This Week in Startups - What will be OpenAI’s IPO price? Place ya bets! | E2202

Episode Date: November 1, 2025

What will be OpenAI’s IPO price? Place ya bets! | EXXX*Polymarket’s sharps are already placing bets on when ChatGPT maker OpenAI will go public… and Jason and Alex want in on this action. They��re placing side wagers on how the House that Sam Altman built will be valued once it finally goes public.PLUS find out why Amazon is Jason’s stock pick for the next five years… Why startups are starting to release their own foundation models, and what this means for major players like OpenAI and Anthropic… AND Producer Oliver stops by with an in-depth look at the music-generating Suno app.Friday shows don’t get any bigger than this.Register for Founder University's Japan Kickoff!https://luma.com/cm0x90mkToday’s show:*Timestamps:(0:00): Jason doesn't trust Sam Altman... Stay tuned!(2:24) Jason’s been using Tesla FSD’s “Mad Max” Mode and it’s a SPICY ride!(8:54) Jason and Alex checked out Higgsfield’s new Face Swap tool… Jason thinks this will revolutionize the movie casting process(10:39) LinkedIn Ads: Start converting your B2B audience into high quality leads today. Launch your first campaign and get $250 FREE when you spend at least $250. Go to http://linkedin.com/thisweekinstartups to claim your credit.(14:03) Here’s what Jason thinks: there are acceptable uses of Higgsfield’s tech but some are OVER THE LINE(17:23) Here’s why Jason says the music industry deals with startups like the Mafia(19:59) Northwest Registered Agent - Form your entire business identity in just 10 clicks and 10 minutes. Get more privacy, more options, and more done—visit https://www.northwestregisteredagent.com/twist today!(21:02) Producer Oliver has been playing around with AI music app Suno… here’s a live demo(22:47) We’re using Suno to make a Dire Straits song about JCal!(30:11) PaperOS - Whether you’re raising a round, launching a fund, or managing a venture portfolio, PaperOS can unlock simplicity and scale across your empire of capital, contracts, and companies. Claim your $10,000 credit at https://paperos.com/twist(39:42) IT’S EARNINGS WEEK! What Jason and Alex make of the fresh numbers from Amazon, MSFT, and Alphabet(40:58) Jason’s checking out his Grokipedia page… could this be the internet’s new go-to resource?(45:15) Why Amazon is Jason’s pick of THE stock for the next 5 years(46:42) Three startups made their own AI models! Is the era of paying big co’s for their LLMs coming to an end?(50:09) Jason’s warning for anyone “dumb enough” to use OpenAI’s API… they are studying you!(57:16) It’s a new Jason-Alex bet… What will OpenAI’s IPO price be?(59:35) What happened with Navan’s IPO? Jason and Alex dig into the numbers.(1:07:37) Why Jason compares some later investors to Wormtongue from “Lord of the Rings”

If I was a developer of any kind, I would never work with Sam Altman and Open AI. Let me say that. This is a warning. People can clip this. This is a warning for anybody dumb enough to use Sam Altman's Open AI API. They are studying you. Remember, Sam Altman's been around the block. I've known him since Loop.

He's an incredibly savvy person. And he wants every bit of revenue from the ecosystem. He isn't taking no prisoners. He's going to study how you're using the API, which he has the right to do. Sam Altman comes from the Zuckerberg School of Business, which is give dumb people, access to your tools, study them, and like the Borg, steal every innovation they have. And Zuckerberg got it from Bill Gates and Microsoft. Microsoft had a platform and operating system. They let people build Lotus 1, 2, 3.

Then they did Microsoft Excel. They let people build a product called Word Perfect and WordStar, and then they built Microsoft Word. This Week in Startups is brought to you by PaperOS. Building an Empire, PaperOS offers the largest library of AI-driven workflows for both funders and fund managers. Whether you're raising capital, launching a fund, or wading through diligence, PaperOS unlock simplicity and scale for your ever-growing empire. Claim your $10,000 credit at PaperOS.com. Twist. Northwest Registered Agent, starting your business should be simple. With Northwest Registered Agent, you can form your entire business identity in just 10 clicks and 10 minutes. From

LLCs to trademarks, domains to custom websites, they've got you covered. Get more privacy, more options, and more done. Visit Northwest Registeredagent.com slash twist today. And LinkedIn ads. Start converting your B2B audience into high-quality leads today. Launch your first campaign and get 200 $150 free when you spend at least 250. Go to LinkedIn.com slash this week in startups to claim your credit. All right, everybody, welcome back to this week in startups. It's Halloween 2025. It's October 31st, 2025.

