TBPN Live - Masa's Epic Rise, New AI Paradigm, The Drone Race, $MSFT Earnings, Evolution of the Tech Bro

Episode Date: January 31, 2025

TBPN.com is made possible by:Ramp - https://ramp.comEight Sleep - https://eightsleep.com/tbpnWander - https://wander.com/tbpnPublic - https://public.comAdQuick - https://adquick.comBezel - https://getbezel.comFollow TBPN: https://TBPN.comhttps://x.com/tbpnhttps://podcasts.apple.com/us/podcast/technology-brothers/id1772360235https://youtube.com/@technologybrotherspod?si=lpk53xTE9WBEcIjV(02:55) - The New AI Paradigm (34:09) - Drone Horizons (Pirate Wires) (48:11) - Microsoft Earnings (01:02:42) - Masa's Epic Rise (01:57:25) - The Timeline

Welcome to Technology Brothers, the most profitable podcast in the world. Let's kick it off with a quote from David Senra. He is quoting sage advice from Li Lu, Charlie Munger's favorite investor, and he says, you can compound knowledge faster than money. If you truly love this game, I would suggest that you don't take shortcuts. It might take longer, but it's more rewarding. That's what we're trying to do today, compounding knowledge. Let's go uh more importantly than lilu david center is not feeling super well right now he's got a little bit of a cold send him a nice comment send him a dm or a comment and say heard the technology brothers

told me you were sick i hope you feel better hope you feel better and uh let's get him back in action the the rss feed needs him at the top of his game. Desperately. Even though David on a sick day is better than your next favorite podcaster on their best day. We still need him back in business. Speaking of next favorite podcasters, we got David Sachs here from the All In podcast. He still has the All In badge. He's monthly now, but he's still putting up huge numbers. He's got a great post here. He says, new report by leading semiconductor analyst Dylan Patelel. You know, we love Dylan Patel here on the

show. Shows that DeepSeek spent over $1 billion on its compute cluster. The widely reported $6 million number is highly misleading as it excludes CapEx and R&D and at best describes. Hey, we're going to just ignore CapEx and r&d and and every hey normally this is the most american thing you could do we we pioneered this they even stole the idea of faking your financials sam sam altman is actually pissed that they figured this out yeah beforehand this is just adjusted ebitda it all comes back to masa yeah and Adam Neumann. He pioneered the community-adjusted EBITDA. DeepSeek puts up a fantastic community-adjusted EBITDA, ripping out CapEx, ripping out AR&D.

And just ripping the American markets. I mean, you do take CapEx and AR&D out of EBITDA. Psychological warfare. That's the point, right? Yeah. And at best describes the cost of the final training run only. And so today we have the semi-analysis post here. Deep-seek debates, Chinese leadership on cost, true training costs, closed model margins, impacts, H100 prices soaring, subsidized inference pricing, export controls, MLA. And so if you're not already subscribed to semi-, you gotta do it folks. You got to do it

It's one of the most embarrassing things that can happen to you in Silicon Valley getting caught you looking at a at a At a semi analysis paywall. Yeah, that's the way to get just kicked off a board. Yeah. Yeah Disastrous what we say. Yeah, if you see a VC out in public walk up to them say pull up semi analysis right now And if they hit the paywall, it's just very, very negative signal. Oh, so you're not raising the next one. Like you're retiring? Yeah. Cool. Enjoy Aspen. Yeah. I hope you picked out some good stuff. So Dylan Patel writes, the DeepSeek narrative takes the world by storm. For the last week, DeepSeek has been the only topic that anyone in the world wants to talk about. Basically true. DeepSeek's daily traffic is now much higher

than Claude Perplexity and even Google's Gemini, which is crazy, but it makes sense. But is that the Chinese market? No, no, no, no, no. I think in America, because the traffic stats are always aggregated by American tracking pixels. But Gemini is super embarrassing

because it's a Google product. They should be stuffing it everywhere, but it's so hard to get to as we talked about before. But close watchers of the space, this is not exactly new news. We have been talking about DeepSeek for months. The company is not new, but the obsessive hype is.

Semi-analysis has long maintained that DeepSeek is extremely talented and the broader public in the United States has not cared. When the world finally paid attention, it did so in an obsessive hype cycle that doesn't reflect reality classic americans we get pumped love it we go to the moon we get fired up crash and sometimes forget how to go to the moon for sometimes decades yeah sometimes you're the world we run it back yeah we run it back masa stop yeah sometimes you're the world versus man and then you lose 99 of your money that happens We'll get to that later.

Very American. We want to highlight the narrative has flipped the last month when scaling laws were broken. We dispelled this myth. Now algorithmic improvement is too fast and this too is somehow bad for NVIDIA and GPUs. The narrative now is that DeepSeek is so efficient that we don't need more compute. Jevons paradox is closer to reality. The models have already induced demand with tangible effects to H100 and H200 pricing. So you can just go look at the spot market for renting these GPUs and you can see Jevons Paradox in action today. And so it's a really strong market signal. Yeah. And RAM saw this with their data, right? As the cost per token drop, the actual demand for them increased, spending on them increased substantially. Yep. And so High Flyer is the name to know. They're a Chinese hedge fund, and they started using AI in trading, and they invested in 10,000 A100 GPUs in 2021. This is

before the chip ban. They didn't do anything. Everything was above board, but they got a big cluster. It was a massive investment. I wonder how many trips to Singapore they've taken. Since then? Yeah. Well, that's the thing investment. I wonder how many trips to Singapore they've taken. Since then? Yeah. Well, that's the thing is that they might not have needed to because they just built up such a huge cluster. But it goes on to say they spent over half a billion dollars on GPUs with estimates pointing to 50,000 hopper GPUs across variations.

So there's the H20, the H800, which is the nerfed one with that memory bandwidth, which they got around by the way. And then the H100, which they can still get through the Singapore loophole and a bunch of other places. You can buy, I think a thousand or 2000 at a time. And so they, uh, there is an interesting angle here, which is that they truly do have goaded engineers on the team. Yeah.acked, I think was the term Dylan Patel used. They have a really crazy talent strategy. They aggressively recruit top Chinese university students, offering salaries over $1.3 million.

And I assume that's per year. Oh, per month? I think it's per year. But in America, some people are making that per month. Got to get your money up. With flexible roles and unfettered GPU access, staff count is 150 and rapidly growing. And so they have, there's an interesting kind of like a counter narrative to the whole like 996, grind it out, work really hard. They are much more like a research lab where it's just kind of people hanging out, trying stuff. There's really like, you know, open-minded,

try whatever you want, don't worry about cost, just run it and it's produced remarkable results. It's more of an American style. And there have been some pushback because, from folks on the DEI side, just because there was no non-Chinese employees, but they seem to have made that work. Yeah, yeah.

We'll see what happens. Maybe there'll be some pushback. I expect the New York Times to come after them pretty hard. Yeah. I would imagine Taylor Renz is writing something right now. So in terms of server capex, it's estimated at 1.6 billion plus almost a billion in operational costs. There's some estimates here, 10,000 H800s, 10,000 H100s with large orders of H20s in progress. H800 has the same compute as the H100, but less network bandwidth. The misleading 6 million figure was widely circulated as the cost of V3, but that doesn't include R&D, hardware depreciation, repeated experiments., actual total spend is far higher. But at the same time-

Wait, and just to break it down, R&D is the cost that they're spending on their team, which is undoubtedly one of, you know, like I would imagine in the billions of dollars a year, based around how many- I mean, they have $944 million in operational costs. So yeah, the OpEx is probably mostly R&D and salaries. They have 150 people making a million in operational costs. So yeah, that's the OPEX is probably the most these R&D and salaries. They have 150 people making a million dollars a year. And so right there, that's $150

million. That we know of. They also have been aggressively lying and now the truth is coming out a little bit, but we still don't know the full truth. I think this is probably correct. This feels correct and believable. Yeah. And so it's easy to lean into this and be like, you know, Dylan Patel is saying it's not 5 million or 6 million. It's actually closer to 500 million or a billion. And it's easy to be like, no, it's even higher. It's 50 billion.

It's probably not that high, but it is interesting. And so in terms of model performance, DeepSeq V3 is the base model, while R1 is the reasoning model. We've been through this. And they have a couple genuine innovations here. So these are the things that you need to see. These are the buzzwords that are gonna be going out

in the AI influencer space. So this weekend, start sprinkling them into conversations at home. They probably won't land but you'll sound smart and then come monday you'll be able to just sprinkle them in in the workplace and really start you know working them into the yeah you got to let people know that you you use these words so frequently you got to use the abbreviation so multi-token prediction mtp tokes tokes tokes multi-tokes prediction multi-tokes mixture of experts just call these

perts specialized sub-networks quote-unquote experts managed by advanced gating networks you could say oh yeah my my perts are gated i mean the funny thing is that real ones will know exactly what you mean yeah and so it's a good way to filter out who's who's LARPing and who's who's really you know at the at the forefront of of these broader trends yeah I mean like the like mixture of experts was was described in detail by George Hotz when GPT-4 dropped uh and he was saying that there were some hardware limitations to how large you could scale the models. And so they were effectively training six or 20 sub models and then wiring them all together with this mixture of experts thing. And so I've heard that term before, like outside of this. So that's not entirely new, but it does seem like combining all of these are what allowed them

to create a great model. And then multi-head latent attention, MLA, you're going to be hearing a lot about that, drastically reduces KB cache memory, 93.3% less, slashing inference costs. And that's obviously very important. And so at this point, V3 reportedly passes GPT-4.0 from May 2024, although AI evolves fast and many new models have also improved drastically since. And there's this reasoning shift going on where post-training synthetic data like RL is getting cheaper than giant pre-training runs, letting R1 quickly match others like O1. OpenWeights Advantage DeepSeek is now considered the single best OpenWeights lab, beating out Meta's Llama and Mistral. Although Zuck had a few things to say about that.

He claims that Llama 4 is going to blow him out of the water. And he's still investing in CapEx. And Zuck loves to deliver. Yep. He's done it many, many times. So that feels believable as well. And so. He kind of has to, too, to save face.

Yeah. He's got to drop something like truly competitive. Yeah. Not just. Oh, great. You have distribution and you'll get users for it. But it's not best in class with the amount that he's spending, which is something like 60 billion a year. He's got to show results as well as the meta executives that are all sitting there making five million plus a year yeah and uh facing that pressure internally to perform yeah i mean people i you got to go back in the timeline when meta dropped the first llama model because i remember the exact same hype cycle happening where people were really rooting against sam for a variety of reasons and people were rooting for

zuck for a variety of reasons because he was like on this giga chad arc and there was this whole this whole narrative of like oh zuck it would be a shame if i destroyed your entire company like opening i was going to zero now because llama's out there and it's free and you can just inference it and then i remember there was a runune tweet at the time saying, like, have you talked to that thing? Like, it's not that good. Like he was kind of playing like defense. Obviously it got better. And at this point it's, it's pretty good, but it had the same narrative as like, as like, oh, if there's an open source model, it's game over for the closed source models. And that didn't really play out because there was obviously the app layer and then all the reasoning. And yeah, OpenAI is clearly not just a model company. They're shipping products against the underlying model. And they do seem to be able to stay six months ahead or so. So V3

surpasses GPT-4.0, but GPT-4.0 is almost a year old now, May 2024. And it's crazy already, despite the mindshare dominance among outside of Teapot is insane with chat GPTs. Totally. Like I talked to a founder yesterday and she's like, it's acting as a co-founder for me. I'm like thinking out loud,

talking to it all day long about what I'm building. Yep. And yeah, she could go and use, you know, Lama or any of these alternative products. And it's just, for some reason, it's not getting traction in that same way. Yeah.

And so let's break down the direct comparison between R1 and O1, DeepSeek's model versus ChatGPT. Again, I think a lot of the narrative of DeepSeek just killed open AI was driven by the fact that it was free. It was free. Yeah, exactly. And so a lot of people hadn't upgraded to the $200 model 01 Pro. And so they were just like, well, I'm not going to spend $200. That's too much. It was the cost of the training run that everybody saw is really bearish because US companies are spending tens of millions of dollars.

And then there was the concern around a lot of people just trading against the news, not really trading against Jevons Paradox. And then there's the cost of like, OK, if you introduce a free product that's better than the free product of your competitor, what happens? But consumer behavior is not always entirely rational. And especially in the context, I just don't see, it's still hard for me to see DeepSeek becoming truly sticky and dominant. Because ChatGPT among consumers is a very loved product yeah and you can't underestimate consumer love because so many people have had magical experiences if you go hey you can use deep seek and it's better but they're like well i already love i already love chat gpt

exactly it does exactly what i want and it's learned a bit about me some of my data's in there already oh and i also there's a little share button. So when I generate something with ChatGPT, I can share the link that onboards you. And the value for us, it's like, okay, for $200 a month, we would happily for the same service pay two grand a month. So by telling us that we can get something that's comparable for free, the value trade-off is not that significant.

I only want the best, most cutting-edge model that has the most features at any given time. But a lot of people aren't in that boat. A lot of people do just want whatever's free. So is R1's performance up to par with O1? I want to read this pretty much, this whole paragraph here. On the other hand, R1 is able to achieve results

comparable to O1, and O1 was only announced in September. How has DeepSeq been able to catch up so fast? The answer is that reasoning is a new paradigm with faster iteration speeds and lower hanging fruit with meaningful gains for smaller amounts of compute than the previous paradigm. As outlined in our scaling laws report, the previous paradigm depended on pre-training, and that is becoming both more expensive and difficult to achieve robust gains with. The new paradigm, focused on reasoning capabilities through synthetic generation and RL in post-training on an existing model, allows for quicker gains with a lower price. The lower barrier to entry combined with easy optimization meant that DeepSeq was able to replicate O1 methods quicker than usual. As players figure out how to scale more in this new paradigm, we expect the time gap between matching capabilities to increase. Note that the R1 paper makes no mention of the compute used. This is not an accident. A significant

amount of compute is needed to generate synthetic data for post-training R1, not to mention RL. R1 is a very good model. We are not disputing this and catching up to the reasoning edge this quickly is objectively impressive. The fact that DeepSeek is Chinese and caught up with less resources makes it doubly impressive. But some of the benchmarks R1 mentions are also misleading. Comparing R1 to O1 is tricky because R1 specifically doesn't mention benchmarks that they are not leading in. And while R1 matches in reasoning performance, it is not a clear winner in every metric. And in many cases, it's worse than O1. This was your actual user experience where you were trying it and seeing, okay,

this output is not on par with what I get from chat GPT. And my prompt was not something that's baked into some math eval or physics test. It was just, can you summarize this book in 5,000 words? Did you give me 5,000 words? And was it factually accurate or not? And O1 Pro gave me exactly 5,000 words, a little bit more. It was great. And Deepsea gave me 1,000 words.

And it's like, how am I going to- You failed the test. It failed my prompt. A functional test. Exactly. And then and then as this is again, from a consumer standpoint, you're not going to notice these small intricacies in terms of, you know, having the model, you know, run at 50 physics problems because you wouldn't necessarily wouldn't have been able to get any of them yourself.

Right. So it's almost irrelevant. Yep. What matters is actually the output, which is what people want and what they're paying for around specific problems that they need the model for. Yeah, and this doesn't even include O3, the latest model from OpenAI that hasn't been fully released yet. O3 has significantly higher capabilities

than both R1 and or O1. In fact, OpenAI recently shared O3's results and the benchmark scaling is vertical. Deep learning has hit a wall, but a different kind. And so then there's also the question of Google's- Great line from Dylan. Google's reasoning- It looks like a wall, by the way. Yeah. It's completely up and to the right. It's fantastic. But it's on ArcAGI, SweeBench, and Frontier Math, Advanced Mathematics, like things that most consumers might not even care

about the benchmarks on, but are potentially good proxies, because a model that can do insanely advanced mathematics might be better at generating a recipe. And so Google jumped into the game to very little fanfare. They did not really break through. Really struggling on the marketing front. Totally.

