary that the substack chat is maybe the Winnie the Pooh Bear with the Tuxedo on and the

in the YouTube chat are having a wild time talking about Fortnite and romantic companions. Tanner in the chat says not quite one hour this time with semi-analysis close enough. Yeah, we always try to get to the 60. I'm trying to be pulled. We should switch to potentially more important than Nvidia.

What's that? Will DePue says he's going to detwinkify. Current weight is 161.2 pounds. See you in eight weeks. Oh, wait. Let me check the date on this post, just three days before bulking season starts. Interesting. September 1, September 1. He must have gotten the memo. We got to send, Nick, can we send

some mass gainer through Will to Will. Let's get his address. Honestly, we might already have one. Let's send, yeah, let's send an eight weeks supply of mass gainer. We'll send him some creatine. We'll send him some creatine. Yeah, we should get a whole, the whole whole army of supplement providers to put together a package to bulk him up. Yeah. I think he's going to look great. And we're going to be up in the bay. We'll have to get a lift in with Will ASAP.

John Holtz-Quist says he's sharing a screenshot, an F-35 pilot held a 50-minute airborne conference call with engineers before his fighter jet crashed in Alaska. And he says, feel the same after my conference call is TBH. That was a crazy story. Yeah, the pilot ejected, so he was fine. It was just a total, total loss of the plane itself. The F-35 does not look good for the F-35. Kevin Kwok says, I entirely judge the Stripe podcast on whether they have actually finished

at least one beer by end and refused to watch one until that's true. I did scroll through the one with Scott Wu, and I was like, okay, the beer's going down, beer's going down. Wait, Tyler, what do you got? Yeah, okay, so I looked to the Scott Wu, and I went all the way to the end, and you see in the frame, it's him talking, and then there's four, basically, full guineas is like three of them are literally full and then one of them is like maybe down to like the logo okay interesting unbelievable well also uh scott said when he posted he said he he went non-alcoholic

because he had to get back to work which i respect but i feel like even if you're drinking a pint of non-alcoholic even if you're having a cheeky non-alcoholic pint you should finish the pint also john colson should be a dog when it comes to ginnis right i mean i agree i agree i think the whole the whole conceit of like hot ones is that like you actually die because you you're eating the hottest wings and that's what brings out the hilarious reactions from everyone from Shaquille O'Neal to you know whoever else is on the show like they have fantastic guests and the beauty of the show is that the hot wings put you can't hide from the heat well you can't hide from the heat um but it it puts you off and so you give more candid conversations and it's funny and it puts you in this un well yeah so the so the challenges they they publish the podcast during the middle of the work day. Yeah. So if they were hitting publish and the boys are getting sloshed.

Yes. So they, I mean, they need to do a Friday night, a Saturday night because a lot of people would be down to have a cheeky pint, but it must be on, it must be on the end of the long work week. Start recording on Friday.

Friday evenings, make it very clear. Maybe release it on Friday. Yeah. I think, I saw Wilman Nitz had a good response. I think he said, said, we want to see five to eight Guinnesses and split the G every time on the first sip. What is it?

Oh, split the G is the Guinness. You're supposed to drink down to the Guinness glass. Oh, I didn't realize that. On the first sip, you have to get down. Okay, okay. Well, they did open, I mean, Cheeky Pint implies the existence of a show called Cheeky Rack, correct? Yeah, we were talking about this. Yeah, I think this might be the one where, yeah, where you.

where we have a tech person on, we interviewed them about financial control and projections and capex spending and depreciation schedules while crushing a 30 rack of Bud Light. Yeah. I think that might be the move. That might be, there might be what we're there. Something there. There's something there. Cheeky rack.

Cheeky rack. It just has a nice ring to it. Shout out to Ilhan. Ilhan over at Boston University. Yesterday was watching with six of his friends from their dorm. Today, they got 10. 10? No way.

The team grows. The team grows. Welcome to the stream. Get all 10 of you folks on graphite.com. Code review for the age of AI. Just go sign up right now. Create accounts. You can get started for free.

Graphite helps teams on GitHub, ship higher quality software faster. Honestly, if you're in college, you should create an account and then you're at least familiar with this. If you get into the workplace, you can be like, yeah, I'm already AI enabled. There's been a million articles

about how there's this delta in early. stage hiring between like early stage it's harder to get a job it's harder than ever to get a job out of college but the AI enabled folks who are able to say put on the resume like I know AI skills and I'm I can use AI tools and get more leverage out of them not having as all three all three of our summer interns actually made yes made things yes as part of the application process without being asked we didn't say anything I mean and then Nick over there was saying he was vibe coding yesterday and I was like really like We didn't hire you as even a programmer.

You did not market yourself as anything related to technology, really. And you've been doing a fantastic job doing what I expected you to do, but then vibe coding was just like added on. The production team is always troubleshooting stuff using AI. We're going to give Nick a mic soon. Okay, yeah. We'll get everyone a mic. Anyone.

Continue. Sucks at Power Bottom Dad with an absolute banger. You're scrolling on your phone, slow day off. online. You look up. Your kids have moved out of the house. You're 65. Your parents are long gone. Panic takes hold. You want your time back. Your youth back. But it's too late. You look down. Three new notifications. How exciting. Wow. This hits like that Rick and Morty sketch about the Roy, like the whole life flashing behind your eyes. This is crazy. Yeah. Remember, folks, you got to touch some grass. Always have some grass. Yep. Touch some grass. Keep it on.

you keep it on you thank you to the ketone IQ folks for sending over some of these yeah we're gonna be we're gonna be we're gonna be taken ketone shots later that guy wound up going to the same gym as the founder which is cool anyway uh Dylan Patel uh says interesting coincidence that canter fitzgerald has first question on invidia earnings after the h20 is unbanned put on the tinfoil hat of course Howard Lutnik is a former chairman and CEO and his son runs the firm now so Hunter Fitzgerald. That's a good spot. First question on the NVIDIA earnings. Yeah, I do wonder, as I was listening through the earnings call, I would have loved to get semi-analysis folks in there asking questions. I would love more media folks on the call asking questions. But it seems to be restricted

to traditional cell side analysts. But I wonder if that's like a hard and fast rule or if that will shift as like, you know, the, the cell-side banking world kind of de- disaggregates a little bit because a lot of what Dylan Patel and the folks at semi-analysis do is very cell-side research adjacent. In fact, in many ways, it's superior because it's more focused. Anyway, we talked about this. Sam Allman said their AI may be in a bubble. And Dr. Parake Patel says MFRO has been telling us we are on the brink of AGI for the last three years. And the moment he ships a bad model, he says, we're in a bubble. Not quite.

I still don't understand where that question or that actual bubble, like, topic came from. But at the same time, it's different to call a bubble when your, when the aggregate value of all AI stuff is at like, you know, 50 billion and then it 10 X's in a year. Like, that is a different, that's a very different environment. Yeah, I mean, Open AI with close to a billion weekly actives or whatever. You can make it a billion month in revenue? They'll be fine. It's the number 10 through 50 labs that have no revenue and no users that are the ones that are really going to struggle. It just feels a lot, it feels more, it feels less like, I mean, there's elements that feel like dot com,

but there's also elements that just feel like 2010. It was like what was going on in 2010? There was like kind of a bubble inflating and there were hot startups that got overvalued. Ultimately there was like a power law in a lot of the things that came out of that era. You know, you got Facebook at a trillion or whatever and then kind of like, you know, the next biggest social network was one-tenth and then the next one-tenth of that. And there was a lot of B2B SaaS that got built for point solutions and narrow use cases. So I don't know.

I think that that talk with Doug overall was pretty white-pelling and pretty like, I don't know, what's in between a white and a black pill? Like something just like even keeled. I felt even keeled after that. Anyway. 80% chance that Jerome Powell cuts rates next month according to polymarkets. Let's pull this up and make sure that that is. Fed decreases interest rates.

Now, this, like, somewhat should be priced in because I feel like there have been slight interest rate decreases already happening. The question is just, like, would they cut rates more aggressively? Well, I don't want to read too much into this, but there was some reporting from the journal about Trump's mood. And you could potentially read into this with this polymarket. Apparently, Trump wants to be at the White House more frequently this term, blaring music with the doors. of the Oval Office. Do you have this article? Working later into the evening

and telling his... Can you pull this article up? Yeah. I want to read through the full thing if we have it. Oh, wait. Do you just have a long post? Because I have to run for a second.

So if you can read through that, I'll be right back. Yeah, let's do it. Let's do it. In Trump's second term, a bolder president charges ahead unchecked. That was where this quote is coming from. this article is going to be fairly political, but this is not a political reading. So some aides to Donald Trump warn the president that building a ballroom at the White House

would force them to tear down part of the East Wing and disrupt daily operations and tours. Trump said he would build it anyway, and the contract was given to builders chosen by the White House. In his first term, administration officials regularly curbed Trump's impulses on matters big and small, including on tariffs, immigration, and controlling the Federal Reserve. In his second term, Trump has been surrounded by fewer people who, We try to dissuade him, according to officials, Trump allies, and observers. I think he's learned that there is not much that can really stop him from what he wants, said Mark Short, who was Trump's director of legislative affairs in his first terms.

In recent days, Trump renewed a call to end mail-in voting, announced a new policy of coercing local governments into abandoning cashless bail policies, threatened to send the military to Baltimore and said he'd like to send, and he'd like to send it to New York and Chicago as well, all of which pushed the bounds of his authority. In one of the most aggressive steps in that direction, he tried to remove Federal Reserve Governor Lisa Cook from her post on Monday, setting up a conflict with the Supreme Court,

which has recently suggested that the central bank is protected from direct political manipulation. Some of his new directives are encouraged by advisors, while others appear to come from Trump himself. Seven months into his second term, Trump has taken to rifling more frequently with authoritarianism after positing during the campaign, he would be a dictator only on day one of his presidency.

Such a wild quote. Wild quote. He's got a lot of... You get the quote about the music? Playing music with the doors open or something? I want to know what music he's listening to. How closely does his playlist match the campaign playlist? And was Trump featured on the Panama playlist?

I want to see his Spotify. That's a good question. I don't think they found him. That would have been like headline. Yeah, the other crazy line in here, and I hadn't heard of this before. I'm surprised. Apparently Trump is giving away campaign-style baseball hats to visitors emblazoned with the phrase Trump 2028, even though the Constitution bars are running for

another term and keeps him in the White House and keeps the hats in the White House office. I mean, yeah, Trump is still at like two or three percent on polymarket because the market is just pricing like, yeah, like legally it can't happen. like the funniest outcomes the most likely, I guess, the 3%. Yeah, so this was the line. I'll read it again because it's just hilarious to visualize. So Trump wants to be at the White House more frequently this term, blaring music with the doors of the Oval Office open, working later into the evening and telling his advisors that he is having fun.

The reason I highlight this is if you can remember times where you're working late, you got music blaring, and you're having fun. I mean, at least he's having fun. At least he's having fun. Certainly more than the first year. There's a comment on substack right now. Max Condrat says, semi-analysis needs the first and last question

on the next earnings call. Leopold Ash and Brenner, too. Clear pill. I completely agree. Speaking of clear pills, there's this post from Nick in News and Vidia's data center revenue from Q2 chopped out at $41.1 billion. Analysts, we're expecting 41.2, billion and then just it's over it's like off by the slightest fraction and and the and this anime

character is just holding this like tiny little pill um i actually prefer more air bubbles i'm a bubbler i don't really understand the reference speaking of uh speaking of speaking of analysts should we get into this uh speaking of analysts should you analyze your data on julius what analysis do you want to run chat with your data and get expert level insights in seconds go to julius loved by over two million users and trusted individuals at Princeton, BCG, and Zapier. Should we get into this post from Grizzly Research? I will never be able to get that right. And we've spent 30 minutes talking to the founder and I still get it wrong every

talk about it. What do you want to talk about this? Grizzly reports, Grizzly Research. Oh yeah, we should go through this. This is fantastic. So I guess short seller grizzly research has come out with a new article as of yesterday. called Archer Aviation, the Nicola of the Skies. Archer, if you're not

familiar, is the previous company of Brett Adcock. It's an EVTenberg research is gone but not forgotten. The idea remains. Grizzly research seems to be adopting, picking up where Hindenburg research kind of left off. So, Grizzly research says.

Archer Aviation, which is a public company. Yes. Flying cars. Yep. They make flying cars. They spacked. in 2020. And surprisingly, they're only down 7%.

They've done well, which is, you know, there were a lot of SPACs that went out, went down by 90%, 80%. But... They're up 159% in the past year. Wow. What's the market cap? It must be in... $6 billion company.

That's pretty, pretty big. And so Grizzly... Certainly has a good, it has a reasonable narrative for retail. It's like flying cars. If that's going to happen at some point, maybe you want to buy... Got a pure play here. You got a pure play here. It's a pure play.