Here we are. A month away from my birthday, I'll be the double nickel coming up, 55. I can't try 55. I have been using, I have been using, FSD 14.1 in Mad Max mode. And we had Elon on All In. That'll be out later today. That's a little surprise for folks.

Elon episode dropping and I was talking to him about it. It's nuts. Mad Max mode is when it is a bit more aggressive, let's say. They already have hurry mode. They have hurry mode. Now they have Mad Max. The Mad Max mode is like you gave, when I used to be in a rush in New York, taxis were cheap, you know?

I'd be like, hey, I'm late to the garden. Get me there by this time. And I'll double the meter as your tip. Yeah. You know, so it's $12.00. I'd give you $25. And they'd be like, absolutely. I'm looking for an excuse to drive that way.

I should be driving up on the sidewalk if you do that. Mad Max doesn't drive on the sidewalk. Okay. But it will go around people and it will make lane changes, you know, to a minute to just get around people, it's awesome. And it does go above the speed limit. So it does lead you to believe, well, it is supervised FSD. So therefore, I should be able to do what I want. And if I want to go a couple of miles over the speed limit and push the envelope, like 95% of people on the highway are doing anyway, I guess I can get a ticket, not the FSD, but it is a spicy, spicy, spicy ride.

And it does feel like FSD is, the 14.1 is, I wouldn't say like a step function higher, but it's noticeably better than 13. So the journey continues. And he said on All In, and I think on the earnings call, he was preparing to take safety drivers out by the end of the year or, you know, pending regulators. So I interpret that as second quarter of next year, the regulators will let him, but maybe in a small section of Austin. but let it happen, but I'm feeling pretty good about self-driving coming on in 2026. It's going to be pretty amazing. I'm so glad that I bought my last car that I'll ever buy that won't drive for me. And I just do not want to buy another one until it's time. So just tell Elon to go a little faster if you don't mind.

No, no. Take your time is what I literally said to him. No shame in the safety driver game because I do think if we have another cruise or Uber from Arizona back in the these tragedies, it's going to make regulators take a whole different approach. So we got a lot to run through today. I thought we should start with Brian Armstrong, punking the prediction markets. Oh, my God. Okay, so I have to explain a little bit to everybody.

So prediction markets, we talk about them to show a lot, Jason. We often use them to take a look at who's going to have the best AI model by December 31st, or who's going to win an election, or who's going to win a sports game? You can also get super granular on call sheet and polymarket and literally wager on what someone will say on an earnings call or even how many tweets Elon will have in a day. There's really a lot of stuff out there to bet on. So at the end of the Coinbase earnings call, literally at the tail end, something absolutely incredible happened.

And I think it's best to just play it and let people know for themselves. So you're going to hear the end of the earnings call with the CFO. And then there's going to be a pause. and then Brian's going to jump. Yeah, I hope we answer your question on that. I was a little distracted when I was tracking the prediction market about what Coinbase will say on their next earnings call. And I just want to, you know, add here the words Bitcoin, Ethereum,

blockchain staking, and Web3 to make sure we get those in before the end of the call. Okay. So what he did was he punked, is so punk rock, Brian Armstrong. is the greatest. He literally, here's the polymarket. What will Coinbase say during their earnings call and margin? Crypto Winter, he didn't say. Prediction market, SEC, custodial, custody, regulator, institution, staking, stable coin shut down. He didn't say. Subscription, he did say. Bitcoin, he did say Web 3. He did say trading volume, Ethereum, blockchain, etc. And yeah, That's how it goes, folks.

Well, Jason, pull up the Kalshi link there. There's a chart that shows just how much this changed the betting at the very last moment. I think you'll like that. Oh, my goodness. Look at that. Yeah. So here's your Kalshi chart. And I guess I could put it at one day.