And so while there is a frenzy of hype for R1, a $2.5 trillion US company released a reasoning model a month before for cheaper. Google's Gemini Flash 2.0 thinking. What a fantastic name, guys. I don't know why it didn't take off. You got 12 different words that no one knows what they mean it's possible that clippy is like no longer able to be trademarked so google should just take for its next model yeah uh so the model is available it's considerably considerably cheaper than r1 even with much larger context length for the model through the api uh and again

i'm signed up for gemini Advanced Pro Plus plan. I pay for it. I have no idea if I have access to Flash Thinking 2.0 and I have no idea how I even go and find it. If I Google it, I'm going to get the paper. I'm going to get some tweet. Like it's so hard to access Google's models. It's so embarrassing for them.

Get it together, guys. On reported benchmarks, Flash 2.0 thinking beats are one. Uh, I, I remember Peter had a fantastic, Peter Thiel had a fantastic, uh, like, uh, distillation of what's going on in AI where basically, uh, all the different AI labs are, you can think about them as like lemonade stands and they're all obsessed with perfecting the formula for lemonade. And none of them are thinking about where to actually place the lemonade stand. And when you're running a lemonade stand, all that matters is that you're on the busiest corner. Location, location, location. Yeah. And that speaks to distribution. Not even the corner, what neighborhood

you're in. Exactly. Exactly. And so you can imagine that at Google, they're like, Gemini Flash thinking 2.0, we beat this benchmark. It's amazing. Everyone's going to love it. It's like, that's not how products work, guys. That's just not, like, no one is, like, I'm going to go figure it out at some point, but it's going to take me 30 minutes to find this product and figure out that I'm using the right model. Yeah, and again, you don't, you have access to OpenAI's reasoning model, so you can just use that. It's easier to access. Your data's reasoning model. So you can just use that. Yeah, exactly. It's easier to access. Exactly. Your data's already there. Yeah.

And then now in every single Google product, there's some sort of pop-up saying like, hey, do you want to use Gemini for this? Do you want Gemini to work on summarize your emails? And I'm like, well, which Gemini? You guys have 17 different products. And like, how is it going to integrate? I know, we're dealing with this email

like related bug right now for TV related to a new domain. And everything is set up properly, but it's not working. And they just keep sending me the pop-ups of Gemini, and I'm like, you know. You can't get the basic level one. Maybe I should actually try it with Gemini. Hey, Gemini, fix this.

Fix this. Yeah. And so on reported benchmarks, Flash Thinking 2.0 beats R1, though benchmarks do not tell the whole story. Google only released three benchmarks, so it's still an incomplete picture. We think Google's model is robust, standing up to R1 in many ways,

but receiving none of the hype. This could be because of Google's lackluster go-to-market strategy and poor user experience. But also R1 is a Chinese surprise, which is a huge advantage. To be clear, none of this distracts from DeepSeek's remarkable achievements. DeepSeek's structure as a fast-moving, well-funded, smart, and focused startup is why it's beating giants like Meta in releasing a reasoning model.

And they did beat Meta. Like Lama 4 is still in training. They haven't released a reasoning model. And it certainly beat Meta, like Lama 4 is still in training. They haven't released a reasoning model and it certainly would be cool if this had just baked itself into- At least Zuck is breathing a sigh of relief knowing that they actually spent however many billions getting this model out.

Oh yeah. He's like, okay, we're not totally crazy. No, no, totally. We're spending 60. And I think that's why the stock popped in earnings. I mean, earnings were crazy, but I think everyone was very excited about his positioning on the AI infrastructure investment.

And so DeepSeek has cracked the code and unlocked innovation that leading labs have not been able to yet achieve. We expect that any published DeepSeek improvement will be copied by Western Labs almost immediately. What are these improvements? We covered a little bit of these training pre and post. V3 utilizes multi-token prediction at a scale not seen before. These are added attention modules,

which predict the next few tokens as opposed to a singular token. And so you can imagine, you know, the old version of autocomplete on the iPhone, it would say, you know, would you like to go to, and it would have like breakfast or lunch pop up, but now you see some of these models where it's predicting three or four words. It's kind of like that, but baked into the model at a very, very low level. So it's just more efficient,

because it says, oh yeah, after this sentence, usually there's just this three words that go together. And all of those go together. There are added considerations like doing FP8 accuracy and training. We talked about this. Leading labs have been doing FP8 training for some time. So that's a little bit overhyped. There's a lot of things where people just went through the deep seek paper and everything

that they were doing, they assumed was novel. And they were like, oh my God, they're using data to train it. Like no one thought of that. And it's like, no, everyone's using data. Everyone's scraping the web. Everyone's, you know, like running it on GPUs. And the synthetic data side as well. Yeah.

And so- And I don't know if this is gonna come up later in the show, but some lab at Cal was able to generate a reasoning model with 450 bucks. Yeah, I love that. So anybody working on models out there, we're good for our 450. Yeah, we're good for our 450. So rather than decreasing overall investment, mixture of experts, which is where one large model is comprised of many smaller models that specialize in different

things. And this is an emergent behavior. Historically, MOE models have faced struggles about how to determine which token goes to which sub model or expert. DeepSeq implemented a gating network that routed tokens to the right expert in a balanced way that did not detract from model performance. So basically you're getting,

instead of activating the entire model, which is very expensive, requires a lot of energy, you're just activating like one little section of the brain. But despite concerns that MOE efficiency gains might reduce investment, Dario over at Anthropic points out that the economic benefits of more capable AI models are so substantial that any cost savings

are quickly reinvested into building even larger models. This is Jevinin's paradox. Rather than decreasing overall investment, MOE's improved efficiency will accelerate AI scaling efforts. The companies are laser focused on scaling models to more compute and making them more efficient algorithmically.

In terms of R1, it benefited immensely from having a robust base model, V3. This is partially because of reinforcement learning. There were two focuses here, formatting, ensuring that it provides a coherent output, and helpfulness and harmfulness to ensure the model is useful. As mentioned in the Scaling Laws article, this is what happened with O1. Note that the R1 paper, no compute is mentioned. This is because mentioning how much compute was used would show that they have far more GPUs than their narrative

suggests. And NVIDIA wouldn't have had billions and hundreds of billions wiped out of its market cap. Yep. And Dylan also says that additionally, a portion of the data DeepSeek used seems to be data from OpenAI's models, which of course makes it easier because you're just cloning the outputs. And so you're optimizing against like a very set benchmark. If you want a GPT-4 model and you're just cloning the outputs. And so you're optimizing against like a very set benchmark. If you want a GPT-4 model and you're training on all GPT-4 data or a lot of GPT-4 data, you're going to get a lot closer to GPT-4 pretty quickly. Which again is interesting because they would have had to spend a ton of money with open AI, right? Which we've talked about before, the same thing that TikTok did to scale. They just spent billions on user acquisition

on Snapchat and Instagram. And it's this weird dilemma that the other platforms get in where they're like, well, we'll take the billions now. And even though we're going to lose out on, you know, potentially billions later. Yep. And so he concludes with a couple different takeaways. DeepSeek's cut rate pricing may be zero or negative margin, partially to gain market share and investor attention. And then he highlights Jevons' paradox again, talking about the AI race is similar to semiconductor fabrication,

cutting edge nodes, command premium margins, everything older quickly commoditizes. And that's like the story of tsmc versus intel uh the dynamic this dynamic spurs an endless cycle of scaling up new models and hardware investment benefiting chip providers like nvidia uh there are exports restrictions h100s are fully banned they've been replaced by h800s um h800s i think were banned and then the h20 was introduced more stringent u.s bands may come it depends where like we haven't seen exactly how the government will react to this

it's a really funny dynamic that the the our government doesn't want nvidia selling them great chips yep so why is it okay with them selling them any chips at all yeah and what you know assuming that they can. I mean, the steel man here is that if you're Nvidia and you're Jensen Wong, and you've been building this company that you control and own for 30 years, and for 29 of those years,

the government has been telling you, go global. We want you to be as big as possible. Get those tax revenues up, hire people. It's cool. These are our allies. The internet's going to make China a democracy. And they're part of the World Trade Organization. So absolutely go manufacture over there, do business with them.

We love this. We'll put you in touch with the ambassador of China. And then all of a sudden they run-pull you. That's kind of harsh. And so unwinding that relationship, it's not that crazy. And if NVIDIA had a blanket ban on selling to China or any countries in close enough proximity to China that they can act as a proxy, it would destroy, I mean, it would wipe potentially trillions out of the market. See, I might actually disagree with that because I think you're right that they

wouldn't be as high of a stock, but there's an interesting dynamic where the best time to deleverage from China was when there's insane demand in America. And that's what happened during the AI boom in 2023, 2024, where if there was ever a time to say, hey, we're going to pull back in China, it's when Elon Musk and, and Oracle and everyone is sitting down and being like, we will buy every single GPU you can make. Yeah. Yeah. So, so, uh, the alt man needs to, uh, he needs 2 million GPUs just in Avalon, right? Yeah, exactly. That's, that's, he'll buy them all. Nvidia is going to produce 6 million total this year. Yep. Right. So just one guy could, could potentially be a third of NVIDIA's. And he's not buying those

all this year. It's going to be spread out over time. I need a steel hat to put on when I'm steel manning. Because again, the steel man here of why would Jensen not want to do that? Obviously, people have been very critical of him not putting America first. But the flip side is, if you do say, okay, I'm only going to sell to these crazy American AI companies, and then there's a bubble and it pops, you are more fragile than if you have a more diversified revenue base. It's not like the United States and China have declared war on each other. It's just this great power conflict that kind of moves along slowly,

and everybody tries to kind of... We're like not allies, not enemies. We're also dependent on so much. I mean... Totally. And so this is an interesting stat. I hadn't heard about this, but the Bank of China announced a 1 trillion yuan,

140 billion US dollar AI subsidy. Let's ring the size gong for China. The one T. Hey. Size is size. Size is size. Size is size. Size gong is nationalistic in does is national size gong rings for everyone yeah um and if anything we hope the the the ringing sound inspires one of our listeners to invest a trillion

into something exactly and so 140 billion subsidy ai subsidy after meeting deep seeks founder so the guy literally goes and pitches the government. This all seems very conveniently timed, right? It's their Stargate. Yeah, you have Sam Altman gets Trump to do his press tour for him. Less than a week later, China meets with the DeepSeek founder and announces something very similar. Yep. But interestingly, this is the Bank of China putting up a subsidy of $140 billion.

Trump hasn't put up anything for other than just his name and a couple minutes in the PR room. Sam Aura farmed him. He Aura farmed him. Whereas the DeepSeek founder gold farmed him, basically. So China has a goal of self-reliance and advanced technologies, obviously, and this aligns with that. And so despite large GPU stocks scaling to serve millions of new users is tough. Future export controls may starve them of advanced chips. And there's a big question about, was this an aberration? Will they be able to keep up? A lot

of that's predicated on, can they, on, can they either get around the export controls reliably at an order of magnitude larger scale, or can they get SMIC and SME, their version of TSMC and ASML? Can they get those chips up and running and start making competitive GPUs internally? Because if they can do that, it's game over. Like they could make a billion chips.

And with something like $140 billion of demand. China is world-class at copying. Totally. And that's all they have to do. And I don't even say that in a way to slight them. No. But that is historically what they've done over and over and over and over. Yeah.

And they have ASML machines. They have TSMC machines. This is what entrepreneurs, you develop some hardware product or consumer product. You go to China, you develop some hardware product or consumer product, you go to China, you see your factory. If you walk to the factory next door, they'd probably be producing your exact item. Totally. Almost a one for one within a year if you're actually really selling well. Yep. So it's just ruthless. They're good at it. I'm sure they're going to have their own NVIDIA TSMC. Yeah. And so let's close with some of these conclusions. Efficiency does not mean reduced investment. This is Jevons' paradox.

Cost savings from new architectures like mixture of experts spur more scaling, not less. Fast follower. DeepSeq lags top labs, but closes gaps swiftly with new paradigm reasoning. Their open source stance accelerates industry-wide progress. God. Sick.

Frontier capability commands premiums, but DeepSeq's low pricing disrupts incumbents, potentially forcing big labs like open AI to innovate even faster. And we're seeing that, you know, there's going to be more pressure. Hey, release. Oh three, get it on the free tier, get an ad product in there, do something to bring this to the masses. So you can cut stem the bleeding of people installing the deep seek app, a long-term outlook, hardware demand isn't going away,ote, more advanced AI, more GPU spending, more scaling. Export controls will likely tighten, but we're going to need to see

where this shakes out. I think Jacob Helberg might be on the team that's working on export controls and tech deals globally. We actually need a rack of GPUs on the set. I think we should get some H100s and just replace that cache with a couple of GPUs. Yeah, great. And so we think the East versus West debate in AI models will continue, and the new paradigm is good for NVIDIA. Very interesting. So fantastic analysis from Dylan Patel. Highly recommend subscribing to Semi Analysis if you haven't already. It is a great resource and we love to see deep dives on the timeline. A great poster too.

Yeah, it's a great poster. Give him a follow. Follow him. DM him. The Technology Brothers sent me. Exactly, exactly. Comment on his post. So let's go to Drone Horizons,

a great piece in Pirate Wires, another outlet that you should definitely be subscribed to. This one by GB Rango, fantastic name. Great name. Pirate Wires says, Drone Horizons, our deep dive on how drones will change the future, not just in war, but in everything from construction, infrastructure,

and interplanetary space. The widespread adoption of UAV technology piggybacked on the inexorable AI surge, could significantly retool both international power relations and the daily lives of regular people in a manner that few understand. Today on PirateWire's GB Rango systematically outlines the trajectory of inevitable drone developments that are going to permeate every pore of society from agriculture to medicine to policing and national security and firefighting which is particularly relevant to the last month we had here in LA and so let's kick it off with GB Rango's lead which is a little anecdote here the stadium is a roar with 80,000 raucous attendees and casts a short

shadow in the midday sun. In the distance, peppered specks rise from nests unseen before gathering over the high horizon like iron filings in a magnetic field. Small payload drones, hundreds of them, race balefully forward. The orchestrated swarm twists and folds in hive-like unison as it approaches its target. In response, nearer to the stadium, another group of drones ascends in concert and rushes to meet its hostile counterpart. The confrontation is swift and casually anticlimactic, with only a few citygoers caring to notice the whir of propellers or the hissing and buzzing of the interceptors, high powered microwave pulses. The sky is clear once again, and the game goes on uninterrupted. The emergence and facile neutralization of drone threats is commonplace here

in a potential future. This cross section of a shocking and interterminate future and those to follow are meant to demonstrate and warn of the transformational nature of an intimate, of an imminent drone age shift. Dargstein. Pretty crazy. But I think this is one of those things where you see one side of the technology development

and you don't realize that there is the flip side that's countering it just as aggressively. Like in 2000 or the 2000 or the, you know, the two thousands, the 2010s, if every day you were reading articles about email spam and you had no knowledge about spam filtering and the fact that that was possible, you would be like, wow, why would I ever sign up for email? Like it's just going to get worse and worse and worse. But of course it's an arms race and there's anti-drone technology, which I think we'll talk about as well. And so I'm not quite as blackmailed as some people on this front. I also find it hard to imagine this future.

If you put it in the context of how dissimilar would it be if somebody launches a hypersonic at a stadium of people and we launch some type of countermeasure and stop it that would be national news yep so i would hope we don't experience this future because um and and and in this illustration the stadium doesn't even notice right uh but just very dark but um this is like the threat so it's worth you know really paying attention to and there are um so he breaks it into six or seven different sections. We'll give you a little tour of the article and you can go read the full post on pirate wires. It says an infrastructure and agriculture infrastructure development will undergo drastic changes with the advent and ubiquitous drone of advanced and ubiquitous drone technology construction site sites,

strangely devoid of human laborers, instead populated by powerful heavy lift drones, nimble inspector craft, and bipedal robots will be more productive, efficient, and safer than ever before. Gargantuan cranes with arms outstretched, fresh metal glinting through measured pivots rotate powerfully over the earth. He's such a good writer.