Retail loves a pure play. So Grizzly reports here says, Archer Aviation has built a reputation as a leading publicly traded E.V.TOL company through misleading projections in PR, reminiscent of Nicola's tactics. Our in-depth analysis conducted with E.V.Tol engineer scrutinizes all major players

and finds Archer's midnight aircraft fundamentally flawed and likely uncertifiable. In contrast, their competitor, Joby, aviation, is making tangible and impressive progress on all fronts. Site visits to Archer's state-of-the-art Covington, Georgia, quote-unquote, state-of-the-art Covington Manufacturing Facility in June. July and August 2025 show little-to-no production activity.

Fun fact about Joby. You know what the founder's name is? Joby. Joe Ben. Joe Ben. Joe Ben Bevert. Bevert. B-E-V-R-T.

Yeah, we need to go back. I mean, there's... He has vertical takeoff in his... his name. The nominative determinism of Joe Ben is so good. And I believe, I don't want to get it wrong, but I think I have a friend who worked there at Jobi and said that they really enjoyed the engineering culture and they really like the company. But even Jobby, I think by their own admission, would say that they're not, you know, fly. It's a very tough business. It's just an early stage. It's a frontier technology. I don't think that they would make the claim that

it's hard enough to make a fly. Yes. And then you, you, you, you know, it's a very tough business. It's a you add in the regulatory complexity. It is fascinating with all of this with the backdrop of what's going on with Zipline, where there's no human, it's not a flying car, but they seem to have solved some of the EV-Tol pieces. And so overall, I feel like there's going to be things happening. There's going to be value created in E-V-Tol broadly,

but it's going to be hit a risk. And some companies are going to be able to nail it. Some companies are going to fall behind. So continue reading. Grizzly allegedly visited Archer's facility and found little-to-no production activity. This is the one in Covington, Georgia. This comes despite claims of current production along with current scaling to production of 50 aircraft per year and a final goal of 650 aircraft per year with support from Stalantis.

That parking lot looks empty and it's bright in the middle of the day. Maybe it's a weekend. Yeah. So they have a $6 billion order book. Grizzly says it's inflated with questionable and fraudulent commitments. There's a company called Air Chateau, which is based in the UAE, has an MOU for 100 midnights, which is valued at around $500 million. Is it MOU something like a, I mean, it sounds like an IOU, honestly, but it's something like

an L-O-A? Yeah, it's basically saying, like, if you can make these, we'll buy them. Yeah. What is it? L-O-I? L-O-I. Letter of intent? Yes.

That's what, but MOIS is based. It's not a full binding contract potentially? Yeah, it's non-binding, but it's meant to show something that you can then go show to investors and say, like, we have demand for this. Yes, okay. So the UAE operator, Air Chateau, has only ever, I guess, purchased one helicopter in 2023, and they're claiming they lack the scale or capital for such a fleet, and they also believe that Air Chateau is now defunct.

There's a picture of the founder of Air Chateau. here, with a good-looking regular helicopter. Future flight globals up to 116 midnight order is tied to a shell company with no operational track record. So that is concerning. Cacao Mobilities, 50 aircraft commitment fell apart after Archer failed to deliver for Q4-24 demo in Korea. Yet the order remains in the backlog. The U.S. Air Force is up to 148 million Agility Prime contract has only awarded 33 million with just 744,000 dispersed and the 110 firm fixed price portion. We were just talking to a friend yesterday. They were saying, oh, yeah, like the DOD did a contract for our product.

And it really accelerated the business. And he was like, oh, we got like, I don't know, I forget it was like $5 million or $500k or something. It was a decent amount of money. And I was like, oh, did they actually like pay that? Like, or was it just like, you know, some sort of thing? And then it didn't materialize. And he was like, no, like, we did the deal. We sent them the products and they paid and, like, it was done.

But that's not always the way it works. So you obviously have milestones with a lot of these, a lot of these contracts. And from Grizzly researchers post, it sounds like the Air Force was, was, you know, headline number was 148 million. But the, but the contract only awarded 33 million and has only paid out around $750,000. And the fixed price, the firm fixed price portion only paid $1.3 million, which is obviously much lower than expected. Anyway, you want to talk about the launch edition in UAE?

Yeah, so they had a launch edition. Grizzlies claiming there was nothing more than a staged hover at a photo op location using an obsolete aircraft recycled for marketing, the choice of venue, lack of real test data, and diversion of scarce engineering resources, all underscore that this was a PR stunt, not a meaningful step towards certification or commercialization. The company also claimed to have delivered their first midnight aircraft there, yet they remain the sole owner of this aircraft that has now supposedly been delivered two times. Interesting.

So the recent defense pivot is a desperate attempt to stay in the race, says Grizzly Research, but the company lacks resources and capacity to be a credible player. In conclusion, Grizzly Research believes Archer's Fragile Foundation will soon collapse akin to Nicola's downfall as mounting deceptions enravel and investors demand accountability. And yeah, what's interesting here is that I remember we looked at the video for Archer and we were going really back and forth on like, is there CGI in use here? Is part of its CGI? It was kind of hard to tell.

And that's just the nature of like launch videos now, honestly. They're all pretty hard to tell because like the CGI is really good. but the bigger question is from our discussion of SpaceX yesterday to actually deliver like we wanted flying cars to actually deliver flying cars it's not enough to build one you have to build the system that builds the flying cars and so from my perspective I don't know about the price of the stock or any of that but if I'm if I was going to be excited about a company that was that was in the flying car industry I would be much much more focused on how does the manufacturing process work?

How can they reliably produce more and more at a faster and faster pace? And are we seeing a true acceleration in the rate of progress, even if they're blowing up all the time? Well, that's the thing I was thinking about, if you're an EV tall company that wants to one day carry consumers, you can't really afford to have what SpaceX, the video we covered yesterday, you know, a highlight reel

of a bunch of... I completely disagree. I completely disagree, because you know, know who's flown on top of that exact rocket that's blown up so many times and they have a YouTube video of the rocket blowing up, the astronauts. Yeah, yeah. The astronauts fly on top of it because eventually it got to a point where, yeah, they blew up the first 50 or whatever. Sure, it just wouldn't be good marketing. And then they did 200 without any blowups and you're like, yeah, at this point they don't blow up. So I'm down. All right. But I think if, if Archer or Jobie

or anyone released a video of a bunch of a bunch of their aircraft exploding, it would not be a good time to do that right now. It wouldn't be good for the stock. But 10 years ago. But I mean, Archer announced that they are the official, they're the official air taxi provider for the 2028 Olympic Games here in Los Angeles. And again, I think that, I think that Grizzly would at least, I don't think they would bet on that.

Yeah, Grizzly says, we think Archer's ambition to fly a certified. aircraft that the L.A. Olympics in 2028 are laughable. I can't imagine how hard it is to get sort of, when I hear about UAE or any of the sort of markets that seem to be more open to technological experimentation, like SpaceX started with quadulene atoll. And I think that the regulation in quaduline atoll is probably a little bit easier than Starbase Texas. And, and so when I hear, oh, the first time we fly is going to be, the most high profile event in the most regulated country and the most regulated state and the most regulated city in America and the world, that seems like a really, really tough challenge.

Like, if Texas would probably be easier to get regulatory approval for, like, there's other countries that would be easier to get regulatory approval for. Going straight to the Olympics is a huge, huge, huge, like, push. But I don't know, it might just be, you know, framed as, like, We're the official air taxi company, and we're here, you know, advertising our business more than a product. But, yeah, lots of skepticism. Lots of skepticism. Yeah, so Joby trades it near double Archer's valuation. Really?

Somebody on Reddit in response to this Olympics announcement says, if anyone from Jobi's team is reading this, please step up your PR game. Archer's way behind you in product, but ahead in marketing. You need both for a new company. Yeah. So retail on retail violence. I think we will see a lot of, a lot of small EV-Tol aircraft like what Zipline's doing before we are actually flying in in EV-Tol vehicles.

Helicopters exist. Like they, like, what is the actual pitch here? It's just, is it just cost that we're optimizing for? Because are we getting a new capability? And what they finally wound up building, Archer, the latest one, the Midnight aircraft, looks a lot more like an Osprey helicopter where it can tilt into a plane mode and go a lot further. Are we trying to say that this is going to be like a helicopter but faster or like a helicopter but cheaper? Or like a helicopter, but you don't need a helicopter pilot?

Is it a driverless helicopter? I still don't fully understand the definition of how is a flying car different than a helicopter? Because this doesn't look like something that I can drive on the 405. And that was what I was promised. That's what I was really promised with the flying car was like you can just drive it as a car at 75 miles an hour down the highway if you want. And then you can also just take it in the air. Take it in the air and fly like a plane at.

hundreds of miles an hour. That was like the high level pitch. But if you make just like a plane or you make a helicopter, those already exist. People that have driven cars fast and canyons will tell you that it's possible to get some air time. Some airtime. Yes. Anyway, if you're planning on building a helicopter business, building any business, you've got to get your brand mentioned on chat GPT. Go to profound. Reach millions of customers who are using AI to discover new products and brands. There are some new demos that hit the timeline from Higgsfield. Swap to Video is powered by Nano Banana. What a wild name for a product from Google. Swap any pick into a video. It's literally bananas. They turn this runner into the queen.

They swap Angelina Jolie into some stock footage. I'm sure she will love that. This is very cool. Look at that. Yeah, that's fun. The queen just running. It looks so realistic now. Yeah.

Even the motion blur in there. Like everything is just mashed. it's the perfect one play it again play it again it's so funny the queen the queen that really that really takes us to the next level that's fantastic oh anyway do you want to scroll back to some of the other posts that we skipped over uh we can go through yeah four chan is suing the UK government toby over at shop the Shopify highlighted this uh there's some funny excerpts um Yeah, well, I tried to read this, and it was late, and I didn't really understand. So, apparently, I didn't, 4chan is alongside Kiwi Farms, which are both these, like, communities online.

But the introduction says, the Internet is a global system of communication between computer servers, located in data centers around the world. Despite the Internet's global reach, it is more or less universally acknowledged that the Internet is predominantly an American innovation, built by American citizens, residents, companies, and that the United States has the largest and most thriving technology sector of any member. state. Foreign governments, particularly those in Europe, which have not managed to build technology sectors of their own for the past half decade have sought to control the American internet and hobble American competitiveness. So they're suing the UK and they start off by just like making the argument that America is the best. And so Toby, Lukie asked Grok to turn the complaint into a green tax and Grox says, be the internet, global comm system between servers, mostly American invention, built by U.S. peeps and corporates.

U.S. has biggest tech sector in G7. Foreign goves, especially Europe, can't build their own tech, try to control American internet with laws and stuff for half a decade, threatens Americans and others with fines, arrest, jail, for illegal stuff on their sites. Forgian and Kiwi Farm Sioux, UK, enough is enough. This is not that great of a green text. I'm going to say it.

It just feels like it just turned it into some punchy pay. paragraphs, and they just threw the little, uh, the little greater than symbols at the front. Yeah. Uh, it, it lacks the, read, read, MFW or like, uh, Sorrow. PNG. Like, I need, I need more 4chan green text. Read the notes from, from Andy over at two cents. Okay. The notes from Andy at two cents. Andy says, this is incredible. Uh, the parties. Um, number seven, plaintiff fourchan is a limited liability company. Uh, and then eight, Delaware was a colony of the United Kingdom of Great Britain until the assembly of lower counties of Pennsylvania that declared itself independent of

British authority on June 15, 1776, thereby creating the state of Delaware. Delaware was subsequently the only state, first state to ratify the declaration of independents, the instrument which created the United States of America. Anyway, I don't know where this is going. They're clearly having fun with the complaint. Yep, I think we can switch gears to Mahal. He is just on a tear wire cutters as the mattock is a whole different kind of robot vacuum i agree yep it is it goes to town yeah my house so this is kind of like a proper sort of like ground up rebuild using modern technology of what most people think of as a rumba or a robotic vacuum cleaner obviously that was from the previous generation of artificial intelligence where there were some pathfinding

algorithms, but Madik has, you know, actually a really solid world model and image generation or image recognition. And so it will, you know, really hunt around your house for dust and dirt and messes and clean it up. Mahal sent us a few of them. We've been enjoying them. This isn't a promoted post or anything like that. But we're just a fan of building cute little robots, which I'm a big fan of.

But, you know, if he needs to pay sales tax, he's got to get on Numeral HQ, sales tax an autopilot, spend less than five minutes per month on sales tax compliance on Numeral HQ. TCP says, are you a small business owner struggling to make ends meet? Then try being a big business owner. Always a brilliant strategy, always the correct strategy. Go big. Just become a big business owner. This is a great post.

And market cap for Ridge will go nuts when Sean has a baby. Imagine all the product ideas. He will come up, including the Ridge Pacifier, which looks crazy. And also, like, a weird rendering where it's kind of like, that's, like, it's shaped like it would be outside of the, yeah, it's inverted. It's inverted. It's, uh, AI genera, but the, um, but the concept is sound. Yes, but I like, I like building the Ridge world.