And, yeah, as you can see, terms that you wouldn't expect to go way up. went way up during the call. So just understand how prediction markets work. You are in a grand casino where lots of different things can happen. And one of the things that can happen is the CEO can be aware of them and say things or the person writing the script could say things. And it's all fair game in the world of predictions markets. Yeah. Brian Armstrong did say it sweet that quote, lull, this was fun, happened spontaneously when someone on our team dropped a link in the chat. Here's a question, though, and I don't want to go too long on this, but like, is it insider

trading if you read the predictions for your own earnings call and do them and bet that you will say them? Well, see, I think that would be against the terms of service of these, that if you are the person who reconciles the trade, yes, I don't think you're supposed to participate. So somebody should, we should ask producer Claude to look that up and we'll, we'll get that answer. Producer Marcus, go find that, or producer Alon, go find that on the terms of service. But can people who are involved in resolving the bet, alter the bet, and then what happens in that circumstances? My understanding, which is based on my memory, is that you're not supposed to do that.

Now, I don't know what they do in that instance, but hey, brave new world. And here we are, folks. Very punk rock. And, yeah, we'll keep watching it. It was the funniest thing that happened in the whole earnings season. So shout out to Brian. All right. Next up on the docket, Jason, Higgsfield face swaps.

Now, Higgsfield, it's a company we haven't talked about on the show to my knowledge. Just to give people a little bit of context here. It is a San Francisco-based startup with roots in Kazakhstan. And they offer two consumers, essentially a lot of AI models you can play with. Some are open source, some are not. And you pay for a monthly and the number. of credits essentially. So we've seen things like this before. The reason why this is blowing up, and I wanted to show it to everybody, is that the face swap is bonkers good. And Marcus pulls

some clips for us. So can I entertain you for a minute? Yeah, please. All right. Go right to it. Here it is. Take a listen or take a watch, everybody, and I'll narrate just a little bit for folks on audio. Here's De Niro turned into Trump, shooting someone in the face. Here is Joker turned into Joe Biden pretty affectionately. Oh my God. And they swapped in Robert Patterson, Patterson Jr. Instead of the Joaquin Phoenix guy. Yep, Joaquin Phoenix, who played the Joker.

And this is all the Joker trailer. And then you get to see what Batman would look like as the Joker. You know, it's pretty effective. Keep going. Here's another one, Jason. This is the character from Squid Game. Okay. And they're going to drop in Samuel L. Jackson.

very effectively, and then they're going to drop in other dude. Okay. And one more really short one. Chris Hemsworth, I think also known as Thor, the mighty Thor. And here is Leonardo as, all right, all right, all right. McConaughey, Interstellar. Incredible. Founders, listen, if you've got a B2B company and you're marketing your product

or service to other businesses, you need to use LinkedIn ads. Otherwise, you're wasting your time and you're wasting your money. LinkedIn knows who the big players are and who the key decision makers are in your vertical. Based on the data they're already collecting. The platform has more than a billion members, and that includes those 10 million C-suite executives. B2B marketers report 2.5 times higher returns on ad spend on LinkedIn compared to other social platforms. That's why my firm launched advertising.

is our own projects on LinkedIn. We want to reach people who are actually doing exciting work at real companies. So here's your call to action. We've got a special offer just for Twist listeners. Launch your first campaign and get $250 free when you spend at least $250. That's one for one. Go to LinkedIn.com slash this week in startups to claim your credit. LinkedIn.com slash this week in startups.

I have to say, hey, Air Director, Juan, you can jump in. on this, since this is your wheelhouse. This is actually putting all jokes aside. What a great way for a casting director to see what a person might be like. So in the Ashoka series, one of the former Jedi who turns to like, not the dark side, but kind of like in the middle side,

or maybe the dark side, he died after the taping of season one. I forgot the names. It's Balin. I think is his character's name in Ashoka. Ballon, the real actor tragically, died. And they replaced him. Ray Stevenson.

Ray Stevenson is the actor you're thinking. Okay. So he passed away, correct? Am I remembering this correctly? And then somebody's replacing him. Yeah, he played Baylon Skull in the first season of Asoka. And then he died in May 2023. So they are discussing, they want to keep this character,

Baylon Skull, in the Star Wars universe. So they're probably going to be cast him. I think they have, if you look at producer Claude. So interestingly, you could as the Star Wars. In fact, it's Rory McCann who played the Hound on Game of Thrones. He's going to take over as the new Baylands. He was great in Game of Thrones. Yeah, it was great in Game of Thrones.