This is why I love this style of writing. It just takes you on like this tour. Yeah, you can really visualize it. It's like, it's really like pleasurable to read. Yeah, and it's already, can you imagine the drones, you know, working this construction site, they run into an issue, they need some type of human intervention

and you're the human that's on call and you get a call from this nice sounding robot voice and they're like, hey bud, you mind stopping by the site for a few we could really use a hand here it's like a human voice yeah and then uh you know hopefully after you finish as a human that the drones are nice enough to say hey thanks bud yeah uh probably won't need you anymore today but uh you know appreciate you stopping by yeah and um this is the flip right now. We tell, you know, we tell

the models what to do. Eventually they'll, they'll tell us what to do. We are the APIs. But, uh, I mean, there is obviously there's like, there's the job displacement question, but the, the flip side is like, I think every, almost every single time a major skyscraper is built, there is a single digit death toll from workers just because it's so dangerous and you put 10 000 people on a construction site for five years or something and like things are going to go wrong and so there is a pretty significant opportunity even even in oil and gas right yeah totally on these fracking fracking yeah every every now and then a video will go you know super viral and it's it's an you know, fracking site and just that kind of

action that's going on where there's heavy machinery whipping around. Super dangerous. Yeah. It's just super dangerous. So he says in the agriculture world in the coming age, farms will not sleep when the sun goes down. Drones will work around the clock to protect crop yields and monitor livestock. Rustic family barns sitting bare on flat plains will look like nocturnal beehives with luminous agents approaching for payloads before dutifully buzzing back into the dark. These small flying machines will flit nonstop through thousands of crop road acres, addressing issues like pest treatment, disease, and water shortage on the individual plant level. Very cool. We've got to get him to describe us preparing for the podcast.

What a great writer, seriously. get him to describe us. Yeah. What a great writer. Seriously. John threw on his jacket. Yeah. And yeah. And this next, this next segment on emergency medical services was actually highlighted on pirate wires.

Drone delivered organs should be standard protocol for transplants. And he says over 46,000 organ transplants were conducted in the United States in 2023. Quicker transport times are associated with better patient outcomes and in the sort of major delay described above is not unheard of. In 2023, a pro-Gaza ceasefire protest on the San Francisco-Oakland Bay Bridge blocked all westbound traffic for four hours. Three UCSF organ transplant deliveries were significantly delayed, one of which had to be

rerouted over the Golden Gate Bridge. But one transplant surgeon at UCSF noted that two other large transplant centers in the Bay Area, Stanford and CPMC, were probably suffering from the same issues as well. Now, while no known complications have been directly attributed to the protest, the proliferation of drone technology will completely eliminate the transport delay risks opposed. Yeah, I mean, a lot of this stuff happens via helicopter now. And there's even there's companies like Blade that actually have a really, I think at one point, Blade's organ transplant business was bigger than their consumer business. I had no idea. I think it was COVID.

Wow. Might have that off. But yeah, this is actually a very big business already, but it's kind of funny to think about, you know, needing to fly in Oregon from one side of a city or county to another and you just put it in a helicopter. Like it's not the most efficient if you're just basically needing to transport the equivalent of like an ice chest. Yep.

Yep. He goes on to explain search and rescue, maritime efforts, larger boats serving as bases of operation for small drone swarms that can expand radially outward and aid in the geographic process of elimination. Larger unmanned aircraft systems with multi-day endurance will work in concert with these teams to survey vast areas of open sea for survivors and identify points of interest. That just seems like a new capability. Yeah, this happens now. identify points of interest. That just seems like a new capability. Yeah. It doesn't seem like job replacement. You have a bunch of pilots that go up in their planes and they take turns just taking, you know, basically running. Yep.

There's some sort of like grid search they have to do, but it's a funny path. It's not. And they have to be high up enough to get enough coverage in a world where you can send a thousand drones. You can be much lower. Almost instantly scan huge, huge surface area. This would have been useful during hurricanes Helene and Milton, both of which struck in the fall of 2024 and drones were pivotal in the

preparation for, in response to these disasters. He goes on to explain that drones will have an impact on firefighting, which I think we discussed previously. Yep. And in America, fire department use of drones is almost exclusively limited to the realms of surveillance and intelligence gathering in the wake of the recent Los Angeles fires, which have reportedly resulted in the deaths of at least 24 people and wreaked over $250 billion of economic damage. This lack of progress.

What? Two weeks ago was at $150 billion. So it's still climbing. So it's climbing. climbing yeah uh compounding this international delta is the fact that across all u.s public safety agencies 90 of drones in use are designed and manufactured by dji a company that was recently added to the department of defense list of chinese military companies while that number is from a 2020 survey and steps are being taken to limit chinese drone, the status quo remains largely unchanged. So he goes into talk about foreign dominance and how drones will be effective in the military, talks about regulation

of drones and drone deliveries, as well as policing in the state, and finally, national security and war. National security, of course, is of paramount relevance in the discussion of drones. The nature of war is changing faster than it has at any point since the post-World War II nuclear proliferation, and drones are one of the major cruxes of this change. It is essential that America not rely on China for any segment of the defense supply chain, let alone UAS design and manufacturing. Yeah, and this goes back back i posted a while back and it resonated we needed a stargate for domestic drone production yep and the reason for this is that if we were in a conflict with china they could produce presumably millions of drones a month

and we can probably right now produce i don't know if we like needed to turn it on and produce locally yep but however, it would be a fraction of what China could produce and the capability would not be there, which is just a terrible place to be in. Like a huge part of the reason that we were successful in world war two was because we were able to turn on this massive, you know,

manufacturing capacity and just out, out produce our, uh, produce the competition. Yeah. This is interesting. He ends with space and visions of the future. Below the arcing of stars and spirited space adventuring of generations to come, the skies are clear, save the birds and an occasional transport drone passing

silently overhead, towering bio-hybrid buildings draped with verdant ivory attended to by swaths of assiduous airborne robo-gardeners looks as if they might have emerged in search of sunlight from the earth itself. A period of peace less fraught with tension than its Dr. Strangelove type predecessor has settled upon the world. This is not for want of bad actors,

but for their inability to find secrecy. And he closes with now, with all of this having come to pass, there seems to be an almost palpable optimism, a sense that things can change, and that we have a significant say in how they change. A weird coalition between humans and their hive-minded autonomous aircraft, once an unnerving prospect now powers the world forward in ways that not even the most visionary among us could have seen in a crystal ball ball. This or something else entirely lies ahead of us with courageous optimism. We step into drone filled horizons. So clearly GB Rango should write some science fiction because we'll read it.

So good. One thing that's interesting too, there's already technology that's been developed to make drones silent. So it's some element of the propeller design. Yeah, the propeller design. It's very crazy. So right now, it's actually nice that you can tell when a drone is. I've had situations where a drone flies over my backyard. Yeah.

I go out there and I look up at it. I'm like, get out of here. Get out of here. Scram. Scram. Scram. Yeah. Get off my lawn. I go out there and I look up at it and I'm like, Get out of here. Get out of here. Scram. Scram. Scram. There's not much you can do.

But imagine when they're completely silent. You're just, you know, somebody could fly it over your backyard on a battlefield. It's pretty terrifying to think about, you know, this kind of assassin that can move at 100 plus miles an hour instantly take you out without you even hearing about it so very dark but uh good time to get so one thing i would say uh kind of a call to action for the audience it's seen there are a lot of new american companies taking on this issue

developing drones for all these different use cases if you look at augustus and rainmaker they're developing drones for cloud seeding and agriculture we like that we could probably have 10 10 to 100 you know 50x more drone companies attacking all these different areas everything from like the production down to the parts level all the way up through these sort of like app layer product layer companies uh for organ transplants you know sorry not transplants transports transports yeah and things like that yeah yeah that's right yeah we'll go give it a read on pirate wires uh we love pirate wires and highly recommend that you subscribe it would be just crushingly embarrassing to get caught looking at a pirate wires paywall in front of someone in tech brutal you don't want that to happen to you. Yeah.

And if you ever run into Solana, he has the ability to just search your email in the database and he'll know. Yeah. So you can't fake it, guys. You can't fake it. You got to be real, real subscriber. So let's go to Stratechery, breaking down more tech earnings from big tech,

the hyperscalers. Microsoft and meta earnings went live and Ben Thompson broke it all down. And I know Wall Street Journal. I don't know what day it was, we already covered a lot of meta earnings. So we can kind of rip through Microsoft and just give you a quick update

from the Wall Street Journal. Microsoft's flagship cloud computing business experienced a slowdown in growth last quarter as constraints on data center supply once again weighed down results. Revenue for the tech giants, Azure Cloud Computing Division, which is closely watched by investors,

grew 31% at the low end of the company's projections. Chief Financial Officer Amy Hood said, Azure growth would again be 31% to 32% this quarter, a projection that disappointed investors. Microsoft stock was down 4.5% in after hours trading. Fortunately on public, you can get in on the after hours action now. They are allowing after hours trading.

So first off, Ben says, Microsoft said the AI growth was capacity limited, but would be rectified by mid year. It's nice to not have to be on the hook for that new open AI data center. But the bigger miss in Azure was somewhat obscure, but interesting in its own right. Basically summarizing the Wall Street Journal,

Ben says, Microsoft pivoted their scale motion. Think small and medium sized businesses and second tier geographies to be very focused on AI instead of boring old cloud migrations. And they didn't get many takers. So they're going to second tier geographies outside of the United States and they're going to small and medium businesses.

And instead of saying, hey, you have this data on this old server, let's get it on the cloud. They are now shifting to a model of saying, hey, can we get AI installed in your Teams integration? Can we get AI installed in looking at all your data, running streaming analytics, looking at your

enterprise. And that's a harder sell for a lot of businesses that are just, hey, like I kind of have, I'm running on vibes. Like I'm the AI. I know if revenue is down, I kind of know why. I don't necessarily need to dig through everything. I'd love to see an AI have a power lunch. Exactly. It's a harder sell. Apparently that's what we're seeing. Could turn around. We never know. So now they will go back to just selling actual cloud services and expect to see a revenue rebound count one for my thesis that bottoms up AI adoption will be a hard sell. And this is also why there's those private equity roll-ups, VC backed roll-ups that are, that are buying companies and then forcing them to use AI because it seems to be a hard sell from an outsider. And so you

could always ask, Hey, why is long Lake rolling up HOAs and then dropping AI on top of it? Why don't they just build AI for HOAs and then sell into the existing HOAs? Well, maybe the sales motion is very difficult. Maybe they want a new model. And so if they own the company, they get to convert them. My HOA management company still runs on paper sent in the mail that has to be sent back to them and then processed. It's like the easiest that has to be sent back to them and then processed. Yeah. It's like the easiest software integration to just allow online surveys. But if you haven't rolled that out. So imagine how many HOA software companies have pitched them. Look, you could be way more efficient. You could save money here, all this stuff. Don't care. Yeah. And they say,

oh, you're trying to pitch me some e-signing software. How many employees you got? 50,000? I'm not interested. It's going to be too expensive. Get those numbers up. Yeah. So on the flip side, Satya Nadella claimed good progress with co-pilot adoption. We are seeing accelerated customer adoption across all deal sizes as we win new Microsoft 365 co-pilot customers. Once again, however, Ben says there was no actual real numbers shared. I'd love to see them. Beyond that, I do want to express gratitude to Microsoft for introducing new terms of art, database fungibility. I'm surprised they're not getting pressed harder on that, considering it's such a major element of their AI strategy.

Wouldn't the analysts kind of push? No, you can totally push back. Until the SEC says you've got to break out the revenues, you don't want to share anything. This is the classic Google, got to break out the revenues. You don't want to share anything. This is the classic Google, YouTube, AWS, Amazon thing. Like Amazon did not want to break out AWS data because even though like you could hire someone from AWS and they could tell you like, yeah, that thing was really profitable. That doesn't ring as true as every day seeing the stock price and every quarter seeing AWS just print profit. And that gets Google and Microsoft off the couch. And they say, well, I know for a fact that AWS is

massively profitable. We got to build a competitor instead of just, oh, I've heard whispers that they're making lots of money. Very different. Because Bezos that whole time was reinvesting all the profits. So it wasn't showing up. It was hiding. It's just tough because on the on the copilot side, they're selling into existing customers. Right. So it's an upsell. Yep. Which is different than, OK, if you want to go compete with them really on copilot. In many ways, you have to own the underlying company wide people infrastructure to some degree. Right. Like you need to own teams so that you can upsell, you know, the co-pilot for teams integration.

Yeah, and so Satya continues to get hammered with questions about OpenAI and the relationship there. And he says... And they posted, so Altman and Nadella, Altman and Nadella posted a selfie together. Yep. Trying to, you know, clear you know clear hey we're buddies we're buddies we're smiling we're happy it's not it doesn't count unless there's the the the

the dela mogged altman on tv he said yeah the bar for like the the ceo to ceo selfie has gotten so high it's like i want to see you doing jiu jitsu together i want to see a jersey swap i want to see that leather jacket on you i want to see i want to see Satya in the passenger seat of the Regera. In the Konigsegg. I want to see shots of them on a yacht together. Hitting 200 miles an hour in that thing. And Satya scared out of his mind. Then I'll be like, okay, yeah, these are boys.

They're boys. They went to the track. They did a track day together. And Satya trusted Sam to put up five G's in the corner. Satya let Sam ride his favorite thoroughbred. Exactly. Exactly. Yeah.

It's not, you can't just take a picture in a boardroom. It's boring. It's boring. Step it up guys. So Satya answering about open AI, he says, but to your overall point,

I, the thing I would say is that we are building a pretty fungible fleet, right? Like, don't worry, we can move our fleet anywhere. We're making sure that there's the right balance

between training and inference. It's geo-distributed. We are working super hard on all the software oppositions, right? I mean, it's not just the software optimizations that come because of what DeepSeek has done,

but all the work we have done too, for example, reduce the price of GPT models over the years in partnership with OpenAI, which is true. They've done a lot of optimization to serve GPT through Azure.

Do we know exactly how much Microsoft owns of OpenAI? Because there was that one weird investment where they basically bought seemingly, I forget the exact structure, but it was like 10 on 20 or something. The way I think of it is like, there's kind of three important parties around the table with the Open AI for profit.

It's like the founders, employees, like the shareholders, the Microsoft and then the nonprofit. And then the actual investors are like pretty small in that. But you could think about like those four things and they're like roughly balanced. There isn't one person or one organization that has like outsized control. I think the whole meme of like,

it's entirely controlled by the nonprofit or it's entirely controlled by Sam or it's entirely controlled by Satya is maybe a little bit oversold and it's a little bit more of a balance because when the nonprofit conversion happened, there was like a big debate and you get four people around the table. Oftentimes if everyone has like roughly equal leverage to kind of veto, you kind of wind up with roughly equal positions. Um, but I don't really't really know. Anyway, in fact, we did a lot more work in inference optimizations on it, and that's been key to driving, right? One of the key things to note in AI is you don't just launch the frontier model, but if it's too expensive to serve, it's no good, right? It won't generate any demand. So you've got to have that optimization

so that inference costs are coming down. He's using some specific communication strategy where he keeps saying, right? Yeah. And it makes you inclined to agree with him, right? Yeah. Right. Yeah. A hundred percent. I was watching some video of someone, who was it? Someone on, it was Vivek Ramaswamy on mad money, on mad money with Jim Kramer. And he's breaking down this biotech deal. And every sentence he starts with, actually, he includes the word, actually like 25 times in this interview. And so Kramer will be like, Oh, like, why are you taking this, uh, drug through, uh, through phase three trials? Like it already failed phase two. And he's like, well, actually, when we looked at the data, like it actually outperformed. And so

he's like really enforcing that this is like true. What he's saying is a fact. It's actually true. And he's just throwing that in. And I just picked up on it as like a rhetorical technique. Got to watch out for these things, folks. It's important. We would never do something like this to you. We would never use rhetoric against you. Never. To get you to buy one of our- To sign up for RAMP. To sign up for RAMP. Actually. Actually.

Which is actually the greatest financial institution of all time. All time. Also, the fleet physics that we are managing. And also, remember, you don't want to buy too much of anything at one time because Moore's Law every year is going to give you 2x. Your optimization is going to give you 10x. You want to continuously upgrade the fleet, modernize the fleet, age of the fleet. And at the end of the day, you have the right ratio of monetization and demand-driven monetization to what you think of as the right training expense.