I don't fully think of the Ridge as like the super tactical brand, but, uh, that's, certainly where people went in the comments, thinking about, like, tactical strollers and stuff like that. Oh, I mean, their initial customer base was the Everyday Carry crowd. Yes. They even bought EverydayCary.com. Oh, really? I didn't know that. Wow.

Yeah, they're very tactical. Well. Citrini research says, NVIDIA down 3% on earnings. I think everything will be fine. Open Twitter. See Satya posting like he's selling AI courses. I think a few put contracts. that's ridiculous let's actually read through what what how satch is using gpt 5 and co-pilot

so i i i have a very different read on this but let's take the current not just says based on first way he's using it based on my prior he's prompting based on my prior interactions with a person give me five things likely top of mind for our next meeting uh nick does this for us very well yeah draft a project update based on emails chats and all in a series, KPI's versus targets, wins versus losses. So the benefit, of course, of running the whole business on 365 copilot is that they have email and Outlook, they have chat in teams, they have meetings, and this is true of Google for the most part, I guess, except Google doesn't really have a Slack competitor, so a lot of

people are having to piece those two together. And that's probably where a company like, what's that enterprise, like search across your entire enterprise business? Glean. Glean comes in potentially. But certainly having this as kind of like the homepage for your, you know, like dashboard for your whole website, all the communications. Like it's a very reasonable, very tractable, like, you know.

use case. Just saying like, hey, here's some prompts that you can do that only work here because we have all the data. Are we on track for the product launch next November? Check engineering progress, pilot program results. Risk, give me a probability. That's cool. All stuff you can do in chat GPT if you sync everything in, but you know, these things come out of the box. Well, yeah, and opening eyes trying to build a lot of this functionality. Yes. And that feels like more of the narrative here, which is Sotia, like, mostly just reminding people, like, every Microsoft product, every Microsoft product has, by contract, the best Open AI model, like, on the day it's released, because, like, that is the nature of their agreement. Like,

they get a copy of whatever. And so, like, if it's easy for people not to remember that. And I'm sure that there's a lot of Microsoft 365 co-pilot business owners who run on 365, and they've been on that for a long time. And then they're hearing this, oh, like, I've got to get my company, like, AI native. Like, I got to get, like, AI tooling. Like, should I bring in another service? It's going to make enterprise and, I mean, even just, you know, mid-market, SMB is like a tough category for Open AI to really dominate. But they've won consumer. And so I think that they should be fine over there, but if you're running

a business and you have, and you're like Microsoft 360, we are a Microsoft company, we use Outlook, we use Teams, that's the backbone of the business, and then you start seeing credit card charges for $20 a month on OpenAI, $200 a month in Open AI, and you're just like, wait, no, no, no, we get

GPT5 for free at this company. Like, we don't need to pay extra for GPT5, we get it for free. And so maybe this is the, maybe this is the solution to the ad free version, is if you get on Microsoft 365, they won't have ads in it and it'll be able to do everything and you won't have to pay? I don't know. We'll have to keep tracking. Anyway, if you're looking to integrate AI into your customer service workflow, go to fin.a.i, the number one AI agent for

customer service, number one in performance benchmarks, number one in competitive bakeoffs, number one ranking on G2, fin.com. Legends. You want to talk about Garmin? Let's do it. Garmin is selling a real-time health data for your horse with its easy to attached tail wrap and censor the blaze equine welcome wellness system measures your horse's heart rate strides gait and more during rest and rides get insights into their health and fitness and make more informed decisions about their well-being and training i need this as well and you know we talk to the the fit bit for cows company well now your horse will have a fitbit or maybe it's an eighth sleep for your horse but a eight sleep i'm getting a hot five ultra this uh this uh this note from

Kane is cool. Garman picks niche but spending activities and goes, I can crush every electronics maker here. And it sounds like a super fun business. Yeah. And hunting. Long range shooting, fishing. And so they make very specific, specific hardware, often ruggedized and figuring it out. But G-photo Or says, honestly, one of the most confusing spam emails I have gotten. I do not have a horse. but I love it I love it and some of these photos look really really cool

these Garmin the uh I had no idea that Garmin is a $45 billion company huge huge I mean there's a lot of fishing to do there's a lot of long range shooting

or bow hunting you want to go out bow hunting without a garment on your on your bow and I think they still crush it in smart watches despite obviously having totally competition

yeah because they've differentiated niche down like Apple's going to be kind of the broad broad consumer choice for a lot of folks, but Garmin has, like, all the trails. They ask the question, what are you running from? We'll help you go faster.

For sure, for sure. Brian Johnson says, one of the best new habits I've started lately is calling friends for 15 minutes. I say, hi, tease them or share something funny. Then it's immediately over, no lingering, no fuss. It's great. It's called flirting. Not with your friends. It's called chirping.

Chirping. Hey, did you bench today? what's up this is this is just a funny habit i mean it's genuinely hilarious it's extremely brian johnson coded to take something that's like just calls like you know uh this is like the classic like tech guy i'm gonna actually do this to him later i'm gonna do this to him later yeah yeah tech guy events being friendly he's texted like once i'm gonna call him out of the blue yeah hey how you doing tease what's teased you buddy the elf what's your favorite color i'm I mean, people like it. 7,000 likes. It's good take. Anyway, starting today, flow by Google. Google AI ultra

subscribers can generate V-O-3 fast videos without using any credits. So you can fine-tune your clips to exactly how you want them and only spend credits on that final upload. May this be the piece de resistance on that new product. And Bone GPT says, I just got unlimited V-O-3 fast. So people are very excited about that. That's fun. I've been really enjoying the V-O-3 model. Continue to continue to enjoy it and continue to use it. So, uh, Bucco capital bloke sounds like we are getting the final Google ruling tomorrow.

For now we exist in the liminal space between it being over and so back via condeos. This is about the Chrome. Chrome divestature. Yeah, what's the news? Has it come out yet? Okay, we'll have to track it later. Um, so, uh,

nano banana, more nanobanana news from Google. Love these. What does the red arrow see? Google Maps transforms with Nanobanana. Levels IOS says if you're not building a mini startup with nanobanana today and launching it tomorrow, you're missing the opportunity of a lifetime. This image model just made hundreds of new startups and apps possible.

The only limit is your creativity and how fast you can ship a user interface and put a stripe payment button on it, I think. And Levels has done this a few times. I think he had some image generation app that allowed you to decorate your house. So you take pictures of the empty room and then it would give you ideas. for decoration and inspiration. I thought that was very cool. And with much cheaper AI images,

it just unlocks new use cases. So certainly high color. Apparently somebody sniped nanobanana.aI. Oh, okay. And read about dot com. Where's dot com go? And it's not Google affiliated, but they're charging up to $6,500 per image.

No, they're just saying, yeah, they're charging $64 for 800 high quality images a month. Wait, isn't it like $10 a thousand or something? Yeah, they're marking it up. Again, this is what happened with chatchipiti early on where people were just reselling chatchipati before they had a lot of focus on mobile. Wow.

Well, if you're launching a foundation model and you're using a codename, buy the dot AI, buy the dot com, don't get swooped, don't get swooped. Tyler, have you played with nano banana yet at all? Yeah, it's really good. Yeah? I think the main... Does it unlock anything new? Like, is there going to be a new studio Ghibli moment?

I don't know. I mean, there's some cool things with just how, like, consistent the characters are, right? So if, like, I take a picture of someone, I can, like, change their shirt and nothing else will change. Like, it's like, you can't do that with, you know, GPD4 image. Yeah, yeah, yeah. And it just in chat GPDD, I feel like it always makes the face just, like, a little bit uncanny valley when it re-renders it. Yeah, I think that's probably the main use case. Because it the face just stays the same, basically. Yeah, for, like, photorealistic stuff. Yeah.

But I think the main takeaway from nanobanana is that, like, you know, they have, like, a run. like they have a poster in the house because like you saw it was like a week before it came out everyone's like oh it's on elm arena what is nanobanana and then Logan is just you know tweeting the banana emoji like I think that is like the real vibes I love it no um the yeah the other question is I wonder I don't know um yeah I wonder if this is going to unlock any like specific viral moments or if it will just be something used by companies because it's cheaper and more reliable. I'm certainly excited. Do you think that there's a potential that it's enough to get people to switch to Android? Why? Because it would just like natively have. Have you used the iOS

remove object? AI feature? Is it really bad? You should go demo it. You should take a picture of something Well, I think... Because you have the latest and greatest, right? Yeah, yeah. Yeah, if you take a picture on the phone and then you go and you say, like, okay, I want to remove this, I'm going to try and remove this Diet Coke from the image. You can go in to edit, and then in the image editor, there's a clean-up button, and you can, like, prepare cleanup in iOS.

And, like, as you kind of draw, it takes a long time, you kind of, like, you know, I guess it only picks things that you can remove. know like it's rough it does not quite work okay here we go now it's finally going breaking news essence has filed for bankruptcy protection what is essence uh yeah Gabe literally called this in the chat he says jordy might know essence don't think john would know about them called it perfectly uh fashion marketplace um surprised uh to see this i'll try to get an article Canadian luxury e-tailor told employees on Thursday that the filing was necessary as tariffs took an unexpected toll on the business and to preempt a forced sale by lenders. Essence's filing for bankruptcy protection after what it described as an attempt by lenders to force a sale of the company.

Chief Executive Rami Atala on Thursday said Essence's creditors want to put it up for sale under the company's creditors' Arrangement Act, a process similar to bankruptcy protection that allows corporations to restructure their finances. Vatala went on to say that Essence will fight a sale by filing its own CCAA application within 24 hours. Recently, we have worked closely with financial and legal advisors to develop our own restructuring plan to stabilize the business and rebuild it for the future. The court will decide which path we follow likely within the next week. Again, I believe that from, I believe, Gabe, correct. me if I'm wrong, but I believe that Essence just primarily partners with boutiques so they don't actually hold a ton of inventory. I might be thinking of Farfetch, but hard, I mean, it sucks when you're a business that isn't a business of just selling goods and you can't

do that. Well, good luck to them. Let me tell you about Adio. Customer relationship magic. Adio is the AI Native CRM that buildscape. and grows your company to the next level. He can get started for free. And we have our next guest here in the Restre Waiting Room and now in the TBP and LF you go. What's going on? Welcome to the show.

Hey guys. How are you? We're great. Good to me. Great to be here. Doing great. Can you introduce yourself a little bit for the stream? Yeah.

My name's Ajay Agarwal, a partner here at Bank Capital Ventures and I've been here for 20 years, started life in the startup world way back when and been in tech for a long time. Were you ever out of the Boston office or were you always in Silicon Valley? I was originally in Boston. So when the fun got started, you know, roughly a little over 20 years ago, it was 100% in Boston. And that was sort of the year of, you know, Route 128 and, you know, all the great stuff happening in Boston. And, you know, the business on the East Coast has really migrated to New York.

I moved out to the Bay Area to open our West Coast office. We launched that effort, you know, about 15 years ago. Now, you know, at BCV, the majority of our team is on the West Coast. And our second biggest office is New York. So Boston, great town, great life sciences town. Well, and YC was there. It started. YC was there originally.

I remember the old days of YC at MIT and then, you know, YC moved. So it's, you know, Boston's a great place, a lot of innovation here. It's just, you know, the application layer of AI, the foundation layer of AI. It's all happening in SF and New York these days. Yeah, when you say like the firm started 20 years ago, you're talking about BCV specifically and not like being capital broadly. Like how do you tell the story of like the venture firm like launching from within that larger structure? Yeah.

I mean, it's a great question. You know, the firm was started back in 1984. Mitt Romney and partners, you know, started, you know, and they, you know, early on, that fund was really small. It's like $37 million. And so some of the early deals were what you'd consider tech companies. You know, we invested, we were the first investor in Gardner Group, way back when, which is, you know, classic creation of the hype cycle. Yeah. Yeah. You literally created the hype cycle. So, so we have a huge poster. We have a huge blown up poster of the hype cycle. We think you can apply it to all things in life.

Yes. It is a fundamental truth. It is true. You know, if you're having like relationship problems, just, you know, pull up a pipe cycle, show your wife. We're going to get to the plateau of productivity. We're just in the trough of disillusionment right now. We are in the trouble. But we can see the plateau of productivity.

Yes. Yeah, no, I mean, the gardener was not my friend when I was running go-to-marketed trilogy, but, you know, Well, they, but certainly, I mean, that business has continued to do great. And so they made a set of early investments out of what you'd consider a small private equity fund, Gardner Group and Double Click and things like that. Even Staples, you can consider a venture investment. It was one store when your bank capital invested in it. So it was truly a startup.