So now imagine you're the casting director, Lon, how amazing would this be? Or if you're doing Indiana Jones, like they're going to reboot the Indiana Jones franchise. You and I have talked about on the group chat recasting Star Wars for a new generation, like the original trilogy, you know, episodes four, five, six, you could just replace folks. This is great. What an amazing way to do casting. Or if you're just a fan of, you know, a certain series to put yourself in it. Like, incredible to see myself as, I mean, you don't have to really stretch the imagination here, Alex, given my weight loss and my increased muscle mass and the haircut that I just got, tight is right.

you could put me in as Daniel Craig you don't need to squint it's just it is what it is folks he's my doppelganger absolutely uh similarly good poker results similarly chiseled jawline similar British accent similar English roots you know absolutely 100% slam dunk uh seriously though Jason how does this work with IP because we've talked a lot about IP and we'll talk about okay next story but yeah what are the rules here so the rules are if you're using this as a tool on your desktop you bought the tool and you put images in it, that's on you. You could do whatever you want as an individual with media you've purchased. So if I bought a magazine and I cut it up and I made, what do you call those? Like a collage. A collage or a picture box. They make those framed boxes. I don't know what

that's called. Diarramma? Sure. If I go do that, I bought the media. I made some new product. Great. I'm not selling it to the public. Now, if I start making art and I use this IP, then you could have a lawsuit because you're infringing on somebody's unique IP. So if I made a thousand posters of, you know, the Star Wars IP because I thought it was fun and I sold them on Etsy, Etsy gets takedown notices. Now, if you do it bespoke, it's one piece of art. It's a little harder for them to come after you. So if you were to release this and you said, put Leonardo DiCaprio as every Star Wars character and make me a montage, that's very illegal. Because this company would be then using this protected IP.

And that would be their opportunity in the world to pursue that revenue stream. So the way IP works and fair use works, if you're doing something for educational purposes, you're doing it on your own computer and it's non-commercial and you're not like making a profit off of it, that would infringe the original party's ability to exploit their IP, no harm, no foul. So me making the collage and giving it to you for your birthday, it's not really like going to affect the Star Wars franchise. However, if the Star Wars franchise makes a Jedi me, as I talked about for the past year here on the program, that's their opportunity.

You can't take that from them. And especially if you're making money. So if this platform lets you upload videos and you happen to have stolen the video, okay, that's on you. And individuals would then be responsible for their use of this tool, just like the scissors and the glue making the montage are not responsible when you make your Star Wars montage. Next up on the docket,

we're going to do a demo of a cool product from Suno in a second, though, Jason. I just want to throw in one little news hook first before we bring Oliver on to show us that, which is that Udio has done what you said they were going to do, which is they have signed a deal with a major, record label, UMG in this case. Terms were announced, I think, was later on Wednesday. The thing to know here is that it appears to be a little bit restrictive. So I would just say that we are now in the lawsuit settlement era of AI music. So just so I understand, Suno did reach $150 million in ARR. That's

incredible. And UMG, the Universal Music Group, and UDio, a different service, settled their suit and formed a partnership. This is critical. Now, the partnership means they own part of the company, I'm guessing. That we don't know. There was a compensatory legal settlement. So we know some value change stands, but we don't know yet that's cash or stock. So I have much experience in working with the music industry, which does take IP very seriously. I would say they take it the most seriously of anybody. And so what they will do is they let startups do all kinds of experimental stuff. They don't say anything. They set a trap. This is my belief. This is my conspiracy theory. The music industry sets a trap for startups. Here's how it works. You go do things. You do them non-commercial.

Somebody gives you advice, hey, you're not charging anybody. It'll be fine. They let you go. Then you get to a certain level of success. Once they see you have success, these are cut-through. wrote individuals in the music industry. Look up the history of the music business for a hundred years. These people are hard core gangsters. And I'm not saying like literally like in the mafia, but there's a history of the mafia being involved in the music business that I think creates an overhang. I'll be totally honest. I think the way they operate is like the mafia in that they will, hey, they'll let you gamble. They'll let you get in debt. And then they take over your sports store and then they have you purchase a bunch of stuff and then they burn it down.

This is what they do to startups. Never use their IP. Never, ever touch it. Not even for a second. In fact, only hire work for hire musicians to work for you and sign a work for hire agreement to put music into your product or make your own royalty free music. Then you can go to them and say, hey, we'd love to have one of your songs in here. We'd like to charge you this amount to be in our product and you come to them with that approach as opposed to using a 30 second clip and saying, I only use a 30 second clip. They will come after you like. Rabbit dogs.