I like the use of fleet because it does. You can imagine that same language being used at amazon where they're like all right jeff you have you know tens of thousands of these delivery vehicles like the depreciation is insane you're putting hundreds of thousands of miles on them every few years like they're clearly depreciating massively and so by positioning it as a fleet standpoint it makes it slightly less painful that a lot of these gpus will be worthless in three years yeah not worthless people are like yeah i i got a fleet of sports cars at home i'm used to this i've dealt with depreciation before you buy the right cars i got an sf90 yeah i'm down i got a bunch of ticons yeah in my fleet i'm not i'm sitting on losses

the only thing that depreciates faster than GPUs is Ticons. Ticons. Some people call them the GPUs of the car market. Yep. And so Ben summarizes this by saying, this is the exact opposite way in which Microsoft was talking about their data center investments in October 2023, where Nadella emphasized that Microsoft would have a durable cost advantage because they could build infrastructure for a specific model. So they were not talking about fleet fungibility at that point. They were talking about going all in on open AI. Ben doesn't forget. He's putting Satya in the truth zone. In the truth zone.

And you know Satya's reading this being like, oh, he got me. He got me. Yeah, yeah. I was trying to be a little sneaky, like a dead sneak past your tech rig. Oh, no. Ultimately, though, that's not a criticism. Mod models seem to be well on their way to commoditization and microsoft has changed its tune appropriately and so yeah and they're hosting r1 for free and so uh you know wait microsoft where credits microsoft is hosting r1 everyone's hosting r1 no one is as much of a jangoist as you no one no one is that america first like the whole like the abstract concept of like it's bad to host the open source model just like that didn't take hold at all as a me yeah

just i you know imagine imagine sam you're you're posting the smiling picture with satya meanwhile he's but they host everything they host llama they do a whole press release i understand i understand yeah it's just and if you're an azure customer you're like i want all of them like give me all the flexibility all the time on cost oh oh like they're hosting it on azure on azure i thought i thought it was an integration in teams or like no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no no other databases. I thought they were like, yeah, load it into Copilot. Sorry, Sam. Sam's got the

smiling face on. He's just crying. No, no, no. It's not that aggressive yet. And one more point of note, Hood, the CFO said, we talked a little bit about what were the main drivers, which was one of the Azure commitments that OpenAI has made. And I do want to say that obviously this is a big commitment. You'll continue to see OpenAI make commitment. So I want to separate the concept that this is one time versus an ongoing relationship that will grow as they grow, which it absolutely will. And Ben says that right of first refusal to OpenAI's AI build out figures to be a pretty handy tool for Microsoft. They can build capacity for their customers.

And if they overbuild, well, guess they'll grab a bit more of OpenAI capacity to fill in the gaps. And so Satya, you know, getting a little tricky with the words, but an absolute dog. He's nice with it. He's nice with it.

He's nice with it. You know, he was the first hyperscaler big tech CEO to mention the metaverse before Zuck. No way. He got to it before Zuck. He's nice with it. He's nice with it. You know he was the first hyperscaler big tech CEO to mention the metaverse before Zuck. No way. He got to it before Zuck. Wow. He was like, yeah, we're really into this metaverse concept.

Like it's gonna be a big deal. We're gonna like. It's called Teams. It's called Teams. But yeah, it was like, you're gonna be, if you're like an industrial company, you'll be able to create a, what do they call it, like a clone of your industrial systems.

So you'll be able to like walk around in virtual reality and see, okay, this machine is green. This machine's red. Let's diagnose it. And it's called Digital Twin. We're running out of Lucy. Digital Twin. Yeah, yeah. Let's fire up the packaging machine.

Print more labels. Yeah. All digital. And I think they have some products for it. I don't think the adoption's been great, but it is just funny that he was early to the trend. He was reading Matt Ball. Yep. Another absolute dog. Okay, let's move on.

What is Matt Ball up to these days? Oh, he just put out a fantastic report on the state of video games, which we should dig into. We should dig into that. It's like a massive slide deck, and he just did an interview with none other than Ben Thompson, the absolute brand. Another absolute dog on dog action. It's a dog fight. It's a dog fight.'s a dog those two guys get in the room yeah yeah

hot takes flying yep it's great okay let's move on to masa yoshi zone hey crack i uh i specifically ordered these uh this morning this is my favorite little japanese soda it's called the kimono okay sparkling yuzu soda it's made in Japan it actually still has all of the it has some westernized labeling in order to sell here but they left the Japanese characters on the side too so if you can read Japanese you can dive in there but these things are delightful I thought it was a great way to kind of sort of kick off this um overview of masa sons masayoshi sun's uh absolutely incredible career and that we are blessed is still raging roaring on the ultimate size lord the gambling man the gambling gambling man there's there's so there's uh john had uh covered masa on on his youtube channel in the past but uh in in diving into this uh so we we looked at some of his research from

that and then i went back and uh started listening to the audiobook uh gambling man which is an incredible overview of masa's life and gets you really into who this guy is, everything like deep family history, going back to his dad, his dad's family, his grandpa, their entire trajectory going from Korea to Japan, back to Korea, back to Japan, going through a very tumultuous series of decades leading up to Masa being born. Well, let's start with some context of why he's in the news and then we'll

kick it off with the proper intro. So Jeff Lewis has a great post here that we'll put up on the screen. He says, you only have to be really right once. Masa has been really right twice. Third time tends to be a charm. And so Jeff's going long Masa. And Hiro Onoda Capital says, 25 at 340 post. This is the rumored open AI round. And it's a quote. It says, there are one business guys.

Bill Gates just started Microsoft and Mark Zuckerberg just started Facebook. I am involved in a hundred businesses and I control the entire tech ecosystem. Masa continued, these are not my peers. The right comparison for me is Napoleon or Genghis Khan or Emperor Qin, builder of the Great Wall of China.

I am not a CEO. I am building an empire. And that's the type of energy we love to see on the timeline. And so we got some photos of Masa, but let's kick it off with why you should care about this guy. He's back. He's in on Stargate. He's ripping huge checks. He lost a ton of money. It was embarrassed by WeWork, but he made a ton back on arm. He's rich, he's powerful, and he's firmly back in the game. And so why don't you kick it off with his humble beginnings and early life? Humble beginning is an early life. So he was born on the 11th of August in 1957 in Tosu, which is the southeastern part of Japan's Kyushu Island. He is a third generation Korean. Zainichi Korean. Zainichi

Korean, which are ethnic Koreans that were at some point or another granted permanent residency or citizenship in Japan. So before going into that, he actually, so going back to his grandpa, I'll pull up some notes here. His grandpa and his, I guess, his great grandpa were a part of this sort of agricultural class in Korea. And so they were not seen as businessmen, but they were seen as people that sort of, they were sort of aristocratic, aristocrat adjacent. Their responsibility was sort of overseeing agriculture in Korea. And what's interesting about this is that his grandpa really internalized this. And so even though that aristocratic

sort of agricultural class was disbanded in Korea and they eventually moved to Japan, he always saw himself as above business, like had a very negative view of business. He had a negative view of entrepreneurs, which is fascinating because Masa's dad, his grandpa's son, the grandpa's son, uh, was very much an entrepreneur and he went on to very much influence, um, Masa, uh, in, you know, in a big way. It's a real catalyze, just do things to real. You can just do things guys. So, he can so yeah so going going back uh masa's dad was born in 1936 uh

they're they're um masa uh he was just 21 when he had masa yeah so has a kid very young masa's one of seven um natalist so his his his yeah very very much um and so uh moss's dad is living through uh world war uh you know the sort of the end of world war ii right sort of this japanese reconstruction era the japanese were deeply embarrassed by world war ii right they went from this sort of dominant empire that stretched all over asia to being forced back onto the four islands. And they went through this reconstruction period. And so as the Japanese were in a bad spot, they badly, badly mistreated Koreans. Even during the war, the Koreans were,

the men were forced to work in like coal mines and do these sort of tasks that were basically slave labor. And the women were like concubines, sort of sexual slaves. So very, very dark history. And so this is what Masa's grandpa is like going through as like an adult, basically. So he has a son, he has seven kids.

And Masa's dad even has memories of American B-29 bombers flying and bombing the nearby cities that they were in. And Moss's dad has this memory of walking up and touching the glass of a B-29 bomber that crashed. But Moss's dad at the time is a kid. He decided that America was awesome. He says like there's a quote, he says they he's like at that point from that point on i i thought america rocked that was like the summary you know not not word for word that's hilarious um and so uh after after the war koreans are being treated terribly there was there was a lot of pushback many of them were asked to leave if you didn't leave you were forced into these sort of

shanty, you know, ghettos, slums. Japan's always been kind of a like light ethno state. Like they've never been big on immigration. Yeah. They're kind of changing their tune now. You can like go live there,

but you, yeah, it's hard to get citizenship. The population is collapsing. So things might change, but it's been rough. Even, even being Korean,

not even being like an American who would really stick out. Yeah. And so um they lived in this illegally built shack on the railroad you saw this so gnarly and so at one point they go so masa's dad uh goes back to korea they realize it's even worse there there's like there's no economy yep then they decide they're going to come back to Japan. They get on a boat there. It's like it's a multi-day crossing. Typically, it's like a couple of days. They the boat sinks and they spend two days at sea just like drifting around and eventually get picked up, taken back to Korea.

And then they finally make it back to Japan. Wow. And so jumping forward a little bit, Japan at this point is still like a completely shattered economy. They're rebuilding. They've lost so much ground. Masa's dad leaves school at 14 and is selling alcohol to basically support his family. He's one of seven. He's trying to put some food on the table. He's an illegal sake business. So Masa's dad starts a moonshine business. And at this point, his dad is a drunk, is beating him like frequently, very dark. And it has at one point badly beats him because he was

bragging. Masa's dad was bragging to his friends about his business success. Right. He's like, I'm making moonshine and making some bucks. His dad overhears it uh ends up beating him for it um so Masa's born um 10 years after the family gets back to Japan um they're completely at the bottom of society so it's it's very much like Masa by the time he's a teenager his dad is kind of on top not not on top of the world but they have a very flashy lifestyle, but he's born while his dad is living with their family in the slums. And, uh, they are pig farmers and they chose pigs for a few specific reasons that I thought were interesting. So they, um, one, his dad had free labor cause they have a big family and everybody's participating. Um, they have free land because they're just living in this sort of like railroad ghetto and the pigs just kind of roam around

chilling. The Masa's grandma, who is a big part of raising him because his mom was very absent, takes Masa around in a wheelbarrow picking up scraps of food to feed the pigs. So this is just like it's hard to even illustrate how rough of a situation they're in. But anyways, they pick, his dad picks pigs because they reproduce super quickly. The pigs are just, you know. It's like very hustle mindset, but very limited opportunities and like low class.

Yeah, and so eventually through this sort of raising pigs, selling them off to be butchered for food, Moss's dad starts saving up some money. At one point, he has the equivalent of 100,000 USD, a lot of money back then. And Masa, this really gets into kind of understanding who he is. His dad starts to get into this thing called pachinko, which is this new form of gambling, which is effectively via slot machines. So this still is very much around in Japan. So his dad just starts getting into the game a little bit. He's gone from pig farming to gambling.

Yep. An absolute dog. But he's on the casino side. But he's on the casino side. He's on the house. He's not gambling to take the money away. This is important. But he still has to take some pretty big gambles.

So anyway, so these are picture like casinos and is not that dissimilar to what Vegas looks like today, where it's just rows and rows of these machines and people just sit there. And over time, he's sort of levering up. He's launching different, you know, parlors here and there. At one point, he takes a huge bet and spends four million dollars on a new on a new club, they called it and it's called lions. And so this was like the vast majority of the capital that he had available. So huge bet. And for the bet to pay off, he needed to get five to 10,000 people per day gaming in this, in this facility, which is an order of magnitude more than his like best performing club. So it's,

it's literally the most like giga Chad long move that a lot of people are like you were absolutely crazy and his dad is on record saying i don't care if i go bust and he even had contingency plans he's like if i go he's like if i go bankrupt like we'll hit we'll go on the run he's like i'll run it all back wow and so this is this is the environment that Mazda is growing up in as like sub 10 years old, which is just crazy, crazy environment. Um, and so he launches this club first two weeks, absolutely flopped. And he's like, obviously freaking out, but wants to, you know, wants to really make a play. So he's takes like a $350,000 loss in the first month. He's like, I got to turn this around. He decides to, he works with his engineers to make it so that the machines like pay out a lot more and people

start winning a lot and he starts attracting this flood of people. And so the next month he changes it and he makes 350 K of profit. And then he, the next month, you know, they're, they're kind of messing with the machines. He loses it back again.'s no these the sons are just like not afraid to like make it all lose it all make it all lose it all um and uh and eventually he sort of slowly dials it dials it in um and so by the time masa's uh i believe it's 14 the family's now living this sort of dual lifestyle where some they have 20 uh members of the family that are working on the family business, but then there's also this sort of older generation. And so it's this weird dichotomy where the people that are working in the business are like driving like flashy cars and like really like living it up, right? They have all these sort of casinos.

And then the grandparents are like kind of still um around but like very much still living in the past in some way um and so we can get more into masa's you know life and career but i think that's very important context because this sort of gambler's mindset yep and an entrepreneur like high high risk entrepreneurship is 100 in his dna it's all he knows. So when people are like, oh, how could Masa put $10 billion into WeWork and lose it all? It's like, well, he's lost more money than any other man in history,

in the entire history of the world and in our sort of modern economy. He's lost more. He was $70 billion during the dot-com crash. Yeah. The highest number ever lost. Yeah, he was also the richest. So he went from being the richest man in the world

to losing like 90. To personally lost more than anyone in history. Yeah. Crazy. So, absolutely insane. Yeah, and so I think through the gambling business and through kind of their family growing, he meets this guy, Den Fujita,

who is an entrepreneur who brought McDonald's to Japan. And this guy is a psycho. He has this quote. He says, if the Japanese eat McDonald's for a thousand years, they'll grow taller, have white skin and blonde hair. Like, it's like they'll basically turn into Donald Trump. It's so funny that the dude's just like, oh yeah, like this is like Trump code.

This is Trump maxing basically. Trump maxing. And so Fujita acts as like a mentor to Masa and tells him go study English, learn about the computer industry and you got to go to America. And so Masa has this outsider complex. He's like, you know, yes,

I'm like the lowest rung of society. Yeah, my family's from this like destitute, you know, shack and we're still looked down on as kind of like gamblers and like, you know, lower class, but I'm keeping the name. I'm not changing my name and I'm going to prove that I can do it. No. And the cool thing. So Masa's dad had a, you know, I don't know the exact dynamic, but very rough relationship with an abusive father. Who's an alcoholic. Masa's dad is awesome though. And like from a young age is training Masa and saying you are the best. You are number one. Anything you want, you can achieve. Like it's honestly like I went from being I was, you know, listening to this being like very emotional because it was so dark and it flips. And Masa's dad is you can just do things. Yeah. Anything you can see in

your mind, you can achieve. Masa really carries that into the rest of his career. Oh yeah, totally. So he goes to California at age 16, joins this college before entering high school and he advances from sophomore to junior to senior in two weeks by just testing out of everything. And he spent two full years at the college,

then transferred to UC Berkeley to study economics while also taking some computer sciences, some computer science classes. And he has this idea where he needs to generate business ideas. And so he generates one new business idea every single day. He would have loved my first million. Oh yeah. My first million or just like, you know, your, your path of just like building a deck, spinning up an idea, tweeting something out and then being like, actually this is a business. Yeah, that idea sucks. But he was just throwing stuff at the wall. I remember in my first startup, we were destitute

and we had a wall of sticky notes of just all the different business ideas. And some days we would just pull one out and be like, can we build an MVP for this? And eventually it started working, which is great. So he sells a patent for a translation device to Sharp Electronics. So it seems like he does have some sort of technical background, even though later on he becomes deeply vibe-based investor.