And then, you know, the funds got bigger. And they started doing bigger deals. And as a result, that meant more mature companies. And it was clear that, you know, in order to do venture and early stage, it necessitated a separate fund. And, you know, the firm's been very entrepreneurial over the course of 30, 40 years now. You know, 40 years now, you know, we've launched a venture fund. We have a crypto fund. We have a life sciences fund, real estate, you know, kind of a whole range of things.

And that's always been the culture, which is let's create it. Each fund stands on its own. So we've got to go raise capital, you know, from limited partners. We make our own investment decisions, but we benefit from the affiliation with the rest of the firm. And we'll get into industrial renaissance. We'll get into AI in the physical world. But our access to the real world economy through Bain Capital, actual physical businesses, manufacturing companies and brands and retailers and financial services companies, that's where a lot of young people want to build today. and, you know, that's a huge source of our competitive advantage on the market. Yeah, before we get into all that good stuff.

How'd you meet Joe? Yeah, how'd you meet Joe? Yeah, I mean, normally we'd be like, oh, no one knows what we're talking about here. But, I mean, he's been on to invest like the best. He was in the journal. Like, this is Joe Lamont week. And so I got to ask. I mean, you know, for my entire career, people are like, you know, we've heard of trilogy.

Yeah. This week has been Pete Joe Limon. I met my sophomore year at Stanford, a professor of mine said, you got to meet this guy. He's starting a company, and you should join him. And so I went over to Tressor and sat outside with Joe LeMont. He was a year ahead of me. He was a junior. I was a sophomore.

So this was when he was like racking up credit card debt, like financing a business? Yeah, this is even before trilogy. So he hadn't started a trilogy yet. And he was starting a company. It was called Fourth Connection. and you know fourth connection is today what you'd consider a CRF sure but you know we would go around up and down El Camino and go to startups and tech companies and say wouldn't it be great if you had one database for all your quotes and contacts and proposals and people are like wow that's so

cool that doesn't exist we'd love to have that that was what we were selling in fourth connection and so when I first met Joe the very first meeting and sorry was that sales builder or was that No, that was all three trilogies. The sales builder was trilogy and, you know, and he eventually dropped out and started trilogy. Got it. Wow. So this was like, this was like, yeah, this was like, yeah, this is like, yeah, this is

1988, I think, you know, now I'm dating myself here. But he, you know, he said to me, you know what the greatest business on the planet is. And this is like the first five minutes, you know, and, and I said, I have no idea. And he said, it's enterprise software. Yeah. And I said, you know, what's enterprise software? Yeah, there you go. Yeah, there you are still, you know, 30, 35 years later.

It's undefeated. It's still enterprise software. And I said, why? And he said, it's the only business in the world that you build it once and your marginal profit is basically 100%. He's like, it's the greatest business in the world. He said, life sciences is sort of like that. But he's like, you don't know if it works for 10 years.

Enterprise software, you know within a year or two. if this thing's works. He's like, let's start an enterprise software company. We did this thing, Fourth Connection. We did that for a year. He dropped out. I just had not to drop out. And then he went on to start Trilogy.

I joined him a few years later as employee 18. So that was the Joe story. Yeah. I feel like there was a lot in the Invest, like the Best interview. The thing that stuck out to me was towards the end when he was talking about how AI will affect enterprise software and just like his overall playbook of reducing churn, switching a company from growth mode to retention mode. And his claim or his hot take was basically that in the age of AI, replatforming or ripping out a piece of legacy enterprise software will get easier.

And so the old school trilogy playbook might be a little bit harder to execute, not just because it's more competitive and everyone has heard about it and everyone's kind of run on their version of it. But just there might be some fundamental technological shift that's making it easier for companies to, you know, switch to other products. And I'm wondering how you see that trend if you agree with that or how you think that affects the next generation of enterprise software companies. Yeah, I think, I mean, I think the things about this era that are the same as the trilogy or in some ways it's it's almost a you know return return to the past is this idea of CEO selling big deal selling selling a vision um you know in in the late 90s we've this

enormous poll from CEO it's just like we're seeing today where CEO said I got to get on this internet thing I got to modernize I just wrote out laptops to all my salespeople and they're on the airplane just playing solitaire they need apps you know and I need to buy some apps for for my And so we would sell directly to the CEOs. And these contracts would be 10, 20, 30, $100 million contracts. And we're kind of back in that era now. And the best companies in AI are selling high to the CEOs. Every CEO, Fortune 2000, they've got some kind of AI counsel.

And they're saying, we're going to fast track all this stuff. We're not going to go through the normal procurement process and everything else. And so I think the companies that are winning today are in many ways able to sell this larger vision. to these large companies to say this is how you don't get left behind, you know, with the AI revolution. So I think that part has changed from kind of the SaaS PLG era, but in many ways return to sort of the 90s and Siebel trilogy and I2 and kind of that world. I think the part that will be interesting to see, and there's been this debate and Aaron Levy's talked about this, is will the systems of record get disrupted or will they get stronger?

Now, clearly, Aaron's got a, you know, biased point of view because he is a system of record, you know, at Box. But his view is the system of records are going to get stronger because they represent the ground truth. You know, models hallucinate, you know, AI's probabilistic. What is the ground truth that's going to guide these agents? And ultimately, is that the data that's inside SAP, you know, and inside Oracle and inside Workday? and will systems get stronger? I tend to be a big believer in systems of record. I mean, if you just look at the history of software,

the biggest application companies in the history of software have been systems of record. And that's just been true. Well, even if you look at Satya's, Satya, people were kind of dunking on Satya for posting this thread yesterday, but he was basically just demonstrating how powerful GPT-5 is within co-pilot because they already have all of your information.

they have your email, they have your calendar, they know who people are. It's just very easy to just like run prompts on top of the system of record and that's just an incredible advantage that Microsoft. Yeah, I mean the flip side of this, just to take like the counterpoint to Aaron at Box is, well, the business owns the data.

And so it's very hard for a company that is a system of record to say, no, actually I'm not going to let you suck all this data out of, you know, my installation into a data lake or into just a bunch of, you know, CSV files if you want, like, if I am paying for a system of record, I will demand that you let me print it out and take it out the door if I want, because that's my data. But, of course, there are different integrations that can be made a little bit more difficult. So I'm wondering if in the age of AI there's, like, the binding between even if

even if Salesforce or box doesn't let me with one click off board and on board into a different service, if I show up with my, you know, my McKinsey or my Bain and company and I have a team of Accenture, but it's all AI. And I say, hey, I want you to pull these eyes out one by one. Take screenshots if you need to. Yeah, one of my portfolio companies building agents for enterprise migrations. So yeah, I mean, can you wrestle with that? Like, how does that play into all this? Look, I think it's a great point. It's certainly even the last five years, more and more companies are putting, you know, trying to get all their data inside, you know, cloud data warehouse, snowflip, their data breaks

or whatever. And so I think that trend is continuing. But the question is for enterprises, where is that on the priority list? Totally. It's just not on the priority list. It's like, okay, that's there and it works. And I now want to move forward. And I want to take advantage of the next set of opportunities.

I'm going to implement the AI for call centers, AI for coding, AI for, you know, healthcare RCM billing, whatever it is. And do I really want to go through, you know, even though it might be possible today of taking all my data and epic and trying to push it somewhere else, it's just low on the priority list. You know, at this point, they've already suffered through the pain of the last 10 years of all these implementations. And so I think it's like a lot of things inside these enterprises are going to build on top. rather than try and rip and replace.

Yeah. One of the things that stuck out to me from the Joe interview was were everyone in Silicon Valley is familiar with the concept like built to grow, like you're in growth mode, go, go, go. Then there's obviously a lineage of companies that are built to last.

And he brings up kind of a controversial one, which is some companies are built to die and that they've kind of done all that they will do. They've built all the products that they will do and they are machines to kind of maintain their customers for as long as they can. but that particular product will not exist, and the people that work there

will migrate to other projects into portfolio. And I'm wondering if you have a unique perspective from your background, a trilogy, and then also the overlap between B.CVs in a unique position. I don't know how tight the firm is broadly, but like Brookside can take a company public. It's like the coolest thing possible.

And then you also have the private equity lineage there that could turn around a company or take it in a different direction. And I'm wondering if that brings you a different perspective when you're sitting on a board to have a real conversation with a with a with a founder that's not just hey swing for the fences and then you just if it if it's not a home run you're never going to hear for me as opposed to actually I can still help you make good decisions in the era where you're going public or where you're going and working with private equity or where you're winding the

company down like do you feel like you have a differentiated perspective there from other venture capitalists yeah I mean I think the reality is that um there are so many different ways to create value. And it's funny, you know, kind of in this era where venture firms are, you know, kind of, you know, there's a stampede towards becoming private equity firms, you know, and, you know, and so, like, I did that first. Yeah. Nobody, nobody. You know, we, we, we, we, we sort of understand that, that desire in capital. And so there's just a lot of different ways to, to, to generate value, to create value. And one of the benefits of being in bank capital is you see, all these different divisions and private equity and public equity and credit and it's incredibly

smart people that are partnering with companies to create value and these companies are different stages different dynamics different competitive situations and you know uh there i mean a lot of great companies end up getting bootstrapped to the point where then they might you know sell estate to to private equity that was the case we invested in survey monkey you know it was a bootstrap company and then, you know, we got involved as part of that business. And so I think that is just the reality, you know, a good friend of mine, founder, we backed early, you know, Nick at GainSight. We were part of that journey. And, you know, he decided that the growth profile of the business was still going to be very strong, but it was not going to be the 60 to 100 percent growth that,

you know, folks like us venture investors wanted. And so he, you know, orchestrated, you know, incredible exit to Vista where, you know, he was able to transition all the venture investors to a different type of investor that was in many ways ideally suited for that more mature phase of Gainsight's growth, you know, over the last, you know, four or five years. So I think that's an important thing. There's different ways to, you know, create lasting value. And I think founders oftentimes, you know, in social media are looking for a pattern or one way. There's just not one way. You know, it's, um, there are a lot of options out there. Yeah, on, on a similar note, I know, uh, I think it's rebranded Sankey advisors. The credit arm of Bain is now being capital

credit. Yeah, you should be Sankey now being capital credit. Got it. Yeah. Uh, and I'm wondering if, uh, if just having any interaction with those folks has, uh, given you a differentiated perspective on what's going on in the venture debt markets. We were reading the other day in the journal that there's been we're back to new highs where we're a drawdown volumes expanded but the amount of deals has has contracted also the number of mega dealers has increased yes yes so market just basically fragmented post yeah so i'd love to know your your you're thinking and what you're advising founders on when is the right time what is the shape of the business like what is the current meta for getting the most out of a venture debt partnership

from, you know, years ago, it was very different. How are you thinking about it now? Yeah, I mean, I think debt is very powerful, but it's also very dangerous. And the reality is that, you know, for an early stage company that's pre-product market fit, I think the best case use of venture debt is some runway extension. Yeah. And not even really runway extension such that you can sort of draw your equity down to zero. and so it buys you a few months.

But even when our companies have venture debt and their pre-product market, we advise them not to actually tap into the debt. It's there so they can sort of extend the runway a few months. I think that's one option for venture debt. I think the second is if it gives you a clear path to profitability, if you're a later stage company and the venture debt can avoid a dilutive equity around and you can get to profitably,

then I think it's very effective. I think for businesses which, you know, are rare in our world, but that have capital, you know, that a little more capital intensive, certainly credit can be a source of financing, working capital financing, you know, is effective. But your business has to be very predictable, and generally credit is best used when you're break-even or close to profitability. I mean, I think I remember, you know, we had this great company, Reich. that was, you know, a PLG company, kind of in the space of Asana and Monday and all those companies. Ultimately, you know, I think exited for north of $2 billion. But I remember Vista was looking at the company and they said, look, you know, given this level of ARR, we can take that amount of ARR and put 2X that number of debt on the business. You know, because the

business is so profitable, because it's generating cash, and that can allow that money to come out, you know, the shareholders. So there are a lot of interesting ways you can leverage debt. I just think you have to be very careful if you are not quite a predictable company that has a path to profitability. Then I think you've just got to be very selective. Yeah, it feels like something where you can get in a lot of trouble if it's if it's not a foregone conclusion that you can raise another equity round. Exactly. But if you're a place where you could always do an equity round, there's always demand at some price. But if you're ever going to be in a situation where there might not be enough demand to clear your next 12 months, 18 months of burn

at any price. You are in trouble, potentially. Give us. Oh, I was just going to say, we had Pete from Astronomer on the show. How, where does that rank in terms of like crazy wild cards in your career? I feel like it was, it just gripped the whole tech community. And I'd love to know a little bit of the play-by-play of like how you experience that internally because it feels like the most unexpected like story to ever emerge out of a company that is managing and well hand and well hand absolutely well handle yes but we

all know about apache airflow now no one no one had in their 2025 bingo card cold play or apache airflow exactly and gwenith paul drop but um you know i i got a call from you know my My partner, Enrique, who's on the board, you know, Thursday morning and you told me what happened. I hadn't, you know, been on social media. Wait, it wasn't Thursday, Thursday, Thursday, the concert was like Wednesday night, right? Concert was Wednesday night. Yeah, exactly. You get this call.