More than that. No, more like in Game of Thrones, the white walkers, they'll come to you like the white walkers at the wall in the north. They are going to come forever for you and they will destroy your company. And the way they do that is they just file all these lawsuits, all these legal letters, and then they get you to sign a settlement for millions of dollars. And there's like five different companies that will sue you. And one or two sue you. You start to settle with them.

The third shows up. They know what the settlement is. I think it's all a rack. And I think it's all coordinated. And that's why we don't really invest in any companies that touch the music industry. Well. I'm glad that they does. But on that note, let's talk about this company and see if they have figured out a way to throw.

Fred the needle, producer Oliver. Talk to us about Soon-Omer, my man. Hey, listen, we meet a lot of early-stage founders here at launch, my investment company, and some, they don't have a lot of traction yet. They just have an idea. Maybe they haven't even finished their product. They've just got an MVP. But they still need investors and accelerators like ours to take them seriously.

And you know what? We can't just wire money to your Gmail or your PayPal. That's not how it works, folks. We need to know that you're a legit and official business. We need to know your company is incorporated. That's why you need Northwest Registered Agent. It's the service that will help you run your business the right way from day one. In 10 clicks and in under 10 minutes, you're going to file for your LLC or a C-Corp if you're a

startup, get a domain name, launch your official website, claim your business email, and even fast track your trademark application, which some people forget to do. We're talking about more than just company formation. This is your entire identity as a business. Go to Northwest Registeredagent.com slash twist and show the world you're in business. and make sure you use that URL slash twist so they know that we sent you. Producer Oliver. Researcher Oliver is here.

How are you doing, Oliver? Great. I've just been spending a lot of time in Sino this week. It's been really fun to play with. And I do come from a background. I've spent a lot of time producing. I've been in a band previously in a previous life. So I had a bit of previous knowledge that one before I'd

of into Suno. So I'm excited to kind of show you some of my takeaways. I think something that's really important to note is that Suno is a company that's making AI-generated music. They recently released their V5 model, which was just their newest, most upgraded model. It's great fidelity. The vocals sound really clear. Overall, the song structures much better. But they also recently release Suno Studio, which is a huge upgrade from what they previously. had because it allows you to dive deeper once you start experimenting with the tracks that you made. So now you're able to kind of get a little deeper. So does it have tracks in there that are known artists?

Can you pick a Lady Gaga track or does it have similar to a Lady Gaga song? How does it work? Walk us through making a song here in real time. What I'm going to do here is kind of walk you through what I'm seeing on the screen. So on the left here, this is the create section. And this is kind of what's, what was there in the previous model before Suno Studio. And this is where you can add some lyrics, kind of talk about what style you want. Then sooner will generate a song for you from there.

This is kind of a library. Got it. Library in the middle. And then this is kind of the timeline where you can kind of edit your project. And then this is detail. So we'll get into it. So one thing I love to do to create the track is get over to chat, CBT, and kind of have it make a prompt in the style. Okay.

Let's do it in the style of dire straits and the song, Sultons of Swing, or. Sultans of Swing and Telegraph Road and Love Over Gold. Perfect. So let's see if it will make a proper prompt or it's going to tell you, wow, look at this. So you're using Chat ChagipT to make the prompt great. Yes, because honestly, for many people, even myself, it's pretty hard to describe a genre of music for some reason. There's so many different styles. A lot of my favorite artists are make EDM music, but they're so different and unique.

so it can be hard. So Chachibouti helped me create a prompt for Dyer Straits. It started a smooth, story-driven rock song in the style of Dial Straits, warm, clean, electric guitar tones with expressive finger-picking and melodic solos, kind of mid-tempo groove with tight drums. So it goes on to continue to describe kind of more of Dyer Straits style. And you do want to kind of put in some lyrics so you can actually generate lyrics directly in Suno. So, okay, so here we'll describe the song. The song will be the story of a kid from Brooklyn who breaks his way into the venture capital industry

and becomes a legendary angel investor and podcaster in 20 years of hard work. All right. This is, I finally have Mark Knopfler making a song about me. No, I'm really interested to see you. I'm like, what's it going to do with that? I mean, who knows? Yeah, I'm curious. I mean, these LOMs are powerful.

We can maybe try this one. Kid from Brooklyn, holds in his shoes, concrete jungle, taught him how to choose, late nights, cold pizza. Yeah, we can, we can select this option here. Chasing a muse, wow. Okay. It gets pretty spicy. So one thing to note is once you hit crate, we filled in the lyrics and we've filled in the style, and then we can create.