He's just not an engineer. And he never identified as an engineer. No. He is the Japanese. In the grand scheme, he's the Japanese Warren Buffett. So you should look at him as like a dealmaker, somebody that's seeing opportunities, putting people, capital, and ideas together

and doing it on a global scale. And so Sharp pays him $1.7 million for that patent, which is crazy considering how young he is at the time. That's a huge windfall back for, this was before 1980. So this is in the 70s. He gets 2 million bucks basically. And so he immediately invested all in importing Space Invaders arcade machines, before 1980, so this is in the 70s. He gets two million bucks basically,

and so he immediately invested all in importing Space Invaders arcade machines, modifying the software and installing them on campus, and that makes him another million dollars, and then he founds a game software company, and he sells that for two million dollars. So he has a bunch of back to back hits really, really quickly.

And so then he returns to Japan in the early 80s and he starts doing extensive research. So he's just kind of monk mode. He's made his money in America enough to buy some time. And he's just spending 18 months doing extensive research. All of his relatives are like, this guy's doing nothing. But secretly, he was building a massive knowledge base. So he developed 20 metrics to measure potential business ideas. And the highest scoring idea for him was personal software distribution. And this goes into his deal-making philosophy. He's not building the software.

That's complicated. That's risky. He's just taking software that already works and just buying it for cheap and selling it to someone else. Yeah, distribution businesses have existed for years. They're great businesses. You usually don't have to take on too much inventory risk. Totally.

You get this sort of toll booth style business. Yeah, not R&D heavy, very bootstrappable. I know some people that own food and Bev distribution companies. Yeah, exactly. That's a nice reference. And they just print money, and they never raised any money. It's like this amazing family-owned business. Yeah. just print money and they don't, and they never raised any money. It's like this amazing family owned business. And so this and this also like just, it's super aligns with his skillset at

this point. Like he understands what good software is obviously, cause he's built a game company, ran the space invaders thing. But if you think about, you know, you get a call from somebody who's like in their early twenties, they've already sold like three companies. Like they're, they have money. They're really like this crazy hustler with this interesting backstory. Like you're going to take that meeting even though they're young. And so that's the ultimate alpha is just, he's like such a hustler has enough of a track record to get the meeting. And then it's like, oh yeah, you're providing me a service. Like I need software to distribute through my stores anyway. So why not? Why don't I go with you? Yeah. And that's why when, you know, because today

the younger generation of entrepreneurs really know Masa for going giga long, turbo long, we work and, and, but that's not the story. So 50 years earlier, he had his 20 metrics to measure business ideas. And so he takes massive risk. And oftentimes there is a sort of vibes based analysis, which, which some of his, his employees talk about. Right. But in general, he's one of the most well-studied, well-researched, most experienced deal makers in history. Totally. Totally. And so there's not like he does get some really lucky breaks, but it's not necessarily luck right because he's taking so many crazy shots on goal yeah and so he chooses the name soft bank think about it as like a bank of software it's not it's not actually a financial institution but it kind of turns into one or a

huge fund it's a certain point and so he he hires two part-time people in a small office he's not afraid to hustle uh and then famously, he would get an Apple box because he's short and stand on them and declare in five years, we'll have $75 million in sales supplying a thousand outlets and we'll be number one in PC software distribution. And the two employees were like, this guy's crazy. We quit immediately. Imagine giving a speech to your only two employees. Imagine we're like, Ben, this is going to be the most profitable podcast. We're going to crush it. We're going to be number one this year. And Ben just is like, I'm good. I'm out. I'm out on that note. I'm out. Yeah. I

mean, if he saw a stand on Apple box, maybe that would be the trigger for that. But hilarious. And so, but he doesn't mind. He, he grinds through it. And so he books a giant booth, the size of Sony Toshiba's at a Tokyo trade show, even though he's like this small company, he's just sending it. It's just like, it's so smart. Cause he's standing out. This is the friend.com, Alvy thing.

Like he's doing the thing where everyone's like, no one would take that amount of risk. And then it gets them a bunch of attention. And so he invited vendors to exhibit for free. And so he's like, I got this booth, I'll let you do it for free. he invited vendors to exhibit for free. And so he's like, I got this booth. I'll let you do it for free. I'm going to advertise you for free. And he's just creating, pulling forward value basically. And so it feels like a flop. There's no immediate leads. But weeks later, Joshin Denki from Osaka contacted him. And originally, this is hilarious

because he's made millions of dollars at this point. And they're like, hey, we're the software firm. Why don't you come and pitch us and we'll see if we can do a deal. And he's like, I don't have enough money to afford a train ticket. Like I literally can't go. Yeah. So we, I would love to know how he's, he probably did the most degenerate thing with all that cash. Probably going long, like Russian corn futures. He just uses it all. Yeah, I mean, he's just investing it. I think he literally spent a ton of it on that trade show booth and just spending it on all these crazy long shots. Oh, yeah.

You can imagine in an era when Tokyo's biggest software trade show to actually get a booth the size of Sony or Toshiba, you might have had to spend a million bucks or something. Yeah, a ton. And so the president, Josh Denke, calls him and says, we need a dedicated software supplier. We want you to be the one to do it. And he says, this is such an interesting time, right? Because this is at a time where companies were R&D product companies, but they wouldn't do their own distribution. So to sell, they would just find a distributor that would be these primary sales channel for them. I mean, this was up until the internet, like all the, not this we're in 1981 or something. And, uh, all the way up to 2000,

you would have to go to, uh, what was it? Comp USA or Best Buy or Fry's Electronics. And I remember buying video games on discs in boxes and you'd take that home. That's why GameStop was a great business. It's because the manufacturers or the game makers didn't actually have direct distribution. And if you just think about it, there's obviously some sort of power law where Best Buy is probably the biggest one, GameStop's big, like you mentioned. But there are hundreds or thousands of individual small software box retail stores. And then there's also thousands of game companies and software companies and somebody making spreadsheets. You forget that before Microsoft condensed all of Office, PowerPoint was its own company. I'm pretty sure Excel was its own company. And they put all these together. Yeah, almost all of that was through M&A. And so at a certain point, if you're the guys who are like,

we're making presentation software, we're going to call it PowerPoint or whatever. You're like, okay, we have some cracked engineers. We built it, it works, it's good. But like now we got to sell it into a thousand different retail stores. Like we have to have a massive Salesforce. It makes perfect sense to disaggregate that

and work with the distributor. And so that's what they do. But Masa gets this phone call and they're saying, we need a dedicated software supplier for Josh and Denki. And Masa says, I have no product, no money and little experience, but I have the greatest desire to succeed. And I will pour my entire spirit into PCs. Let's go. Let's go. Full set. So one thing you'll know about masa that that's important to know is from a very young age he intensely practiced manifestation and visualization so he would have loved la in the 2000s because this guy uh i'm sure he's a big feng shui you know you know

enthusiast as well but he he would in his his view from early in his career, if he could visualize it, he could make it real. And so that has kind of bit him a little bit throughout the last few years, where he'll just announce a huge deal before it's done. And in his mind, he's like, I can see it in my mind. It's done. We're going to do it. And that bit him a little bit with WeWork because I'm sure when he was investing at WeWork at a $44 billion valuation, he just imagined a world where WeWork had office buildings, hotels, apartments, you know, all these, you know, campuses. Right. And so he's like, of course, he was not analyzing it based on their EBITDA at the time. It was entirely based on what he visualized and he believed in adam newman in

many ways adam newman was an amazing ceo from a storytelling narrative brand building standpoint yeah yeah and so in his mind we work was already a hundred billion dollar company so it seemed like he's buying it for cheap right he's like you're the idiots right yeah i mean you can see it in his mind even even we work like it's not like we WeWork, people like to put WeWork in the same bucket as like Theranos, but it's just not true at all. Like this is a benchmark company. Like this was like, they were very much like a serious contender in Silicon Valley,

whereas Theranos is like this weird zombie. The other thing with WeWork is the fundamental consumer product was amazing. Do you remember walking into early WeWorks? I remember being, you know, in college, probably going to my first WeWorks? I remember being in college, probably going to my first WeWork. I'm like, the coffee's free and good. Like, that's amazing.

It was like a fun vibe. Everyone was friendly. Maybe it's not the highest productivity space in the world. It was on a mega trend of like more remote work, more distributed teams, more small teams, more startups. It felt like the pricing, part of why the product was so good is because they were not selling it

for what it costs to deliver. Totally. But it still was like, okay, at 400 bucks a month, it'd be stupid not to get this because I'm getting so much value. Totally, totally. And so Masa, the deal maker comes out in full force with the Joshin Denki deal.

He says, first off, you got to break off other vendor relationships. You got to go exclusive if you want to be with me, but I will work extra hard for you. And then he also really pitches them on this idea that the software is key to them winning in hardware. So the idea is like, if, if Josh and Denke wants to sell a lot of hardware, they're going to need the best software in their stores and people will come for the software and then they'll buy a new computer and that, And that virtuous cycle will drive value for Josh and Denke. He's not wrong there. That is a good pitch. And so he secures the deal. And he also, importantly, gets Josh and Denke to pay for product inventory covering Masa's shortfall. So all of a sudden, he doesn't have a working capital

problem. And so even though he's very small and doesn't have a lot of funding raised or anything. He can basically bootstrap. Joshin deserves, hopefully Masa went back after all this and was just like, here's a boat. Here's a boat or something. Something to pay him back. Cause it seems like Joshin just made it a really risky bet. I mean, look at this growth. He was always a force.

Always a force. And so Masa takes SoftBank from $10,000 in monthly revenue to $2.3 million in monthly revenue. And remember, there's no inventory cost here. So it's all like, effectively, software margins. He doesn't have the working capital or the debt issue that normally comes up. He's printing. And so he IPOs in 1994 at a valuation of $3 billion. And by the early 1990s, SoftBank was Japan's largest distributor of PC software. He manifested this.

And he was extremely wealthy, but far from done. Should we move on to the dot-com boom and bust? This is like, we're just getting started. We're just getting started. Buckle in folks. We got another hour,

at least on this baby. So he moves to Silicon Valley in 1995. He sets up offices in Mountain View, seeing the internet's enormous potential. And 1995, this is still early on the dot-com boom. It's not like he YOLO'd in at the top in 2000. Like he did that also, but he also got in early and that was very, uh, that was very valuable. And so the dot-com boom was in full swing. Companies were rushing to harness the web's explosive growth, but this was still pre-IPO craze, oh, raise, you know, go public with just a slide deck and a domain name. And so in 1996, he meets the Yahoo co-founders, Jerry Yang and David Filo. He offers them $100 million for 30% of Yahoo. That's a big

stake. Typical VC funds at the time rarely wrote $100 million for 30% of Yahoo. That's a big stake. Typical VC funds at the time rarely wrote $100 million checks because that could alone be 40% of the fund. Yeah, this is so normalized now, but the biggest funds back in the day were like the classic solo GP fund today. Yeah, yeah, yeah, exactly, yeah. And it just, yeah,

like you would just raise from VCs for a seed round or a series A, series B, and then IPO. Google IPO very early. Obviously it was a different thing because they were cashflow positive, but even the money losers would IPO and use the capital markets in the public capital markets to raise money. Nowadays when companies IPO, they're oftentimes, it's not even about the money. It's about the liquidity for the shareholders. It's like, well, I raised a couple million dollars just from angels.

Then there were a bunch of huge venture capital firms that could write a $20 million series paycheck, no problem. Then there were growth funds. Then there were crossover hedge funds. Then I started just going to the direct LPs and being like, can I raise a couple billion from a sovereign wealth fund just directly in the private market?

All before IPO. And that's why you see companies IPOing at 60 or a hundred billion. So listen to this quote. Masa gave the, after he made this offer, he gave the Yahoo founders an ultimatum. He says, if you don't take my money, I'll invest in Excite or Lycos, which were competitors, and kill you. He's just making death threats. The funniest thing about this quote is that he didn't even know who their competitors were. In the meeting.

He asks, who do you think your biggest competitors are? And they say, probably Excite or Lycos. He's like, okay, well then if you don't take the money, I'm investing in them. Thanks for doing my DD for me. Ultimate deal maker. Didn't do any research, still wins the deal. And so it's this, it's this interesting because he breaks all the, all the trends of, he's exactly the opposite of Warren Buffett, right? Totally. Right. Warren Buffett is, is heavily researched, sort of slow, uh, moving, builds these positions, you know, over time, make, takes a position, then starts marketing and things like that. He's the opposite where, where he's coming in and he's clearly brilliant and gifted and a brilliant entrepreneur.

But from an investment standpoint, he breaks all the rules, right? He's not even doing market research before offering to do a hundred million dollar investment. Crazy. To not know the competitors. Yeah. And just be like, yeah, like you guys, but also I will kill you if you don't let me invest. And so Yahoo accepts the hundred million million offer. They IPO in 1996, like very shortly after this deal goes through and the share price soared on day one, Masa made $150 million immediately. Which again, at this point in time, doesn't really make a dent in his net worth, right? Like he had taken SoftBank public at a $3 billion market cap. Sure, 150 extra bones is not a bad thing. But I mean, Yahoo continued to rip post IPO all the way. No, no, I know. I know. But I'm

just saying, I think you'll realize through looking at Moss's, the reason that later in his life, he starts to make these crazy size Lord, giga Chad, mega long bets is because he's had instances like later when he invests in alibaba and prints tens and tens of billions of dollars you know that he wished he just massively sized up that position totally totally and so that that leads to sort of 10 you know he understands concentration you know right now there's too many vcs who are like yeah i'm in this amazing company i got a quarter quarter percent, you know? Cause I got, I got diluted down. And I, and I got crammed down on this one round. I couldn't get to the ownership percentage

and the founders only wanted to do a 10% round and I only took 2% of it, all this stuff. But I mean, Yahoo, like the wealth creation at Yahoo was really underrated in the Silicon Valley story. Like when I got to Silicon Valley, you'd meet a lot of Gen Xers and a lot of them were like, oh yeah, I was early at Yahoo. And then you're like, okay, this person's like cool and like kind of smart, but not like generational talent. You go to their house,

like a $40 million mansion in like Menlo Park, because there was just so much wealth created from that IPO and a lot of liquidity. And so it was a huge, huge thing that led to, you know, a rarefied air around anyone associated with Yahoo. And then also just a kind of a Yahoo mafia that spawned and built a bunch of companies. Um, and it was kind of like the Facebook of the day, but it didn't, didn't have staying power whatsoever. And so it eventually unwound. Um, but, uh, so once he gets liquidity from that, he's, he's hooked on deal-making. And so from 1996 to 2000, he invests in 250 internet startups doing a deal a week.

And he emphasizes... And you know there were some weeks where, you know, some weeks he probably did five. Oh, yeah. And other weeks he was back in his Tokyo pad. Yeah, he's making calls. Okay, yeah. What is it?

Oh, who are your competitors? Okay, I'll invest in them. Okay. Yeah, one thing that's interesting, a lot of his career, you know, he had his time in the's interesting, a lot of his career, you know, he, he had his time in the Valley, but a lot of his career, he spent working and living in, in, in Tokyo and just sporadically traveling to places like Davos or California and things like that. And so he's

basically just making a, uh, he's not trying to pick winners. He's just trying to ride the internet wave. So he's not, not doing a lot of due diligence enormous volume and it's a good bet he created the tiger the tiger model was the same thing tiger just timed it terribly but uh tiger tiger actually did way more due diligence because they would hire like bane yeah consulting it was still i mean pretty quick still like they were doing market research totally totally moss is just vibes based ripping checks and i mean we should do the tiger story at some point but they were very active in in asia around this time shortly after this and so made a look made a ton of money there moss is not just scaling the investments he's scaling the firm so he launches basically teams in south korea japan hong kong

australia india mexico brazil argentina so all all you know over the last you know decade you had vcs being like oh yeah we're doing a lot in latin we're doing you know we're doing stuff in india we got bets in pakistan and uh and uh meanwhile moss has already been in all those places just spamming checks it's like you on the, you know, with the lat. On the Dell machine, yeah. You know. Spam them. And so he makes $15 billion in just a few years. And he starts to double down once he meets Jack Ma of Alibaba. And so Alibaba's scaling.