Hey, how's it going? So I get this call, and I'm like, oh my God, that's crazy. And so then I went social media and watched the video. And then later that afternoon, you know, I ran into my wife and I was like, oh, this thing happened to her company. And I was explaining, and I assumed it was sort of. of this thing, like in this narrow tech world that I live in, you know, and she was like, oh my God, everyone's talking about that video. It's got totally viral. My kids were picking me. And I've, you know, the part of it going viral in tech that's happened and good news, bad

news and various of our companies over time. But the part where it became part of the cultural zeitgeist for the entire country for literally three days, never experienced anything like that. And, you know, I think the board and the management team, Pete, obviously, as a founder, they, you know, they did a really good job of, you know, one being, you know, number one, you know, we're responsibility to the employees and our customers and let's do what's right. Number two, you know, this is obviously a cultural moment, but they're also real people involved and their families. and things like that. So I think they handled in a way that I think reflected a bit of that balancing act. And so, you know, I think you had Pete on the show.

It was great. And, you know, and I think the company came out the other side with more visibility. Certainly, I think our customer base feels very good about the business. The business, you know, was growing very rapidly. and has continued to do so. So, you know, we feel really good about the business. We feel good about the investment.

And, you know, thank God for Gwana. Yeah, as always. Absolute masterclass. Give us an overview of how you're thinking about what some would call American dynamism, but just this industrial renaissance, how you're thinking about what's important but maybe not investable versus what's important and investable from a venture standpoint. Well, we don't like to call it American dynamism, but we call it industrial

renaissance. But, you know, we think that, you know, it's a really important trend and opportunity. And certainly, you know, I think that there are enormous tailwinds, you know, that are, you know, heading in this direction. Certainly the advances in technology and computer vision and the sure cost of robotics and automation has come down. you know, the geopolitics, onshoreing, tariffs, labor shortages, all of that, I think, is creating the tailwind. Certainly, you've got similar set of dynamics if you extend into, you know, the military and Department of Defense. You know, our strategy here, you know, really starting with our investment in Kiva systems way back when, I think it's been to invest in companies

where the core innovation is still software. You know, and there are elements that involve hardware and robots and things in the physical world, but the core innovation is ultimately software. And I think he was a good illustration of this, you know, when Mick showed me the prototype way back when 2004, he said, you know, inside an Amazon warehouse, a worker is spending 90% of their time walking to the shelf

and 10% picking. And so he said, and he said, having a robot pick one of a million skews that are different shapes and sizes and weights and colors and t-shirts versus like a broom, he's like, that's an impossible problem. It won't be solved for at least a decade. You know, it's now been, you know, now 20 years, it's still not a solved problem. But he said the problem of getting a robot to move a shelf from point A to point B, he said that's a trivial problem. You know, and so his insight was instead of having the worker go to the shelf, I'm going to have the shelf come to the shelf. the worker. You know, and that alone, triple productivity, built this great company, powering all of Amazon today. It's really core to their fabric of how they run their warehouses. And so

we're looking for opportunities like that that are fundamentally soft with their heart. It's more of a systems approach. You know, really thinking about a problem holistically, these companies tend to be more full stack as opposed to just selling a widget of some kind. I think that also prevents you from being commoditized by low-cost, you know, offshore manufacturers trying to come in. So those are some of our thoughts, kind of in this industrial Renaissance world. So you've backed every humanoid robot company. No. Sounds like, sounds like, yeah, we were talking about this. Like, yeah.

You see the exquisite demo, and it must be a jump to try and understand, okay, can this person, are they just a genius and they can build one? Or are they best? building a system that delivers to the tune of a million. We were talking about this with SpaceX. Obviously tons of rockets blowing up. I think they're on ship 36, but like the goal is not just get one rocket to space. Like we did that, we went to the moon. The goal is like get it really cheap and reliable so it's going every 30 minutes.

Same thing with what we're seeing with Zipline. Those folks are just focused on like scaling and manufacturing, the Starlink stuff's the same. It's a completely different dynamic and problem that you're actually solving. Jordy, where do you want to go next? Good question. Yeah, I guess. Also, so, yeah, I think Aaron Slodov had a post yesterday, having some, he has a manufacturing tech company and this idea of like, basically, yeah, understanding specifically these

industrial renaissance companies, I have to imagine you're oftentimes meeting entrepreneurs where you're like, this seems like a good business. It should exist. But like what is your like actual investment criteria for a company that wants to make things in America? That's not not a company that is a system for making things, but but something where it's an actual physical product as the output. Yeah, I think if you're actually a core manual, as opposed to a technology company that's automating manufacturers, I think the bar is very high to invest as, you know, from a venture standpoint, because I think there are a couple of things you got to believe. I think number one, you got to believe there's truly proprietary technology. And that's just a high bar in manufacturing. I mean, you know, you've seen it, you know, with you guys were referencing the Rumba earlier and, you know, Shark Ninja. You know, like anytime, you know, every Dyson product, every, you know, product that's out there, they have some equivalent.

And Shark Manager is a great company. I'm not disparaging them at all. But, you know, you realize you're going to get competition very quickly as a manufacturer. And so the question is, what is that source of, you know, competitive differentiation? And oftentimes it's software, some kind of software or network effect or data, you know, that allows this business to make. maintain proprietary margins and a huge competitive advantage over time. And so I think you need that or you need just an extraordinary, you know, entrepreneur, a founder, you know, an Elon-esque, you know, founder.

I just think it takes a very special kind of founder to build a physical products company. You know, you're just dealing with the physical world is far less forgiving on so many dimensions. One, just making money, but two, having it work. On that note of software, is it only? Is it always software that enables, like, the bootstrapping of a network? I'm thinking of, like, Apple gets power from software, but not just software. It's like iMessage, integration, a network. NVIDIA gets power from software, but it's CUDA.

What is CUDA? It's a network of talent and standards. It's not just you copy CUDA and you're good or your feature parity. Tesla, they have self-driving. It's great as a single thing, but it really is like a scaled network that gets better the more people use it. Is there some sort of dynamic there that creates a flywheel effect? Absolutely.

Yeah, I think that flywheel is really important. You know, we're investors in this logistics company called ShipBob. Oh, yeah. And, you know, shipob is like a, you know, it's a tech company up and down. But when you think about why is Shipob so powerful, it is ultimately a network effect because we have so much density with so many warehouses. You know, we know, and we have so much data, we know that, you know, customer A, in New York City is going to order this product from this vendor, so we're going to make sure

that product is sitting in a warehouse on the East Coast. And by the way, we're not going to get that right 100% of the time. So the 30% we get it wrong. We're going to be able to fill up an entire truck and stuff from L.A. and get it to New York so that we can inject that package at the last mile and bring down the package, you know, the delivery cost by $2 or $3. And so it's a classic network affects business that's underpinned by data and software, but certainly relies on, in that example, relies on physical warehouses. You know, we have a certain number of physical where we don't own those, you know, those are through partner sites, but those are physical assets that are out there that create the density. We're using physical trucks to move things from

point A to point B, but ultimately our ability to do that is driven by data and software, you know, at the ground level. And those founders made a really interesting decision. which was we are going to partner with third-party warehouses, but they have to use our entire stack of software. We're not going to rely on their side. If they don't use our warehouse management, our pick and pack software, they don't adhere to all of that. We're not going to work with them because we need to have that, you know,

a paint of glass that sees every bit of information from soup to nut. So I think your point, John's a good one, which is, you know, these businesses that ultimately create some kind of network effect. Ultimately, you have that flywheel that keeps going and allows you even, you mentioned DoorDash, I think, on a previous, you know, podcasts a few days ago, these businesses that people think of as low margin, once you get that network effect and that flywheel, you can just compound forever. Well, thank you so much for joining. There's a ShipBob fan in the chat, spamming. ShipBob will be a 10 billion dollar company. Let's hear for ShipBob. Good to see you guys. Awesome. Have a great

Great, hang in, AJ. Have a good one. See you. Let me tell you about public.com investing for those to take it seriously. They got multi-asset investing, industry-leading yields, and they're trusted by millions. And our public.com update is that BightDance just made more money than META for the first time in history. Yeah, I think this is a boo because we root for Zuck. We root for America. We are team, Zuck, but you got to give it up.

But you got to give it up for big numbers. Congrats to the team over at BightDance, putting up some heroic numbers. $48 billion in Q2. to META's 47.5, just, just over 47 billion. Oh, it's got to be rough over there. Well, up next, we have the founders of Framer. Welcome to the stream. How do you guys?

Gentlemen, what's going on? Hey, good to be here. Big day. Big day. You guys are up late. I'm so, yeah, thanks for hopping out late. I'm so trained to when we have two people, they got to debate. I think we've got to get these guys to debate.

What are we debating? I don't know. We do that all day, so. Yeah, yeah, yeah, yeah, that's the life of a founder. Give us an intro, the update, and... I got the gong ready for you guys. Yeah, what is top of mind for you guys? What do you do with a gong?

Just wait. You'll find out. All right, here we go. All right, so the update is... So we're Framer. We're a web design platform for professional, or like professional web design platform where companies can run their entire.com.

And we raised series D that we announced today. Let's go. Wow, that's what it's for. That's what the gong's for. 100 on 2 billion. Wow, that is a big. Big number. Congratulations.

Big number. Are you guys in sales? We just celebrate, we just enjoy. Huge capitalism fans, honestly. Yeah, it's really just a love for business and big numbers. But yeah, it's a staple. And it's our pleasure to hit it whenever we can. But give us, give us, first time on the show, give us quick background on the company.

Yeah, yeah. You guys want to start. Yeah, so I'll go pretty quick. We're both product designers and we sold our first company to Facebook where we were a product designer for a couple of years. And after Facebook, we figured, hey, let's do another company before we get old. That's more than 10 years ago. And we always love building creative tools.

So we figured, okay, you know, like starters are pretty hard. Why don't we pick something that we really like working on? So let's build the next Adobe. And everybody around us said, like, don't do that. You'll probably raise some money, but not for design tools. Just don't tell any investor you'll be building that. But we did. And then 10 years later, and a big pivot, here we are.

Overnight success. Overnight success. Overnight success. I basically would Framer is really good at. Or, yeah, like, you know, I can talk about the product market fit. What we're really good at is we allow folks to basically designers ship websites themselves for companies. So you have your, let's say, your Figma where you do your design and you turn it into an application. You've got tools like Squarespace and Wix where you make your personal sites.

But a lot of company websites are still built with code. It's kind of slow. You need a developer team. And over the last couple of years, we saw more and more companies just wanting to ship sites, much faster and do it in a visual tools. That's what we built. What's your take on the vibe coding boom? Where, what are the, within, I mean, we're all familiar with the growth of vibe coding stuff.

What are the overrated use cases? What are the underrated use cases within the vibe coding boom? Yeah, I think we're about to find out. I think everybody's just really excited with where it's going to go. Because in the beginning of this year, I'm sure that if you tied. in, you know, like make me a nice front page for the company that you're at, then, you know, you're not really going to use that result yet, but you can see it's going to get there, right?

Like, we're close enough for the LLMs to, like, it's not going to win any awards. You can't really use it for your side yet, maybe for your personal side. So I think over the next year or two, somebody is going to make that work for companies. Somebody's going to make that work on brand. Somebody's basically going to allow designers to sort of like teach the LLMs how to generate just like an engineer kind of handholds an LLM to write the code. And I think we're kind of on the on the on the on the crisp of figuring that out. And that's obviously what we're spending a lot of time.

How how important is the is the kind of network around Framer? I know there's people that make businesses designing sites that can be adapted and things like that. Is that a core part of the business in the way that maybe sort of like the Shopify app ecosystem might be? or is it more in the direction that the platform is heading? Is it more about allowing people to generate, you know, sites entirely based on their own brand assets and things like that? 100%. So we call it creators and it's a whole economy by itself at this point.

People building more than just sites for clients, which is typically what you would think of is, you build a website for clients. Next to that, they build templates, they build plugins, they build all kinds of assets that can be sold on the framework marketplace. And we're starting to see people make serious money off of that. The most impressive examples are folks that basically start with a new Twitter account with zero followers, and then set out a challenge to get to 10K in, let's say, 30 days just by making sites and selling assets on our marketplace. Are they successful?

It sounds like they've already done this. Is that what you were saying? Yeah, yeah, yeah. This is happening every day. This is the whole thing now on design Twitter. And I think, yeah, there's a couple of outlier examples of people are now doing like three or 400K in year just building businesses on Framer by selling these assets. Yeah.