They need additional capital. Goldman Sachs, who was Alibaba's previous investor, was reluctant to keep funding it. Didn't really see the massive potential there. And was looking at more on a spreadsheet basis. And so Masa meets Chinese. First mistake. Yeah. Ignore spreadsheets. For sure. And so he meets Chinese tech entrepreneurs and he sees a glint in Ma's eyes. And one thing about Masa is like you need to understand this run as he is a degenerate gambler because like he grew up in casinos like you know these pachinko clubs and he had so many wins and his dad was fully comfortable with making you know going up

one month and losing it all the next month just keeping it rolling yeah beta beta yeah beta and so uh they shake on a deal 20 million on $100 million valuation, but he winds up getting 34% of Alibaba through as the deal progresses. And so the critical advantage here is that Masa sat on Cisco's board and saw router sales to China exploding, indicating massive internet expansion. So he got some insider intel, understood the market was growing, and went in, found a targeted bet, and ripped a check. What's probably missing from this story is that he probably wrote a hundred other $20 million checks and none of them went anywhere, but it doesn't matter because he got the elephant. He bagged the elephant. And so the dot-com crash was rough for him. Wait, let's talk about the Alibaba

investment real quickly. So he turned $20 million into a $72 billion position, which I think is actually the greatest... I think it might be. The greatest venture bet of all time. Yep. It's hard to measure because I think it sold off a ton during the dot-com crash, but at the current valuation, and I think he still owns a huge chunk of Alibaba, right?

Because that was one of the positions that they were thinking about unwinding to do the Stargate deal. Yeah. So interest rates rise in 2000. The stock market plummets and many dot coms with shaky business models went bankrupt. SoftBank's market cap plunged 99%.

Classic. Yeah, so to be clear, so this was in March of 2000. In January, he was giving a talk in Tokyo at this very kind of cool underground sort of disco type club to all these. And the Japanese culture at the time was very much salary, men, you worked a good income, and it was sort of seen as boring and so masa got all these guys in a room and it's been described as being almost like a get rich quick scheme because

he was pointing around the room being like you're going to be worth 50 million dollars you're going to be worth 100 million dollars you're going to be worth 100 million dollars and just like going around the room like that and it really seemed like that and he genuinely believed it totally and um i'm sure many of those people went on to start, you know, important companies. But he was he I do believe that at every market top. Yep. He still believes. Totally.

Which is why he gets smoked. Right. He's not the guy. He doesn't really. No, I don't know. He never sells early. Yep. Which is which is a drawback for a lot of big investors who have the ability to find

a fantastic entrepreneur, see trends, but then they get pessimistic at the last second and say, Oh, this must be capping out. I've made so much money. I mean, the whole benchmark Uber story, people are like, you know, maybe that was like they were like looking at, well, you know, we're each going to take home a billion dollars individually. Like this is retirement money. Maybe we got to get out, something like that. And there's a little bit of internal devil on the shoulder telling you,

hey, this is too good to pass up. But even though he's losing so much money, he personally lost $70 billion, which is a record at the time. Alibaba is surviving. Did we hit the size? I think we take a moment of silence. A moment of silence, but it's almost like losing $70 billion means you made that much, which should be self. Gone but not forgotten.

Let's hit the size gone just a little bit. Boom. There we go. For the run up and the run down. And so he makes it through with the Alibaba stake and then he starts launching more companies. Yahoo Broadband, Japan Telecom, Vodafone Japan, rebranded as SoftBank Mobile. And this was smart because he found the area of tech that had durable revenues. Totally. People would pay for an internet connection. They would pay for cell service, you know, things like. And disaggregated from the

application layer. Yeah. So it doesn't matter yahoo wins or google wins or excite wins if you own the broadband broadband demand is going to increase today he still has a huge position in t-mobile so there's certain positions that you just merge you know kept kept riding but so the the chaos of the post 2000 um you know the the the actual like dot com crash was because there were all these businesses that masa was hugely invested in that were like dogs.com it'd be like basically a service where you could buy dogs online and it turned out that people were still just buying dogs from like you know regular kennels and stuff like that so you can think about the early 2000s as kind of like back to the basics rebuilding with like the less frothy companies, riding the Alibaba stock, rebuilding a soft bank

backup. And then in 2010, he releases this 30 year vision presentation. And this thing is a bombshell. In June of 2010, it's 133 slide deck with outsized predictions like advancements in artificial intelligence will surpass the human brain. Life expectancy in Japan will reach 200 years and a new era of hyper-connected communication will be upon us. And so his central philosophy is the information revolution will bring happiness for everyone. So I love that even today he does similar things. Well, he's a master marketer and salesman, right? He's a deal maker. So he's got to sell these deals.

And he would even, despite his limited English, he would know the simple words to say, information revolution to bring happiness for everyone, right? It's very sort of simple. It's very intelligible. It's not techie mumbo jumbo about scaling curves and H100s and all this other stuff. Yeah. And even his more recent decks have these very simplified people. They're so simple. People make fun of them. People make fun of them, but then they end

up being correct. Yep, they do. In many ways. And so the Vision Fund is born in 2017. He conceptualizes it in 2016 during a flight to the Middle East. He wrote down a target of $30 billion, then crossed it out in favor of $100 billion. What did he say? He said, life's too short to think small. Yeah. And so just tripling the size of your raise. Just, yeah, let's go big. One of the greatest, probably the greatest fundraiser of all time sam altman's very good at fundraising yet he still has to go to masa to really get the job done yeah elon's great too but it's a very different style of pitch more driven by the science and and the technology so he has this famous uh 45 minute pitch where he he's meeting with the crown prince of Saudi Arabia,

Mohammed bin Salman. In Masa's words, he says, I want to give you a trillion-dollar gift. If you invest $100 billion, I'll turn it into $1 trillion. So MBS ultimately invested $45 billion on the spot, catapulting the Vision Fund into existence. They also add a bunch of other investors, Apple's in the Vision Fund, Qualcomm,

UAE's, Mubadala, you know, sort of sovereign wealth fund. SoftBank puts in $28 billion. The funny thing about this is that he goes to MBS and is like, you got to invest $100 billion and I'll turn it into a trillion. And MBS is like,

that's crazy, but I'm still going to give you $45. It's like a total shoot for my 45 i'm good for my 45 and it's funny because his original target was 30 he sized it up and still exceeded that with the first check yeah and and there's been this uh idea in tech that middle eastern money is is easy to get yeah and that perception is only because there's a lot of capital out there these relationships take a lot of time yeah i know i know a ton of of allocators that have had to go visit the middle east 10 plus times before getting anything done and so it's not like this was easy you know masa made it look easy because he just rolled over there. He was thinking about doing 30, crossed it out. He liked the 100 to a trillion. But this was, again, a very

credible feat by itself. And then SoftBank commits $28 billion of the fund. What Masa has done historically, I don't know exactly, but he also makes venture investments using leverage, which is kind of a tricky situation to get in because one venture investments can go down all the way to zero, right? And yet he still owes that money to someone. And so he would use, he would use, uh, soft banks, you know, actual earnings to, to help, uh, back that up. And then also would lever up other positions. so he would raise debt against you know some of his specific positions and so all throughout this even from you know from the early 1990s to now he's gone almost bust in like smaller ways that don't even

make it into the story because he's just levering up just going long like yeah he doesn't you know i don't think this man has ever been short. You know, he doesn't have any middle of the road outcomes. It's all 10 baggers or zeros, which is great. Well, in the case of Alibaba, you know, a 10,000 bagger. Yeah. So he focuses on AI robotics, IOT and consumer tech. He makes bets on Uber, WeWork, ByteDance, DD, DoorDash, Grab, and then quirkier startups like Zoom for robot pizza and WAG for dog walking. And it's funny because three of those are companies that have been like, oh, this is such a silly idea.

It never was going to work. He was so stupid for doing it. And then Uber, ByteDance, Didi, DoorDash, Grab, all banger companies. Yeah. All very, very good. Yeah, and one thing you'll notice here, he has no issues investing cross-border. So he has big presence in the U S he has a big

presence in China. And so he's one of the few that's actually threading. I mean, of course, some American venture capital firms have, have gone and raised Chinese money or invest in China, but he's getting into the best Chinese companies, you know, ByteDance, Didi, the Uber of China. And so really he's a truly global dealmaker. He can go into any room and make stuff happen. And he's really good at being the capital that bridges these huge deals. Like I'm pretty sure the Uber and DD deal, like having SoftBank bridge those two companies and put money into both in like this interesting way is like what kind of facilitated that.

And that's just like a wild card. It's kind of the same thing that's happening with Stargate where he comes in and all of a sudden, like Larry Ellison is with Sam Altman and it's like more of this like Avenger style team coming together because there's so much capital on the table. Everyone's like like okay yeah I'm I'm marching to the same beat of this drum because he's oftentimes has the conviction to make stuff happen that otherwise would have never been a part of history totally yeah um I I recently invested in a founder uh who will go unnamed that took a couple hundred million dollars from from a vision

fund during 2021 and the pressure that he faced immediate you know immediately after investment to deploy it because in hindsight he's like how did i spend that much money in such a short period of time company ended up uh you know selling it wasn't It wasn't a good outcome. But he just said the relentless pressure of like, we gave you this money. Yeah. 100x your business this year. There's also this interesting dynamic with SoftBank. I've heard from some folks that have had SoftBank on their board that even though Masa might come in, there's this example with WeWork, he gave Adam Neumann, the founder of WeWork, $3 billion on the first meeting and told them to be crazier. So the CEO and Masa, they get together, they have a short meeting and

it's like the deal's done. But then SoftBank actually comes in and puts in like crazy infrastructure in the board. And so if you sit on a board with a SoftBank board member, there will be audit committees, compensation committees, like they run it like a public company basically. And so a lot of investors like hate that because they're like, I'm used to having like a one hour strategy meeting with my boys. And now I'm like doing like paperwork, which I'm not down with. I just want to let the CEO cook and just do it. But it's interesting. And in many ways, early on, I think investors would say this is amazing. Vision Fund, very early, just gave a company that I invested in at an $8 million valuation. They just gave them $100 million. It's a crazy valuation. So the markup's great. Company's super capitalized. But then over time, it ends up being seen as more

of a kiss of death because a lot of these companies, unless you were this generational outcome like Uber or DoorDash or ByteDance. And you're really ready to deploy that money. Yeah, and you're legitimately a phenomenal business that already has some maturity versus something like Zoom, like the robot pizza company where there's tons of competition.

There's clearly no real network effect. It's called a suicide round. I like that terminology. It's great. And there's been a few of them. They lost money on WeWork, obviously, which we've talked about. They lost money on Wirecard, the German payments processor, which collapsed after accounting fraud.

Greensill Capital melted down. And many vision fund startups lacked profitability. What was Greensill? Was it a fund they invested in? Greensill, I don't remember. I thought it was like housing or energy or something like that. And it was a big blow up. And so there's been a lot of those, but he's still in the game. Yeah. And it's tough. Like the reason, you know, the reason why these rounds ended up

being the kiss of death is if you're not profitable and you raise $300 million and you spend it all, then you're sitting on this massive preference stack. Nobody else wants to back the company because it's going to be tough to get any money out unless like everything goes perfectly. And the number is so big, you think like, oh, wow, he gave Adam Neumann $3 billion and like it basically went to zero. Like that's so embarrassing. And then you realize that in 2021, SoftBank posted an annual net profit of $46 billion. It's like enough to cover, what, 10 of those investments, 15 of those investments without blinking. And so you look at the GP commit on the Vision Fund,

they put in, what, 26 or something? $28 billion on 100. Yeah. They could do that all day. They could do that all day. And it's just like, the numbers are so big, you can't analyze them like it's a traditional fund because it's just like, he's just, the numbers are so big. You can't get, you can't analyze them like it's a traditional fund because it's backed by this organization that's just printing cash, printing cash from all of their whole. Yeah. But now I think their,

their actual net income is in the single digit billions. So that was a specific moment in time where they had a lot of things go right, which was just due to COVID and interest rates. And but he doesn't quit. You only need one. You only need a couple wins to make it all back. Yeah, that's basically what he does every single time. And so in 2022, 2023... This history, going back again to his childhood, he was conscious when his dad was willing to bankrupt the family to go when they were already super successful, right? So a lot of people look at this and it's like, Masa, you've made all these amazing investments. You built one of the biggest companies in Japan. You've been the world's richest man. Why are you still taking

on all this risk? And the answer is he's playing the greatest, biggest game that he can play. And so, I mean, you were right, a year later in 2022, they announced a $27 billion loss. Uh, and so they lose money on Klarna. There's a Chinese tech crackdown. And so Alibaba and DD get hammered by PRC regulations. Very hard to make money as a private investor in China, especially if you're a Japanese instead of Chinese. Uh, and then it emerges in 2022 to 2023 that he personally owes softbank five billion dollars due to side deals gone awry he's just ripping he's so he's got his main deals and he's just like he's got his side deals and i'm like oh yeah i'm good for it like all right i'll put this on the ramp card yeah yeah yeah uh yeah just just uh just hit up softbank they'll take care of this

billion dollar loss yeah uh And so Masa says... Yeah, and this is evident of the way that he operates, which is he is such a force that he's traveling all over the world, doing deals, committing to things, announcing things that haven't been actually finalized yet, and then telling his team, all right, just make it happen.

So it's not... It shouldn't be that surprising to anyone that after the chaos of 2020 to 2023 he just happened to like owe his own company five billion dollars because of like some other stuff that probably didn't even make it into the news it's remarkable and so he shook it off he says when it you open an umbrella. And he's repeatedly insisted that he's embarrassed by the fund's performance, but remains an unshakable optimist. And we saw that come back. And we didn't even get to Arm. He turned something like $20 billion into an $80 billion gain.

He's back in the game. And he is making AI deals now, ripping checks into OpenAI, on stargate it's just he's just not going he's so he's he's the deal maker tech mogul that has the he clearly has a very powerful uh very well known in our center you know corner of the internet yep but if you actually go ask people that maybe even are in tech but don't really follow the game closely they won't know him yep and the story is just truly incredible grows up in not just poor but like abject poverty living with pigs yeah like he has he he said he has a consistent nightmare around waking up, you know, having these nightmares

that he's smelling the pig sort of fumes. Right. And he wakes up like, you know, in, in, in shock. Right. And so to go from that abject poverty to the richest man in the world made his, was born in poverty, made his first million in America before he turned 20, runs it down to zero again, runs it back up, back down to 99% down, runs it back up. I mean, this man, he can't be broke. He's like, he should launch a course, you know, like at this point, like if he ever got down bad enough, he just could launch a course.

Yeah, it's so funny, because this week that Hacker News comment went viral. That's like entrepreneurship is like throwing darts at the dartboard at a carnival. If you're middle class, you get one throw. If the rich kids get unlimited throws and the poor kids are working the carnival. And this guy has gotten so many throws and he came from nothing. Just complete invalidation of that. Yeah, he makes his own throws.

Yeah, he just figures it out. And it's funny because that always goes viral and people are like, oh, this is so true. Like this hits so hard. I looked up the dude who posted it. He's never launched a startup. He's never even tried. Yeah.

And so it's clearly just cope, just being like, oh yeah, like I can't do it because like I'm too poor or something. And like clearly there's a million ways to start the flywheel. And then once you're on the flywheel, you get extra shots. Cause it's like, Oh, well like your last company didn't do well, but yeah, you probably learned a lot and you were a CEO. Yeah. He went in that, in that, when he was in that, uh, nightclub talking to all the Japanese

at that time, it was not normal. You know, venture capital didn't really exist in Japan in the same way it does in America. And he gave the stories like in America, an entrepreneur can start four companies. They all fail. And on the fifth one, they become a billionaire. Yep. There's so many stories like that. And so the thing that's the criticisms of the risky governments and the just crazy risk tolerance are real. We're not, you know, this shows obviously for entertainment purposes only. but the thing that I think everybody should take away from Masa is that extreme radical optimism yep and he's really a deal maker yeah it's really just a story of deal-making in my opinion yeah it's it's it's so accurate

and tech mogul without seemingly ever written, writing a line of code. Maybe in the very early days. Yeah, like when he was. But very quickly moved to just, how do I assemble capital effectively? Yeah. How do I put the puzzle pieces together? And this is Sam Altman too.

This is probably why they. A lot of people. Elon hired Gwen Shotwell to run SpaceX. Like it's fantastic. It doesn't take away, it actually adds to their abilities. It's leverage. It's leverage. It's leverage for the team for the mind it's

fantastic well that wraps up our deep dive on masa yoshi sone and soft bank let's go to a promoted post from your buddy steve oh yeah small note yeah uh he's got two daughters i don't know their relationship status but brothers why don't you if you send it, send Masa DM on LinkedIn and say, hello, uh, I, you know, uh, if, if you review my profile and you find me compelling, I would love an, a warm intro to your daughter. There you go. Um, and he's still, he's married to his wife that he actually met at Berkeley. So wow. Masa always gets the last laugh. He does. He does. Uh, let's get a promoted post from Steveve i got a promoted post from my buddy steve uh simone so yesterday you sent me this which was a chinese drone light

show and can you actually play it no you can't no we can't play the video we're working on we're working on tech here at technology brothers we want to get to the point where we can play videos but i will just share so ben uh benjamin craker cracker who knows who knows the cracker says uh how do you defend against this not a rhetorical question and it's a video of these drone crazy drone light shows steve says you don't you have a robot do it for you and so what acs is developing i'd love one of these to put on top of my my g-wagon uh but uh he's developing this sort of autonomous uh counter drone uh gun that can uh i've seen you know a bunch of demos at this point but it can take out people have just described it as acs gun on truck gun on truck yeah yeah they should really buy gun on truck gun on a truck gun on a truck and that's uh but anyways

go check out add steve sim Steve Simone and check out their turret. It's pretty awesome. The bullfrog. Fantastic. We have another promoted post from Ryan McIntosh promoting our show and promoting an ad

in Arena Magazine. This is also a promoted post for Arena Magazine. And there's an ad for Ramp in the ad for Technology Brothers in the ad for Arena Magazine posted by Ryan McIntosh. There's levels to this Ramp in the ad for Technology Brothers in the ad for Arena Magazine posted

by Ryan McIntosh. There's levels to this. Yeah. And so we whipped up an advertisement, full page, full bleed, beautiful, full color, printed in the latest Arena Magazine. And why don't you read it, the whole thing, top to bottom? Sure. So I said, technology needed a podcast.

Last year, we noticed a problem. While we were listening to the all-in podcast on vinyl, technology needed a podcast. Last year, we noticed a problem. While we were listening to the all-in podcast on vinyl, we realized something shocking. It doesn't have ads. No ads for watches, no ads for private jets, not even a plug for ramp. It's hurting the tech community, but the madness ends today. What if instead of focusing on audience size, a podcast focused purely on profit? What if instead of big political questions, it answered the ones entrepreneurs really care about? Should I get a Ferrari F40 or a Lamborghini Miura? Should I upgrade from a Gulfstream to a BBJ? What's the best Amman property?

Technology Brothers is that podcast focused, profitable, capitalist. Visit techbros pod.com today and add it to your, uh, RSS feed. Uh, we, before we dive into the rest of the timeline, we got some questions. Uh, Tyler, Tyler Ron Gioni with a phenomenal suggestion. He says, get a medieval night helmet for steel man arguments. Oh, that's fantastic. So much better. That's so good. So much better than steel hat than steel yeah than just a steel hat i mean we have our tinfoil hat but the this the the medieval knight helmet is a good uh opposite yeah we need both of these clearly and uh anyways uh just j the letter j says which one is the technology brother's favorite yolo seed round or guac a round uh i i think's mango seed round or the suicide a round suicide no it's more of suicide b suicide b is typically what happens 300 million i just die i just spend it all on people i love

i love a yolo i love a yolo i actually like when the first round is just 20 yeah you know kind of like that incorporation round 20 on 100 100. Those ones always work out great. For sure. Next question, Venturi says, I think he's responding to Masa on this deal making. I don't know exactly. He says, sales is massively more valuable than engineering. So Venturi is clearly mogging the engineers listening. So feel free to duke it out in the comments.

But in Masa's case, he probably had way more impact on the world by being a dealmaker. He might have been a top 10% engineer. He's a top 0.01%. Yeah. I think it's clear that he did the 80-20 on the computer science thing, took the classes, got really pilled on technology, and understood how exponential curves will affect the future of the world and humanity. And he can still think like a technologist while not actually needing to lead a tech company, be a CTO, be a programmer, do any of that. But he understands the science at a deep enough level,

which is important. Yeah, he's still not using that knowledge for diligence. Yeah. Purely biased. I mean, we have friends that buy small companies where people say, oh, did you do like technical due diligence? Look at all the code. And they're like, no, I looked at the Stripe account because if the, if the customers are paying, that's evidence that the code is working.

Yeah. Or just use the product yourself. Exactly. Is it a good product? Is it a good product? I know there's, this has been happening with, with really smart sort of AI ML trained engineers that are now building products at the app layer. Yep. AI ML trained engineers that are now building products at the app layer. Oftentimes they know so much about the actual technology, but can't translate it into products that people love.

Oh, cool. Like your training and inference cost is so low. That's amazing. But like, what's your product sense? Like, are people actually using it? Did you get viral market entry? Exactly. Exactly. As R says, should companies stop paying dividends to reinvest their dollars into custom letterheads? Yes, obviously you their dollars into custom letterheads? Yes, obviously. You need to have custom letterheads.

We're working on it here. We're actually behind. We need our own actual stacks of paper, even though we're sort of evolving. I got the business cards. We need TB business cards. I got some personal stationery. I need some TB stationery for sure. Hanil Patel says, I'm planning to do my master's at Berkeley in civil engineering. Should I do it?

Should I do it? He's an international student. He says he wants to go become a founder or break into Founders Fund. Both good options. Hopefully you get both. Well, you definitely have the first option, Founders Fund. It's tough to get in, but

we have at least 50 of this podcast works for founders fund so if you're listening to it that's a good start um i you know if you want to be a founder doing a master's in civil engineering might be awesome or it depends on what you want to build it could be a huge distraction it could be a huge distraction so if you want to you know go into that field and I'm sure there's a bunch of really crazy, cool opportunities to apply like modern technology within that. So I don't think it's a bad idea, especially if you're, you know, not a U.S. citizen and you want to spend time in Berkeley. I mean, so many. Yeah. Berkeley is doing a bunch of cool stuff in AI. I don't know if it's in the timeline, but they I mentioned it earlier. They were able to create a reasoning model for like 450 bucks. So just being in the Bay, I was born in Berkeley. It's a funny, funky town, but you're right there next to San Francisco. Definitely long going to the Bay. I guess the question is, will this pull forward finding a really, really valuable idea? And sometimes being in a master's program can

be a conduit to industry where you might identify a big problem and identify customers. Sometimes you can bootstrap that without that on the resume. You could just go work straight at a company and realize, oh, this stuff's way broken. I got to spin this out and build something. So there's a bunch. Yeah, you're probably going to have to go from your civil engineering master's to work at a company to truly just figure out. So the question is, can you just go straight to the company and get those learnings? But starting companies is not a race. There's just so many examples at this point of people that started their first company at 40

years old and they needed all that experience to get to the point where they had the knowledge. One, two more questions. Do the technology brothers have a favorite pen or pencil brand from azar again i actually don't my handwriting is terrible i'm using it on the show before i'm using a bowery hotel pen here um but also hire a calligrapher yeah for me it's just got to go higher um full time sergio says do you guys publish anything in the paper form or do you have a website yeah we have techbrowspod.com and a new website

that is already in the works with a new domain that we're excited to share soon and then we don't do anything in paper yet we're experimenting

auction these off they're valuable kids books we've talked about stack of random tweets we talked about making a book that's just

why you shouldn't use leverage. And then it's just 100 empty pages. But yeah, we stick with Arena right now. That's our magazine of choice. And Colossus. Colossus, of course. So thank you guys for the questions. Fantastic.

Well, let's go to Grant. He says, substituting three hours of daily scrolling through x.com with three hours of playing factorio while listening to the tech bros pod gets me up to date on the news i care about plus teapot bangers of the day while destroying biter nests and juicing mining productivity i have a similar hack for work instead of writing code for eight hours like a pleb i spend five minutes writing github issues for mentat alt tab and spend 10 minutes expanding my green circuit board, then flip back to GitHub and review PRs. Mentat is the fastest, most capable AI agent for coding

that plugs directly into your GitHub workflow. Log in at mentat.ai for $30 worth of credits free and redirect your engineering talents toward expanding your base. Brilliant, brilliant post. Talks about the show. Yep. Talks about the show. Yep. Talks about the content. Yep.

Puts an ad in it. I love it. I love it. This is peak technology. One post, you're going to have a ton of different content in a post,

including ads. Love to see it. Well, we got another promoted post by none other than Lulu Maservi. We got a fantastic opportunity. She says,

wanted a founding engineer to join Rostra to work with me on a novel tool for founders. Like a Waymo, you will be fully self-driving that like the media's portrayal of a Waymo, you will go through walls. If it's the shortest path, you will choose your own tools for the job, but experience with LLMs with tool use will help greatly, especially if it involves building long chains of structured and unstructured outputs. You are creative and tasteful with a strong sense of style and vibes you have a finely tuned cringe detector and ability to read the room if you believe you are the person to do it apply at

the link below fantastic job opportunity uh if you're looking for your next thing hit up i mean that's actually the craziest opportunity that i've seen in a while. Yep. And truly, if you are Omega Cracked, go apply for that or DM us and we'll let you know if we think you'd be a good fit. Lulu is phenomenal. And the kind of companies you'd be building products for are legitimately some of the best founder mode companies in the entire world. Totally. Crazy opportunity. Well, we got some personnel news. We got some breaking news here. Scott Belsky, after seven years at Adobe post acquisition of Behance, he's shifting gears over the coming months and jumping into the fast evolving world of filmmaking and storytelling, a longtime passion.

He's joining A24, an independent studio he has long admired as a partner, and he will be kicking off a few special projects within. Fortunately, he'll remain in the extended family of Adobe and look forward to working with them as a future tech partner. You want to break it down, Jordan? What else you got for me? This is absolutely massive. If it's non-obvious, he's going to work on AI within A24. He doesn't necessarily say that explicitly, but A24 is one of the greatest film production media companies in the world. Josh Kushner also made a big investment in A24,

Sizelord that we've talked about on the show before. And I'm just so excited to see what comes out of this. This is exactly, you know, for so long it feels, you know, we live here in LA for so long. It feels like Hollywood has just been so behind the times in terms of actually innovating. They obviously there's a, you know, a massive sort of CGI industry here, a cottage industry around LA, but the opportunities of applying generative AI to

filmmaking are so obvious, yet we still need the best talent in the world to actually implement that and make that happen. So Scott going over to A24, I imagine he's going to build out a massive team of really bright technologists that are going to allow generative AI to actually compete with traditional filmmaking, which is going to be incredible for consumers. Yep. An absolute dog. An absolute dog. Belsky on the move. Stay tuned for what's next. And yeah, this is massive.

What a pickup by A24. You love to see it. Well, we have another massive announcement breaking here on the show. You definitely haven't seen the tweet before. 11 Labs has raised $180 million Series C to give every AI agent a voice. Let's go. Oh, with authority. With authority.

You'll love to see it. The past year has been about building the foundations of AI audio. now they are focused on making speech the standard for how we interact with technology yep and so if you haven't checked out 11 labs they have some fantastic abilities to turn text into audio that sounds perfectly human here's a homework assignment download a transcript of this show all three hours put it into 11 labs, generate it in some exotic sounding voices. Give us Irish accents, Irish accents, like majority Scottish, make me sound like Pat old Patty Collison. Yeah. I want to sound like, uh, it's interesting. Some

of these models, Sean Connery voice, please, please. Uh, some of these, uh, you know, voice generation tools have some limitations. They let you do Irish, Scottish voices. They don't always let you do Japanese, which is unfortunate. We were trying to generate some masa. But maybe 11 Labs can do a masa-adjacent voice that we could leverage because we have a lot of uses for that. For sure.

And we would happily pay a few million bucks a year to get access to that voice. Fantastic. But congrats to that voice. Fantastic. But congrats to the Eleven Labs team. Congrats to the Eleven Labs team and everyone involved in that deal. Hopefully there was some cheeky secondary in there. Hopefully. Let's go to an evolving story. This is heartbreaking.

Yesterday, if you were watching the show, we gave Mass, brother of the week, for his amazing work putting up a swing in a public park in san francisco he says they took the swing down apparently fun is illegal that's insane isn't this crazy and the picture is of someone like with the government saying no swings it's like hey we know somebody's actively breaking into your car right now, just like 10 feet over there. This is the priority. This is the priority. How it's just so bad. It's so bad. It is the perfect symbol of everything that is wrong with the government prioritization here. It's really just so ridiculous. Just the mass brought joy to the world, went viral, and it was immediately shut down by the long arm of the law. And so mass, please don't go full Ross Ulbricht mode. Keep things legal. Maybe there's

a way to get through this. We'll get the swing back up. Maybe a petition. Maybe we'll go viral. Maybe we will upend the entire San Francisco government, but we'll do it entirely legally, and we won't need to create a massive illegal drug network to do it. Great. Let's go to Jason Yenowitz. He says, Tether, 13 billion net profits last year. Every financial institution will read this and ask, why don't we have our own stable coin? So I don't know that much about Tether. I know there's been like rumors swirling, but the Tether organization seems to be on a tear. Yeah. So they've had a lot of people that sort of question their business, right? It's sort of this, they control tens of billions of dollars

of assets. I don't know the market cap of USDT, but whatever it is, if they have a hundred billion dollars in market cap, that is a hundred100 billion of basically cash that they have on their balance sheet. And they invest that in treasuries, so they earn some interest rate. They give you some of that, but they keep a little bit. They give you some of it. That's generally the business model. It's a good business model. Whenever you can be in the flow of money, it's easy to just take a little slice.

And so Tether has been also pretty aggressive on the M&A front. If you're doing anything sort of adjacent to them and getting traction, they'll probably make an offer. And they make typically control offers. So they'll say, we want to buy your business, but we want 51%. And so they've been on a tear, very under the radar. They don't have as much hype because nobody's printing, you know, thousand X returns off of tether. They did have some, uh, crisis points where if you remember after, after all the FTX stuff, there was a lot of questions that USCT was trading at less than a dollar, which a lot of people were like, this is free money, but people were worried that not

chaotically it wasn't down to 10 cents. It down at like 99 cents yeah seven cents or something i think it got even more than that at one point but um it's rough but but yeah still they were you know people were but for those people it was free but but the interesting thing is because it's an international organization there's not as much um there's not necessarily you know the fud around tether has been are they just sort of taking a dollar and printing $2 on chain or something like that, but they've weathered every storm to date and we'll see what happens. Um, and, uh, absolutely meteoric numbers. You'd love to see it. Should we hit the size going for that little one, little one, 13 billion in net profits. Pretty good. Uh, let's go to varunam ganesh he says if you're not subscribed to pirate wires already you really should missing out on bangers like these uh from

riley nork he says in the wake of trump's recent executive order renaming the gulf of mexico after its namesake superior northern neighbor i said what i said um google maps announced it would indeed rechristen the ocean basin as the gulf of America. But don't pop bottles on your boat just yet. Right after the change was announced, Google reclassified the U.S. as a sensitive country, a label reserved for nations with strict governments and border disputes like China, Russia, and Iraq. In other words, they're virtue signaling. Google wants to pretend they're here for tech's vibe shift, but behind closed doors, they're rebranding Trump's America as an authoritarian dictatorship. Bold strategy for a company with an ongoing antitrust suit,

a laundry list of foreign governments that want to rob them blind, and a president, their only real hope against those governments, who famously, to borrow your phrase, sensitive to public ridicule. Good luck. And the reason I just like this ad copy is sharing this is because there's some fantastic ad copy, which we love to highlight sponsors. We love to pass through ads when we're reading content. If it has ads, you're going to hear the ads.