Are there features or functionalities that you think you want to add based on the generative AI boom? I'm thinking, like, we've seen these, like, not just, okay, build a landing page for a business, but build a game or use 3JS or WebGL. And there's, like, there's more tools that feel previously, it was like, okay, that's, if you're going to build a video game in the browser, like, that's going to take you a long time. And now people are vibe coding them. Are there other features or, you know, web components that you want to, like, surface more easily to, you know, prosumers? yeah if you have like a big web platform you can go a few ways like one is obviously the marketing sort of that we're trying to figure out just help startups ship the most beautiful websites really quickly that go very fast and then you can always go into e-commerce but all i see

is shopify i see a lot of people try but all i see shopify so you've got to be pretty confident to go after that and then the third thing is that there's something between an app and a site that people can't really make today right so there's websites we're really good at marketing websites now, but there's obviously websites that do more, right? So they're more like a database, a catalog, they could have a feed, they've log-in state. And that's, that was always constrained by people that could express that, could express that logic. And for that you had to code. And I think, you know, like we're in an interesting time where the LLMs can maybe help with that.

So it's definitely something that we're also going after. Yeah, how, how, it feels like your guys's strategy is, uh, the, the entire, market for creative tools today is extremely loud right every every day there's new tools launching you guys seem to be extremely focused still on on this core value prop of helping people make beautiful simple professional sites is how much how much what do you how do you think not to get too far out in the future but how do you expect like do you think the like my view is that the dot com is like a hyper durable asset that will continue to have a lot of value for for deck you know for the for the you know the next decade two decades et cetera and

there was a time when people thought that URLs and everything would just become applications and things like that and I don't think that's panned out how what's your guys's vision for basically the future of the website itself right at a at a time when you know already we're seeing in different tools, you know, different, you know, consumer LLMs that will generate UI and scrape websites and things like that. What's your kind of, what does the website look like in 10 years? And not like a chatbot. But you'd be surprised we'd be pitching obviously for this round. You'd be surprised how many VCs think that that might be the case. Yeah. But the web as a like as a form to present your brand and build brand equity just isn't

really going anywhere. But one of the things that we really like, I mean, we're both product designers. So obviously we have business goals. But if you ask me why I get up in the morning, it's also to make the web cooler. And the browsers can do a lot of really cool stuff, right? The LLMs unlock people to build stuff for that. And the website's kind of look boring. Like they've only gotten to look more boring over the last 10 years. So I think tools like framework can really help the designer sort of express more interesting websites. So I think that, you know, advent of more people being able to execute on that but as well as sort of like you know showing more like doing more cool stuff that the browser can do like combining 3d and video

on websites making much richer cooler experience i think that's really where things are going so i think it's only the browser still such an underrated sort of piece of technology in in terms of how it's used for presentation and if we give folks the tools to build like cooler output for that then they will yeah i think i think it says a lot that all you know every major AI company company wants to have a browser themselves or they want to buy Chrome, right? That to me says, you know, all you need to know, at least for now, around where the category is going. And it's pretty interesting that all the AI brands are very heavy on their personality on the internet, right? So they all have these websites and beautiful brands, really nice.coms.

Yep. Dotcoms forever. We're big dot com enthusiasts here. Huge enthusiasts. Thanks so much for hopping on, guys. Yeah, congratulations to you guys and the whole team. Thanks for having us. Thanks so much. Cheers. All right, cheers. Bye. See you.

Up next, we have a great.com, super.com. But before I tell you about that.com, let me tell you about adquick.com. Out of home advertising made easy and measurable. Say goodbye to the headaches of out of home advertising. Only ad quick combines technology, out of home expertise, and data to enable efficient, seamless ad buying across the globe. And we will bring in super. com with some news.

One of the Best.com. How are you doing? What's happening? What's latest? Good. You're doing well. Thanks for having me. Thanks for hopping on.

Give us the news. Well, first I saw your comment. Thank you for the domain shout-out. Of course. It's fantastic. Before we dive into the news, how did you get with any crazy stories getting the domain? Yeah, definitely crazy stories. So we had a different company name.

We were called Snap Travel. We wanted to rebrand. We did this whole activity and this whole exercise. And we liked the name Super. because it's how it made our customers feel. Like when they saved money, when they found a great deal, they felt super, and we loved it. It was such a generic word. If you Google Super, you're just like getting Super Marrier Brothers, you're getting whatever you're getting.

So we're like, okay, we've got to go and get the dot com. We reached out multiple times, didn't get any answers. I'm like, okay, forget this. Meta at one point launched some product called Super. And we're like, oh, they probably are going to try and buy it. And it was just like, we're not getting this domain. And then we got a domain broker. and out of nowhere they were like okay this person's like interested but it was uh put the intense

negotiation process it was kind of in the in the single digit millions uh and then it got to the board level uh and it was split and it was 50 50 half the people were like i don't think this is worth it people don't really care about domains anymore they just like search on google so not true it's so like people are people i mean domains are still i think wildly under price like dot com specifically and it's because people have just been so trained. If you've been on the internet for any amount of time, you know that when you are on the website of a big business, it's a short.com,

and when you're on the site of a less legit, less established business, and consumers are really smart. They just pick up on these things. And, yeah, it's interesting. So you guys, I imagine you guys have trademark super.com because you can't trademark like super or who knows. That's right.

That's right. So we're trademarked super.com. Anyways, the story was it was 3.3 in the board. And ultimately as a CEO, I get to make the hard decision. We're like, we're going to do this. So we bought the domain. You're right. I think for consumers, it does matter. Right.

Totally. Yeah. So the big news, we just wanted to introduce our incredible momentum. So we crossed over 200 million in annualized. Crazy. Crazy. And if you think of the evolution, so we've been around for about nine years now, We started the company and it was just like travel hotel deal.

We were doing hotels like four or five years. And then we like completely, we ran into super.com, expanded the product, introduced this membership program. And that's just allowed the business to completely take off. So we're now, you know, hundreds of thousands of paying members as part of this membership club to like save more, earn more, build credit. And just experience more of what life has to offer. So how is it, how is it trying to get, uh, coverage from the legacy media for just doing a lot of revenue. I feel like they still

are like, I don't know if it's a story. Like, come back to us when you raise like a million dollars. But you're like, ma'am, or sir, we're doing hundreds of millions of revenue. I think this is more notable. Yeah. No, it's so, it's so interesting. The traditional media, they don't, they don't have the time to like think about that. And they're just like, if it's a fundraise, I'm going to, I'm interested. If it's not a fundraise, I'm not interested. So, but there's a lot of companies are doing really well and driving a lot of revenue. And I'm happy you guys are talking to us. Yeah, what are the other business models in the category that you deliberately avoided, like lessons from like the previous cycle, previous strategies that

maybe like might have product market fit loosely, but not real sustainable business models? Yeah, that's a really good question. So let me take a little bit of a step back. So the company, super.com, we have a membership program called Super Plus. Super Plus is 15 bucks a month. And the idea is when you're a super plus member, you just get 15 to 20 benefits. So you can save on hotels. You can save on gas. You can save insurance. You can save on pharma. You can earn money playing games, filling out surveys. We have a master card that helps you build your credit score and earn 1% cash back. And some people come in, they use like one or two products. Some people use like five or seven products. But ultimately, it's like you think this is

worth 15 bucks a month or you are getting more than $50 a month worth of value. Right. And for some people that just I just book hotels like once a month and I'm saving you know thousands of dollars for other people it's I'm using the cash advanced product I'm earning money and building my credit score so everyone kind of uses the app somewhat differently and we use AI to kind of customize the app so you'll see what we think is most relevant to you that will give you the most value from your membership now you asked a question which is like what hasn't worked at right so we are like okay what can we add to the membership what can we try we tried

something about two years ago where we thought we could get into selling discounted like physical goods like actual products right yeah it's found and what we learned yes it's very very tough it's almost impossible to be with amazon but the hardest part was that you can't consistently get a large supply of discounted physical products because it's either it's like it comes and it goes and then it's like oh this is a hot item now and then it's not or it's like you get outdated stuff and then it sits in a warehouse and you couldn't keep that steady supply of discounted goods and it became a very unreliable customer experience so we actually acquired a company tried to build up the goods business had a physical warehouse in in Miami and then we ended up

just shutting that whole division down and shutting down that warehouse so not overnight success what's working on the customer acquisition side what where are you guys you know spending money on acquisition most aggressively and and then i have another kind of related question Sure. So each product has its own acquisition channels that work well, right? So we don't necessarily advertise the entire membership. It's a hey, come join in and get like 15 plus benefits. The way we acquire customers is we go for like one product and then they come in and they're like, oh, I'm going to become a member because this product is so worth it. And then hopefully they stick around and do more things. Right. So for hotels, it's a very high intent product. So what that means is it doesn't matter how much advertising you do it. or Facebook, it doesn't really work, right? What works is when you're on like Google or kayak or TripAdvisor and you're like

typing in like best hotel deals in New York and you have that intent to purchase. So like on the travel side, it's a very high intent channels. Yeah. If you have something like make money playing games, that actually works really well on social. Because people are scrolling on TikTok. What's the, what's the actual economic model for that? Because if anytime somebody says that I, that if you tell John, you can make money playing video games, I'm a little bit worried he's going to get on Super and have a little bit too much fun.

He likes video games. Yeah. So I'll tell you the model on that. And first of all, it's not just games, right? So I'll give you one that's really easy to understand. So if you come to super.com and you want to make some money and you go to earn, you may see something that says, if you've never taken an Uber before, download the app, take your first ride, and get 20 bucks. Pretty obvious, pretty simple.

So they'll go in, they'll download the app, they'll take their first ride, and we'll just deposit a $20 into their wallet. And for Uber, that's obviously a user acquisition strategy. They believe that customer. And Uber's probably paying us 20 bucks and we're just, you know, or 30 bucks and, you know, and we'll pass most of it back to the member. But for Uber, that's a user acquisition strategy. And if they can then get that customer to eventually take more rise and that,

that's worked out for them and they were happy to pay that 20 bucks. For game developers, it's kind of the same thing. It's like, you know, you get paid for like downloading the app and getting to a certain level. Now, if you get to that certain level, then potentially that game developers, developers generated some revenue from you because you've seen some ads, clicked on some ads, maybe you bought some gems, maybe you did something else. So it's just almost like an ad network play where it's just another source of user acquisition

for a lot of these companies. Are you guys, do you guys spend time thinking about how you're showing up, how the different products are showing up in LLMs? Is that a meaningful acquisition channel for you guys yet? It is. It is. So it's something that we're looking at closely. You've probably seen there's a ton of AI companies that's working to help you optimize how you show up in LLM.

So some of it is almost traditional SEO-ish, where you just kind of need to have that basic infrastructure and red folder structure and red information. But what we're seeing is that the LLMs are pulling from a lot of user-generated content as well. So things like Reddit and like TripAdvisor forums and, you know, even YouTube for that matter, like wherever you. user-generated content's happening. They're pulling from that as well. So it's some traditional SEO, and it's actually spending a lot more time and energy thinking about user-generated content and how we can help, you know, be part of and moderate some of those conversations. Totally. Great. Thank you so much for hopping on. Yeah, congrats on the milestone. Massive. Thank you. Appreciate it. More air horn. Enjoy the rest of your guys. Cheers. All right. Thank you guys.

Congrats. Let me tell you about Bezell. Getbezzle.com. Your bezelconcieres is available now to source you any watch of the planet. Seriously, any watch. Go to getbezzle.com. And we have another great dot com coming in to the studio, keychain.com, simplify your supply chain with AI-powered CPG. Good to meet you. How you doing? Thank you. Would you mind kicking us off with an introduction on yourself and the company and maybe some background and how you got into this particular business? Sure.

I'm O'Sheaun-Hanmerhan, co-founder and CEO of Keychain. I spent the last 14, 15 years building online marketplaces. I built a company called Handy. Handy's like Uber for Handyman and Cleaners. Filted it up to nine figures of revenue, sold it to Angie's list. Congrats. Yeah, that was a good outcome. He's the largest home service company.

Hey, let's give it up for good outcomes. Congrats. Sorry, continue. what gets a ding in a belt what are the what are the rules never know there are no rules it's whatever i feel like but keep keep keep keep it rolling we'll see i i feel some more i feel some more stuff coming um i i took on run and Angie so uh i'd be given the CEO of Angie and his public company sorry sorry I just had to mess with you I mean it is found your that's right yeah it was fitting in the founder seat it's good not enough not enough not enough not enough uh

founders do that, is get acquired and then take on the top spot. I don't know if I'd recommend it. It's a lot more work. I feel like when you sell the company, like there's this thing where you can just rest and vest. And that's what my co-founder wanted to do. He was like, could we not do all the things? You're like, get ready to be important at work, buddy. Every day, not just some days. Yeah, that's why. I advocate that for all my friends who get acquired. I say you should become the CEO of the acquiring company. should be the goal. But yeah, anyway, you found your way elsewhere.

Yeah, so I was a CEO, my co-founder, was Chief Revenue Officer, and we ran it all the way to the end of 22. And we both left, and we took a little break and then started another company. And we, you know, second-time founders tend to do this thing where they think they're going to start a studio company where it's going to have like four or five companies inside it

and they won't actually have to run many of them. We tried to do that very unsuccessfully for all of about a month and a half. And there was one idea inside there that just started to work really well. And we doubled down on it and really got behind it. And we've been running it for about a year and a half. And the idea is to build the platform that CPG runs on. So you guys know this, but most CPG companies don't run their own facilities. They have co-men's or co-producing facilities that run.

all of their production and there's a whole industry of about half a trillion dollars in the U.S. of 20,000 companies that make, you know, all the products that you eat, drink, wash your hair with, and we are building the ecosystem and platform for those companies to run on. So we're starting with search and discovery at Keychain and we've built the deepest database of how brands and retailers find manufacturers. Yeah. So I am like super intimately familiar with this space. I ran a consumer package goods company. And it was always a struggle finding finding supply chain partners. It was always some kind of like loose network of asking friends for favors and whatnot. The best in class database about a decade ago was just from a news website called BevNet. And they did,

they kind of aggregated up some manufacturers. I think they let people. run ads. And it wasn't their primary business. They were mostly like a news and reporting organization, a trade, a trade publication. So it makes a ton of sense. I would love to use this product if I was building a CPG company today. Yeah, it's funny. The process, somebody has an idea for a CPG brand, at least pre-keychain. And it was like, you just start asking your friends, people, random people. Who knows? Hey, how would you make ice cream? How would you make a protein? I met your co-founder Jordan during, I think probably back in 2020. because it was investing in random consumer brands, super sharp, a kid at the time, now clearly a grown man.

But yeah, this process is still. I mean, the challenge is in how do you make sure that the incremental consumer products entrepreneur and I would say it's easier to go acquire the customers that are making a lot of net new products or have like a product release schedule. It's also how do you make the incremental? CPG founder aware that keychain exists so they don't go through the traditional process of just asking friends and asking around, how do I make this thing? Yeah, I think you're right. And that's where we started. So we built the product for enterprise.

So at Handy and Angie, some of our biggest partners were the Walmarts of the world and the targets of the world and the Costco's of the world where we did their home service. So we had those relationships and they obviously make most of the, private label product in the country comes out of the biggest retailers. So we actually built the product for large private label brand owners, the people who, you know, churn out product every single day. And that's where we started. So today, eight of the top 10 retailers in the country use keychain. And the same is true, actually, of the large brands. So you think of the large strategic brands, some of them have invested in keychain like Hershey's and General Mills.

they obviously have a lot of facility that they make some of their own product but they also contract outsource a lot of it so we built the product for them to be able to search for manufacturers and it's funny so because we built the product for enterprise we almost got the startup for free like we got the fact that the startups are out there and they're constantly looking you know it's harder for them but because we built this data acid and you know I heard you in your last segment talking about SEO and LLM optimization. We started with the data asset, which is the deepest asset anywhere in the world of U.S. CPG manufacturing.

We've invested millions of dollars into building it. And it frankly can tell you who can make what better than any broker, any trade show, any, you know, phone a friend. Yeah, are you guys, do you feel like you're, is it trade shows that you're disrupting, right? I mean, if you wanted to, if you wanted to go, if you wanted to make. CBG products and your target or someone like that, you're going to a trade show and you're like, what can you make for me kind of thing? And this just seems a lot more efficient. I went to Expo West once and I almost overdosed on caffeine because it was like that year, every single product

just had maxed out caffeine. It was protein bars with caffeine and all that stuff. This is deeply bearish for Las Vegas. We need to get Taylor Swift to go to the sphere immediately because there's going to be no more trade shows after this is done. well we if you if you had gone last year i think you would have overdosed on uh psilocybin because everything seemed to have some form of like mushroom in it or you would have overdosed some protein there's always a trend protein protein psilocybin oh my god there you got you got it yeah for when you want a mild muscle while you hallucinate which is first but anyway um so we don't so much think about it as disrupting trade shows

because we really do think there's a role for trade shows and it's just not what we think. Trade shows are a truly awful way to do structured data discovery. And that's what some people do to trade shows. Like they walk around looking for a particular thing. Trade shows are a freaking amazing way to deepen a lot of relationships quickly. So if you know who you're meeting and you've done the research in advance on KeyChance, and you've set up conversations, you can meet the right 15 manufacturers

in 30-minute segments and have the best trip to Expo West ever or you can wander around getting absolutely loaded on caffeine because you're going from boot to booth to booth and have a pretty ridiculous time and maybe you meet two of the right people. So it's like how do you take structured data and figure out how to appropriately use the right tools

so that you can have a great trade show experience? And that's the first, that's effectively the first product the keychain is launched. So that product today is in month 18 and it does about a billion dollars worth of search volume per month. Wow. And we're in the process now of launching our second product, which is a bridge into the operating system for manufacturers. Yeah. Last question. We got to let you get out of here.

But we have one more sound effect. Do you have any fundraising news for us? We do. Yeah, we raised a little bit of money. How much did you raise? We've raised in total, the last round was 30 million. So we've raised $68 million. Oh, there we go. So it's almost like you've done this before.

The ding and the bell thing? Yeah, that gong hit is authentic. That's, uh, that one's a analog. We have a real gong. Oh, that one's real. That one's real. That one's real. Yeah.

That was in a video. of you doing that before no no no no we have a real gong yeah anyway thank you so much for hopping on the show awesome yeah great to meet you thank you so much guys we'll talk to you sir cheers be care bye uh that was a very cool segment um I would have killed for that in 2012 um anyway let me tell about wander find your happy place book of wander with inspiring reviews hotel great amenities dreamy beds top tier cleaning and 24-7 concier service it's a vacation home but better and up next we have a guest in the stream waiting room Let's bring him into the TDPN Ultra Dome.

John, welcome to the show. What's happening, guys? How are you? How are you? See you. Good. This is, I'm just looking, I mean, this is what the fortress looks like, man. Welcome to the fortress.

The fortress of finance. Welcome to the Capitol. Thank you for having me. We love you guys. We love you, too. It's phenomenal. Kick us off, which is a brief introduction on yourself, a little bit about your career. And then I want to go into the hope for common.

stock, the bulk case for startup common stock. Yeah, for sure. So I'm John Callahan. I've been an entrepreneur in venture capital's for a long time. We started True in 2005. We're an early stage success. There's also a stand effect. 20 year, 20 year overnight success. Yeah, exactly. That's the way it all is here in the valley. So yeah. So, you know, we we started the firm with a vision of, frankly, our tagline was more venture, less capital. We thought the world-needed venture capitalists, you were going to, you know, take big risks and entrepreneurs, you know, were coming out at that point, by the way, post sort of bubble crash, post-1.0 crash. There were a lot of new technology changes, Dhtml, open APIs. So all of a sudden, all these really

creative founders were coming up with really cool ideas. You guys remember mashups. And, you know, it was, it was, you could do a lot of things with all of this infrastructure that had been built in the 1.0. And you could build a lot of application layer value. that is a pattern I will talk to you about here about what we're seeing today. So that worked out really well for us. True's been extremely lucky. We've been investors in category defining companies like Fitbit, Peloton Ring down in L.A. Jamie Semenoff is awesome. Also in L.A. Sweet Green with John Neiman. Which ring? Which ring? Isn't a ring the security company? I can make a joke that say there is only one. There's only one.

Security company is only one. The one that protects you and your family and your neighborhood, range. Yes, yes, yes, of course. Yeah, the company is required by Amazon. Yeah, that's right, that's right. And Jamie's up there. I have a funny ring story. Someone on our team was dropping something at my house the other night.

Oh, yes. I'd see somebody with a hat on and a backpack on outside walking around my car. I was already in like about to go to sleep or whatever. And I fully thought Dylan on our team was somebody doing a break. break in the early years of that in terms of what jamie you know i think at one point they caught a criminal a day they did a big thing with the lap it was just awesome but also like the wild animals and all the create anyway yeah ring was great we're also big in enterprise uh we'll do a security hashy corp all sorts of other things we've had seven ipos 60 mnax so we've been doing

this a long time um and we're the biggest in our category of of pure seed four billion in capital 16 partners. It's all we do. What we do. It's amazing. Yeah. So take me through like the last few cycles from your perspective, what's changing around where the founder sits in value accrual, common stock, how you process ZERP and the last bubble and crash and all of that. Yeah, for sure. So that's kind of what caused me to write this piece. And it's like like all things I've been, you know, it, it, The good ones always take like an hour to write, you know, kind of thing and the other things that you kind of gnash around.

So I've been doing this a long time and have seen a lot of waves and we've been a part of a lot of waves. And the things that we are experiencing now, and I can talk through them around AI, but it's really capital efficiency, which we'll talk a lot about, that drives a lot of the sort of solution to the tragedy or the fixing of the tragedy for common. a capital efficiency, the size of the Tams, you know, the founder ethos today, your question right now, John, like, what's happening with founders? It's all leaning much more towards this, do more with less. The conversation now at, you know, Great White in L.A. or at Coupin, Palo Alto, or, you know, is, or Blue Bottle at South Park. It's all about, like, how much I can build with how few, you know, employees or capital. Sure. And that's a good thing. You know, you guys talked about the earnings release yesterday of Invidia,

but Jensen said yesterday there's two to three, I think he said three to four trillion of capex. Our numbers internally are sort of two to four trillion of capex going in to the AI wave over the next five years. So any other wave, Web2, mobile, cloud that we've been a part of, the value that is created at the application layer on top of that KAPX is on the order of 5 to 10x. So we're talking about new TAMs, like markets that haven't been invented yet that we don't

know about yet, net new consumer behaviors, net new enterprise behaviors at a level that is, you know, I don't say that like unprecedented, but like it's just really big. And as you guys have noted a lot on the show, the waves happening really fast, quite a bit fast. The waves kind of keep coming in faster increments. And we're just seeing that combination come together and create what I think is the vintage of a lifetime as a venture capitalist or as a founder. We've got the most powerful tools we've ever seen. On the TAMs, you know, like we talk a lot about software. It's one or two trillion dollars globally.

I think it's like 1.2 doubling in five years as a market. It's super big. Yeah. Services are 17 trillion today. Yeah. Doubling to 30. So like you have these. Yeah.

I want to push on that common point because I feel like you're you're correct that revenue per employee has never been higher. But I feel like with like maybe mid journey is the exception that proves the rule. But founders have not turned. turn their, turned a cold shoulder to preferred equity in this cycle. Like, there are plenty of founders who are saying, I'm fine selling 10% of this company every six months. And they've done very well.

15% every three months. And each deal has been accretive to the share price, potentially. But, you know, they are getting diluted. And I'm wondering if there's more nuance to that. Yeah. Yeah. More opportunity than ever to do more. with less, right? Yes. But at the same time, more opportunity to do more with more.

The opportunity is so large. There's a lot of reasons. If I can do more with less, imagine how much I can do it way more. I can do it way more. I think it's a matter of magnitude here in context. And you mentioned ZERB. So we have to kind of talk about ZERB. It's been super well covered. But like that distortion is at a level, you know, remember it was like a deck of corn a day was being crowned. There's three, again, trillion dollars of limited partner capital locked up in zombie unicorns, according to Bill Gurley's estimate, which I referenced in the, in the, in the piece and, and are we not, do you think this time is, do you think this time is different? I think that time was different. So the answer is no, I'm not, I'm counting on this time being

a time that is like other waves. The reason the distortion, by the way, like we should talk about it for a second. So, and here's why I think this time won't be ZERP. How's that for nuanced, right? There you. Yeah. Yeah. It's the same people. Like the founders, if you look at the senior leadership at startups and companies, they were the ones who, you know, were at different levels through ZERP. They've been stuck under the stack. I'm not saying zero equity. And in fact, if you look at the last couple of, I don't know, maybe four quarters, there's probably been 20 mega seed rounds, the $350 million seed round or whatever, 20. There have been 2,500 seed deals done first half this year. And, you know, it's power loss. You're going to have lots of failures and

stuff. But just when you look at ZERP on a magnitude basis, so much distortion happened. And I'm not saying there's not going to be those companies to your point that just keep raising and raising. But I think founders are going to be a lot smarter about post money valuation because most of, as you guys probably know, that the 14. There's a company that there's a company that I won't name that is a perfect example of this they've raised under a million dollars so single yeah hundreds of thousands of dollars they're doing millions of dollars in revenue they have logos from every single top AI company and they're just growing like a wheat like it's just clearly they're onto something and the founder has no real interest in raising he could probably like he feels like he can

take it as far as it can go just off of just raw customer demand you guys had weighed on i think last week right um the great example of a company that you know raised a little um but kept making the choice i think that term you guys talked about was super cool seed scaling i hadn't heard that but it's but it's what we're seeing so i guess john the answer is like i'm not saying no venture capital i'm just saying yeah of course that's me uh but i'm saying yeah and zapier zapier's funny because the the category that they're in which is like agentic workflows is like they're you know dominating you know having having been building this business for such a long time and yet I can think of like a hundred companies that have raised on

the same kind of core idea of what they're doing so it that category is like turning into a crazy capital war but hopefully Zapier is in a point right now given that they've been at it for so long that they can just keep playing their own game. Yeah, and in a world in which, if it's true that on top of all this CAP-X, all these net new companies and behaviors will be created, I just think there's a lot of room. One thing we've done as technology is all of us, we have always underestimated the potential,

and we've underestimated the tail, like both the curve and the tail of everything. Web 2 is still booming along. Mobile is still booming along at massive numbers. And so- Look at Oracle's stock price. Oracle is still crushing. It's got to be the MAG-8. It is.