Sponsor acknowledgement. It says, Warp fixes this. If I run the FBI, we're shutting down headquarters on day one and reopening it as the museum of the deep state. Cash Patel's words,

not ours. Here's the thing. Your business has a deep state too. It's called the HR department. You know, the bureaucratic overlords who bury you in paperwork and pointless meetings. Warp shuts them down cash style by automating payroll, navigating state tax compliance, and making international contractor payments effortless so founders can focus on building, not busy work. One of my portfolio companies got kicked off of rippling with zero notice

and Warp got them live again the same day. It's great. I'd love to see it. Love to see it. Let's go to Bology. He's talking about our favorite topic, technology brotherhood. He says, we need the tech bro. Massachusetts equals tech.

Virginia equals bro. The synthesis is greatness. Lift weights, yes, but train weights too. And I agree. He goes on to expand on the analogy, doing a little bit of Coogan's Law here. Coinages, bolligy. Fantastic coiner. He loves coining things.

Big coiner. Big coiner. Bit coiner. And worked at Coinbase. Coinbase. Coinages. All of the above. To explain, Virginia's time ended long ago, and today it's just another U.S. state, but

it was originally the home of the masculine, noble, and brave cavaliers. Massachusetts' time is now ending too, but for almost 400 years, it was a center of higher learning. That's why Harvard and MIT are there. These were two of the most important American colonies. In a sense, they compete and cooperate every hundred years. In the mid-1600s, they fought back in England itself. itself first the cavaliers the moose for virginia then the roundheads migrated to massachusetts where does he learn this stuff i've never heard those terms before uh roughly speaking the cavaliers were right and the roundheads were left and the english civil war was over the same themes of hierarchy versus equality in the 1700s they worked together to beat britain and found the united states In the 1800s, their descendants fought in the Civil American War. And in the 1900s, they worked again together to win

World War II and found the American Empire. Now in the 2000s, they are fighting once again over the same themes of hierarchy, red versus equality, blue. Even if the geographic center of red has been displaced somewhat to the south and the geographic center of the blue has been dispersed across the northeast. Others have written about this as well. See The Fourth Turning, Turchin, Yarvin, Walter Russell Mead, Albion Seed, and Scott Alexander's review. My view is that red and blue will unfortunately smash each other to pieces over the next several years, and then the gray will have to pick up the pieces. Gray is tech in its modern incarnation of the tech bro. Tech is the intellectual successor to Massachusetts as the migration to the internet of Harvard's best

proves. Bill Gates, Zuck, and Paul Graham all moved away from Harvard in the literal sense. And of course, Larry Page, Steve Jobs, Vitalik, and many other college dropouts moved away from Harvard in the metanomic sense. But in its more recent tech bro incarnation, tech is also becoming the virile successor to Virginia. Elon, Zuck, and Palmer Luckey are examples of tech bros respectively becoming leaders of men, packing on muscle and getting serious about the military. Amazingly, this synthesis, the tech bro did not exist years ago, but it was memed into reality by blue attacks and has the balance that blue and red both lack. Train weights, yes, but also lift weights.

I think this synthesis, the Tech Bro, will be crucial to what follows. After all, the Tech Bro founded America. Let's see what they found next. What do you think? I don't want to say anything at all because I feel like it would take away from that dramatic, powerful prose.

I think he nailed it. Yeah. If I had to fight in the culture war, I would want Balaji to be my Napoleon. Yeah, my Napoleon. I love it. It's great. I think, I mean, we literally worked out today. It was leg day.

I'm super sore. And then we dove into semi-analysis and talked about training weights and AI semiconductors for an hour. And that is the fantastic dichotomy of the tech bro that is so enjoyable and wonderful. And I think value creative as well. And it's interesting because during the mid-2010s when Tech Bro was a slur, we could have just said, no, it's actually fine. It's fine to be a brogrammer and lift weights and program. But there just wasn't enough like mimetic or social power to actually have a post like this that everyone says, yes, I'm on board. It took Coinbase. It took. And now you have people like WordGrammar who did one of the best analyses of the DeepSeek research paper. He's an amateur powerlifter. Oh, he's jacked. And I would actually push him to say,

you know, keep being a professional programmer, but become also become a professional powerlifter. Don't limit yourself. He's diced. He's he's diced. He's diced. Jack. That's fantastic. We'd love to see it. Uh, follow word grammar. Well, let's go to this post by Andre. Uh, funny meme. I like this meme format. It's Sydney Sweeney and Ana de Armas recently had a spontaneous exchange about which characteristics would constitute a sixth generation fighter aircraft. Sydney, no standing vertical tail surfaces, the ability to operate AI-enabled drone wingmen. Anna, but those will be first fielded in F-35s, so drones alone would not constitute

a sixth generation fighter. Sydney, well, fifth gen, sixth gen, these are marketing terms, not technical classification. Northrop Grumman is calling the B-21 Raider. Anna says, a heavy payload strategic bomber. Sydney says, a heavy strategic payload bomber, the world's first six-gen aircraft. Also, China doesn't recognize the same designations at all.

They recognize the J-20 and the F-35 only as fourth-generation fighters. Ana says, okay, so what else? Adaptive cycle turbofan engines, open system software architectures, suborbital flight. Sydney says, LMAO. I wouldn't hold my breath on that one.

And just a great way to deliver a little bit of information. I definitely learned a little bit about fourth, fifth, and sixth generation fighters from this. Hadn't heard the terms vertical tail surfaces before, adaptive cycle turbofan engines,

but now I know to go look them up. So thank you for turning into a meme. Ready for the next one. Yeah, I love when any new video just like emerges on the internet of a potential six gen fighter and then everybody just like breaks it down in a million different ways and you just get one thread and learn a lot about fighter jets and about 20 comments.

And it's just, it's so fun. It's so fun to learn about this stuff. So we got a post from Kyle Somani, a huge investor in Solana. He says, Polymarket parlays, it's so fun. It's so fun to learn about this stuff. So we got a post from Kyle Samani, a huge investor in Solana. He says, Polymarket parlays, who's building this? And the reason I wanted to highlight this was because six months ago in October 23rd, I said, does Polymarket allow parlays? Because I want Lando Norris in Mexico City,

four rate cuts, Trump Rogan over two hours, and US confirming aliens exist. I would have lost this parlay, but I could have made a lot of money if they had that product. I did talk to the polymarket team about parlays. It's very hard from like the crypto perspective on the backend to actually implement it. Um,

but it would be very fun. Uh, and I, and I, on the, on the pirate wires podcast, I talked about the, the Coogan parlay,

which was, uh, you bet whoever does whatever, whatever presidential candidate does the longer episode of Joe Rogan will win the, uh, the popular vote and the, and the national election.

Because basically I was saying that like, I was predicting that this would be the podcast election and that, and that now that three hour podcasts were on the table, it was really going to be a battle for attention. And whoever could be in the ears of more americans for longer would draw more attention and authenticity totally which you cannot hide your true self beliefs etc across three hours an hour totally you can hit talking points exactly have your notes but by our we know this by hour three you're just that's when the real geordie

comes out yeah that's when the real real dog comes dog the real dog yeah yeah you gotta put the by hour three your dogs are yeah your dogs are barking yeah and uh and so yeah there was a lot of debate over oh Trump's too old he couldn't last two hours three hours on Rogan there was a whole polymarket for it and then obviously Biden people were saying he couldn't do Rogan but then Kamala they were like well she's gonna go on call her daddy that'll be really big but her caller daddy experience was, I think like, uh, her appearance was like maybe an hour, maybe slightly over. And it just wasn't as viral because it wasn't as long. It wasn't as deep. And I think that really hurt. I really, I do think that really changed the election. Um, let's, uh, skip this next one. Uh, did you see this, uh, the Derek, uh, die workwear the watch guy getting in a fight this is yeah i think i

share this uh alex alex broke it down so basically uh derrick uh derrick guy at die workwear posted a handful of watches that he was uh thought were for were interesting as as well as a belt um somebody quote quote uh quote tweeted it and said, if you ever need a reminder that this society deserves to be destroyed, it should come from the fact that absolutely anyone cares in the slightest about any of this.

Nonetheless, are willing to pay many years salaries for something you can get for five bucks. So this guy's basically saying, society should be destroyed because people like nice things. They go back and forth a little bit. Derek immediately responds with a bunch of evidence that the guy that quote posted him, who actually has a lot of followers, is a pedophile, convicted sex offender.

And this guy is astronomical. And so this guy, 5,000 likes to just zero likes. Yeah. So this guy actually has the audacity to respond and tell Derek, one of us will be remembered for trying to save millions of innocent lives. This is the same guy who said society should be destroyed. And the other will be remembered for embodying why this society doesn't. Well, your only God is luxury. You're emblematic of absolutely everything wrong with the society. And Derek says, I i like watches you're on

a watch list it's such a good ratio again you know we talked anyways i think this guy he's a fantastic poster he's a fantastic poster anyways the guy he's going back and forth with uh should delete his account generated generated a bunch of regretted user seconds oh totally but at least die work where managed to turn it into entertainment for the rest of us but it's fantastic um anyways absolutely brutal brutal highlighting that alex cohen we appreciate you let's go to andrew mccallop friend of the show he's been on before he says how embarrassing must it have been for all these car manufacturer dynasties such as ferrari to get completely dominated by a caffeine wrapper company in f1

talking about caffeine. Red Bull is just a wrapper on caffeine. And they've crushed it. Max Verstappen's an absolute beast. The car is beautifully designed. Obviously, it's not Red Bull manufacturing the engine, but it's a lot of the energy and the brand of Red Bull definitely comes through with the team. And the budget and stuff.

Yeah. Did you see Lewiswis hamilton just crashed in the ferrari for the first time i mean he's fine yeah but it was a practice run okay crashed we might have to start we might have to start live streaming yeah f1 i think that's the real failure mode for this show you know people always criticize oh all in they used to talk about tech now they just talk about politics i think if this show goes down the drain it's going to be because it just becomes an f1 show and we're just talking about sports all the time yeah just sports all the time and everyone's like didn't they used to be talking about tech like why are they covering the

nba finals somebody somebody a uh a friend's company told us uh very recently that they were uh it'll all go on a name but they're like we're working on this you know partnership with this athlete turns out to be a very famous athlete neither of us knew who they were you're like i was like is that like an actor or something this is part of golden retriever maxing i'm getting into sports because the politics is too much it just rots my brain and i want to be able to just chill out and just rip about the game. Yeah. Rip parlays. Exactly. I've never, I've never ripped a parlay. The first one's going to be electric when it hits. What if you hit it on the first try? You're going to be addicted for life. Awesome. Yeah.

Let's do it. Uh, speaking of ripping parlays, we got Ross Ulbricht update from our deep dive. Most recently, uh, he lost $12 million on pumped out fun. Hilarious that he has $12 million to lose. No, no, no. Further evidence that he has 12 million dollars to lose no no no further evidence that he has hundreds you can read the post i don't know the full story but that's never stopped us from talking about anything on the show uh so it says ross olbrich or someone with access to his wallets just accidentally nuked the price of a pump fund coin sent to him while trying to provide liquidity to said coin and so what happened is somebody basically made a coin like a free ross coin or some type of coin that that ripped because

obviously the vast majority of ross's biggest fans are heavily involved in crypto and uh so anyways they then sent him a lot of the token because they're like hey this is your token uh and then he made a sort of technical error in terms of trying to actually he was from what i can tell he was trying to um support the the project basically by adding liquidity to it which presumably would have made the the price go up but uh he ended up blowing um more yeah, $12 million, which was effectively, I don't know exactly how it works, but some type of bot that's entire job is to find- Road coins?

No, like- Just find big traders, like whales. Yeah, if you're a super talented software engineer, you can write code that just sort of, it's the same, similar to high-frequency trading, right? And so anyways, basically wealth transfer of $12 million from, um, from him or the wallet that, that he controls to, um, uh, this bot. So the bot did well, everybody else got smoked. Good day to be a bot. Good day to be a bot. Yeah, maybe consider going digital. But somebody, I saw people in the comments saying he's a little rusty. He's a little rusty.

But even in the era, even when he, even during the height of the Silk Road, there were coins flying around everywhere. This is part of the course. He would lose 10,000 Bitcoin in a day and just shrug it off. He'd be annoyed. He'd be like, I'm going to fix the code. I'm going to kill the person that did it. Just a little bit. Or try. You're going to order the hit you're not necessarily gonna try to try to do

it uh let's go to luke metro he says aws gov cloud is safe y'all because breaking news in a major reversal amazon is ramping up ad spending on x after pulling much of its advertising spend more than a year ago and this is i want to cover this because this is a major narrative violation like after elon bought x all the all the advertisers pulled out, he told them to F off. It was very dramatic. And there was this question about like, okay, are they going to be able to get the premium side running? Like, is the business model going to change completely? But advertisers are coming back. We're set. I'm setting up a, uh, an advertising program with them, uh, for Lucy and they've, and they brought a ton more advertisers on. They just did a deal with Visa

and just in general, the vibe shift shift is is perfectly timed because people realize that yeah there's a lot of valuable people on X and it's worth reaching. Do you remember after the whole ad crisis all the advertisers were things like buy these Donald Trump socks you know it became like the bottom of the bottom of just like people that will like pay for any type of, but then quickly I upgraded to X premium plus pro plan or whatever. And I never see any ads now. And it's actually a lovely, it's,

it's totally worth it. I don't know. I kind of miss, kind of miss the ads in an ad here. Well, if you see a good ad screenshot, it's send it to us and we'll read it on the show. Yeah.

As long as it's in a five-star review. Yeah. So, uh, peak journalism. Uh, this is, this is just hilarious. Uh, everything I say leaks Zuckerberg says in leaked meeting audio. You love to see it. Absolutely. Get it on the archive. Uh, very funny. It is, it is interesting. Like there's, there's some sort of shift that happens as companies scale and get more media attention. And it obviously matters a lot more if it's consumer and it gets clicks. I'm sure this is the case that all the

Elon companies, no matter how small they are, a reporter will be dedicated to like the Elon Musk beat or the SpaceX beat or, you know, the meta beat. And they'll be like, yeah, I was at the Wall Street Journal for four years. I was covering meta. And so I cover their earnings, but I also cover any press release. And I'd be talking to employees all the time, just trying to get information on what's going on. Oh, it's telling a story, trying to get clicks. Also just doing good journalism. Sometimes occasionally it happens.

And and it's funny that this leaked. It's almost like he baited this out. Yeah. Like he knew it was going to leak. And so he just said that because it's hilarious. But it really it really must be so weird to be Zuck. And that's something I'd never seen an interview hit on is like, what is your life like? You're this, like they made a movie about you when you were in your twenties. Very

few people get a biopic by Andrew Ross, Andrew Sorkin. Yeah. No, Andrew Ross Sorkin is the Aaron Sorkin. Aaron Ross Sorkin is the, is the journalist. Um, and so just, just a bizarre life, uh, but he's living it to the max, getting jacked, wakeboarding, and doing UFC. You'll love to see it. He's grinding on. And to end on some really positive, awesome news, we got a post from Adam Singer. I thought this was amazing. When people talk about the Roaring Twenties, I feel like it's so often about the markets and just looking at charts.

But this is the roaring twenties folks. We can go on a carnival. That's a mini Disneyland on the water. We got a sphere. We got planes going Mach one. We're catching rockets with chopsticks and it's all out there in the open. We should celebrate it. Uh, we're also working on some very cool stuff with Adam, which we'll announce soon in the ad quick team. But look, we have seven minutes until the market's closed. Okay, let's do it. Just got the notification from public. So cool. That's our time to call it for today. And I cannot wait for Monday. Yeah. So thanks for watching. Go leave us a five star review on Apple podcast and Spotify. Please do

both. Leave an ad in your review. Leave us five stars. We have two ads that way. And stay tuned for the next one. We'll see you Monday. Thank you, folks. Bye.