Yeah. So there's, I think there's a lot there. And then the last point on the common is just that like, you know, that has fueled Silicon Valley, right? It is fueled Silicon Valley for all sorts of different outcomes to, you know, sprinkle wealth or gains down throughout a company's stack. And then those, I mean, literally I said in the process, this happened thousands of times in my career where that engineer, or, you know, VP or director, then takes more risk. And that's what we need. And that has fueled the Valley.

And so my thinking is that it's, it's that the ZERP distortion is what was different. But that Silicon Valley is, you know, it would never bet against the valley. It would never bet against rate of destruction. When, when companies, you know, as an investor, when I was looking at trying to figure out the why now investing in 2021 and 2020, And in fintech, it was like, okay, Stripe exists, Platt exists. You have these new kind of like infrastructure players that allow you, you know, there's better banking as a service platforms. But the Y now was like it wasn't necessarily, it didn't feel fully like a Y now.

And then you'd end up paying 100 times revenue for basically a lending company that like, yes, they benefited from like, you know, Plaid being, you know, like what it was at that time. in terms of, you know, market penetration, but it wasn't, you know, in the end, a lot of those businesses didn't really, they didn't end up really going anywhere, whereas I think having a real why now today, you know, it's not, founders very easily talk, and investors will talk themselves into being like, okay, they'll talk themselves into feeling like there's a why now, even if they're just kind of fabricating it. Yeah. Yeah. Yeah. I mean, the why now, you've talked a about it in the enterprise, but the why now in the enterprise has caused, you know, a lot of

new entrants to have great success against incumbents, right? That's super interesting. They're all sorts of new, again, this whole like, if you were talking about enterprise software a few years ago, kind of pre-AI, you'd look for these like tiny little verticals where you could have an advantage. And now the whole playing field is open to start up founders. And they're having incredible success, either in, you know, but you've had a bunch of them on your show. And I think that's also really interesting. Not without it. So it's coming from the customers, right?

It's coming from the buyers of technology, want this new thing. And incumbents have a hard time delivering that new thing. Not going to be smooth. Like, you know, I like how you're talking about where you're on the hype cycle. Like, it's going to be bumpy throughout the curve. But it's just a remarkable time in all of these markets. And I think that's spectacular. I mean, again, I think it's the chance of a lifetime for a builder.

to start something now with these dynamics. Yeah, I remember being, I mean, I graduated college in 2018 and feeling like I had, you know, missing mobile, right? Like not being an adult or like a massive tech trend like that felt, you know, I read the tech crunch headlines for years of all those companies, but not having been actually in the trenches feeling like, praying for another, praying for another bubble. Yeah.

Certainly got one with, with real, you know, potential that we're already saying. Well, thank you so much for hopping on. Great having you on, John. Awesome to see you. Hey, guys, we invited you to our thing in San Francisco, the Connected Stack. We'd love to have you. Oh, fantastic. Huge Enterprise AI thing.

Can't wait. You can do the show from the floor, whatever. Amazing. A lot of founders, whatever. Thank you for having me. It's awesome. We really appreciate what you guys are doing. Come back on again soon.

Yeah, we'll talk to you, Sarah. We'll follow up about that. Have a good one. See you. Bye. Up next, we're going over to air. com. Have you used this product before?

I used this and was obsessed. It's like, I don't know, I'll let him describe it. Should we be using it? Let's bring him out. We absolutely should, and that's why I was so excited to talk to it because it was like, you can pitch us directly on the show. Yeah. Use this product.

Sales call. Hopefully, we didn't keep waiting too long. How you doing? What's happening? Sorry, we're running behind. Great to meet you, Shane. Nice to meet. Yes. Thanks so much for having me. Yeah, thanks for hop it on. Can you just introduce yourself

the company, a little bit of the backstory quickly? Of course. It's great to me, Bill, Shane Hagdei. My co-founder, Tyler and I started a business called Air. It's an enterprise software company. We help creative teams manage all their media assets. So we help them store and organize all their content and then execute on the day-to-day workflows of making quality images and videos, putting them out in the world. Yeah. I was... The anti-slop company. Yeah. I mean, like, there is some, I feel like there's some crazy link between, like, your clients

or your company, your customers and, like, the brand, like, the culture at your company feels, like, particularly design forward. Like, I'm looking at your landing page, and it's incredible. I don't know, like, where this. They're talking to talk, and they're walking the walk. Yeah, exactly. Like, were you guys to sign this beforehand? Like, did, did this, was this something that, like, came from a particular?

person inside the organization is just like everyone in the company is so focused on design? I think, you know, look, if you're going to sell a product to creatives specifically, that's going to be your ICP, you really have to live and breathe the market. You have to put out great content on an ongoing basis to talk about the narrative of what you're doing, and you have to meet those expectations with how you deliver the product. The marketing site is just an extension of the products. And I think in many ways, it's come from an obsession about who our customer is and how we meet them where they are with what they care about. Yeah, talk about the marketing.

I've seen a few of the stunts that you guys have done, the latest one with the Rizzler. Like, what? Yeah, when did you know you wanted to work with the Rizzler? When did you know that you wanted to be the first enterprise software company to probably partner with him? It's just a wild choice. Look, I think if you look at. great generational enterprise software businesses. They all have a singular go-to-market motion

that contributes to 50% to 60% of growth. And every enterprise software business chooses a different channel to really focus on, depending on who their customer is, what they're good at, and where there's arbitrage in the market. For us, it's content marketing. And there's a nuance in how we approach content marketing that we call culture-led growth,

which is all fancy words to describe, the format of content marketing that we do. It's not HubSpot SEO or Gong social all the time. For us, it's great, you know, seminal campaigns that we put out in the world that espouse, you know, the narrative of what we do, whether it's product marketing or just a viral campaign with a, you know, 14-year-old. It doesn't matter. You're pushing a story out to the market and getting a group of folks who care about that to resonate with it and begin a conversation. with you. So that's how it took years to get here to a point where we knew working with the Rizzler or Kareem or Taylor the Wrenz or any of the other differences worked with

the Rory ones. But it's been a fantastic motion and wildly efficient for us. How much have you, you know, I appreciate, I remember when Air launched and I remember thinking, you know, I worked on, worked with a ton of different companies on, you know, marketing campaigns and stuff like that over the years. So I understood the value prop, even though I felt like it was almost a, it was a contrarian move to launch, like, what the original product, like, helped you just store assets. And a lot of people, I'm sure even VCs back then would have said,

well, like, people get Google, they get Google Drive off the, they're getting Google drive for free. It's going to be tough to compete. But, but clearly there was, there was demand for this. Now, how are you guys thinking about it? about not to ask like a very VC coded question, but I'm genuinely interested. How are you guys thinking about leveraging generative AI

if you house all the brand assets for a business and you can understand kind of like aesthetics of a brand, it puts you in a good position to not just store assets but help people generate new ideas? Just like automatically matching assets to tags and like creating like an ontology on top of the data. Totally. I think, you know, in this moment in time, there is a rush to value.

Everybody, every enterprise out there is trying to immediately get to the thing that's going to compound their business on an accelerated basis. The last conversation you guys were in, I love that tail end of that conversation around there's never been a more open environment in the enterprise for new entrants like us to come in and rebuild the entire sack of how a company operates. But when you look at different areas of the enterprise, in order to scale, you have to have or automate to your question around AI, you have to have a place to automate from. You know, set another way, slop in is slop out, and the richness of your solution is going to depend on the richness of the data that you build on top of. In every area of the enterprise, there has been a system of record that has structured and,

organize the data so that you can automate and centralize. This has been playing out for 25 years. It started with Salesforce building out what we now see as a CRM as a must have inside of a business if it's trying to scale sales. Same thing with development cycles with GitHub or marketing work with HubSpot for multi-touch attribution. There's never been a system of record built for creative work

or images, videos, PDFs, visual forms of data. At the highest level, if you asked a CMO or a creative director, how are your assets performing? What's performing best? Where are you going to invest in next? That's like a six-month job for an analyst. And at the lowest level, if you're trying to talk to a designer about what they do every day, they spend 75% of their time collecting content, approving content, sharing content, and doing all of that all over again. And so our ambition at Air is to build a system of record for creative work. It's one product. We're never going to build a second product at air, we're just going to go deeper and wider in what we do. And yes,

you got to start with storage. That's the hard problem to start with. You have to sink and organize millions. Today, we're up to about 150 million assets that are being managed in our product. But the magic is on what you can do on top of that. You know, the workflows that you can automate, whether that's rights management or whether that's like generating variance of something that you could plug in to different ad units, it shouldn't matter. It can all sit on top of the baseline architecture. You should create a workflow to help legacy brands not get canceled by Zoomers. So it's like it'll pop up an alert, hey, boomer, if you put this on social media, you're going to get canceled by the Zoomers. And I think that could save billions of dollars

of lost market cap. I hope that the, you know, it's funny, we had a hackathon that finished this morning. And one of the things that came out of it was a newsletter that goes out on an automated basis to admins inside of a workspace. And because if you're thinking about our product, about 15% of our customers have more users on air than they have full-time employees. Starts with your creative team, expands to marketing, sales, product, partnerships, your agencies, your influencers that you work with. And then the other dynamic is that, you know, about 11% of our customers spend more than 25 hours a week inside of our product. And so this newsletter propagates what all everybody did and spits out a summary of that with

all of the work to the admins in the workspace. And so visibility, I think, you know, to go to your point around boomers and zoomers and not getting canceled, visibility I think is really important. And our opportunity to do that in a system of record is no different than Salesforce providing a dashboard to your CRO to understand, like, which reps are performing the best or where to lean in segment-wise or why, you know, win rates have improved. Yeah, totally. It doesn't sense. Amazing. Anything else?

No, this was great. Come back on anytime you're the only person in the world so far that has made just storage interesting on the show. so anyway thanks so much for hopping on and yeah we're gonna we're gonna actually sign up for this because we were you're out of very we're very much yeah can we put you guys in the game here come on you guys are you guys are the sort of at beacon

yeah well we were we were iMessage scale where like our version of air right now is just the group chat yeah it's literally all I message it's a mass come on the intro to Soylent we're in the mix here you know put us in the game yeah let's do it thanks man great to me and I'll talk to you later All right. We're going to cap this off by just pulling up this post of what the Garmin horse tracker actually looks like. Okay. Yeah, yeah. You want to end with the Garmin horse tracker? This is the most important story of the day.

This is the most important story of the day for sure. The Garmin mid, it's, uh, yeah, I see it wraps around the horse's tail. Where, how else do you think you were tracking the horse's health, Jordy? Garmin is now selling a wearable you put on your horse. It goes on the tail, obviously. No, no, no. It's further down. Further down. So that's what it looks like when it's not on the horse, but when you apply it to the horse, while you're wearing your Garmin watch. So you can be riding your horse, check your horses vitals from your Garmin watch. That's the future I want to live in. That's horse power, baby. This is important. I mean, this is a trillion dollar company to me. They're sitting in. Look at this thing. Look at it on that horsetail tracker. What is it actually tracking here? Midnight. Outdoor training sessions, indoor training sessions, recovery, transport, max activity heart rate, it calculates the heart rate, how many strides, you can view the data live, and then sugar's got a bunch of strides. That's hilarious. This seems amazing. This is why we love technology. As much of a breakthrough as Waymo, you know, it's already.

Arguably more important. Arguably more important. Horses are self-driving. Thank you for tuning in today. Max Conrad from the substack chat. We need it. This time is different sound effect. I agree. This time is different.

This time is different. Yeah, there must be some clip from like a movie where someone said that or something. Where does that this time is different actually come from? Maybe we can get Warren Buffett saying it or something. I'm sure he said it at one of his meetings or something. Anyway, Thanks for you for hanging with us today.

We love you. We will see tomorrow. have a great a show mansion section oh you're ready cheers bye