And it always makes two versions of whatever your prompt is contained artist's name. So we got a little bit of an error here. So this goes speaks to IP. We have to take our dire straits. So I would say in the style, oh, yeah, you're just going to take out the band name and you've got to take out Mark Knopfler's name. A storytelling tone, right. You just take out the like. And let's see if it does it this time.

So like any LLM, there's or any model, you're able to kind of get around it. I have written artist names in here. It doesn't flag it. I think it kind of just depends on how it's feeling. So you can see it. Well, it did it, which is incredible. Okay, now let's see what you came up with here. This is going to get interesting.

So we can listen to these. It does sound like a little bit like Joshis. It's like bluesy. You got the finger picking of the guitar. That's very much like late stage, Mark Knopfler, like privateering. Oh my God. Late nights cold pizza, chasing a muse.

Dream stacked higher than the skyline views. Oh, yeah. but a notebook a name and a spark pitching big dreams on the street light dog from the man. We can check it the other one too. I mean, Jason has never been happier. I think you really-

This is hilarious. Oh, no, this sounds like John Mayer, which by the way, John Mayer, one of his muses is Mark Knopfler. So his entire album,

sob rock, including Wild Blue, the best track on, Saabrock by John Mayer is a specific tribute to Mark Knopfler's 80s, finger-picking guitar, blues style. It's incredible. Yeah. Now, what can you do in terms of if you want to change it?

Like, if I wanted to change it around. Yeah, so you can drag it directly into the timeline over here. And so you can hear it. This is the song that we chose. I think, Jason, you liked the version one as well. So what I love to do is. You have to click on the track, and then you can click extract stems. So this is something that I've kind of been seeing more with different tools where you can actually, it'll use artificial intelligence to extract each channel, each instrument.

So this usually takes a little bit to load. I see that. So it's breaking out the vocals, the B vocals, the drums, the bass, the guitar, the keyboard, the percussion, the strings, the synth, the fx. And the, you know, synth was a big part of the 80s there. brass and woodwinds. And it's interesting that it's doing this after it made the complete track, Alex. It made the complete track,

and then it's using AI to break it down. I thought it would have made the individual components and then made it into a track. This is wild, Jason, because I remember playing with Suno and Nudio back when they were pretty nascent, and it was like a 30 second snippet. You couldn't change much.

Now it is so much more granular. Oliver, this reminds me quite a lot of a digital audio workstation like Ableton that I'm familiar with. Yeah, so Ableton is the tool I spent a lot of time. Not that familiar with it. Well, never said it out loud, apparently.

Yeah, no, Ableton is a tool that I've spent a lot of time in. This is like, it's called a digital audio workstation. This is where a lot of producers spend most of their time, even when they're, you know, whenever they're recording, they're typically recording into something like this. You can see each channel. You can connect a guitar, a bass, percussion. There's a ton of different knobs. Here you can see you have all your plugins, effects.

and you also have all of the different, like here you can have a MIDI scale. Here you have your different effects. So different tools. I got it. Yeah, so that's kind of what Suno was looking to implement here as producers are familiar with this style. So here you just split it all apart.

All of the. How cool is that? Now you can isolate just the guitar and play me just the guitar. Is that correct? Yep. So we'll do it. Go ahead and just hear the guitar here. Now can I just hear the guitar here?

Now, can I just change just that with a prompt, the guitar? Can I, like, give it a prompt and say, do a little more finger-picking? Yes, you can. So that's something that's really cool. So let's try it out. Okay, so we can- I would say a little more finger-picking and slower finger-picking on the electric guitar. That sounds like a stratacaster.

Or perhaps sire drates. Don't do that. You highlight this section, and then you can just click or place. So I'll actually highlight the whole intro here. You want to make sure that you're selecting guitar and then click replace. And it'll start to download that. But we can listen to the soloed vocals here as well. Okay, so let's do one on there too.

Let's do for the vocals, just pick the entire track if you can. I don't know if you can highlight the entire track, yeah. And then say less country, more raspy, less country. Oh, and I should have said English too, but yeah. I did try and add a accent to a track that I was working on, and it didn't really work. So hopefully that's coming soon. This is unbelievable. I'm kind of blown away.

Yeah. I want to play with this immediately. The world's greatest moderator and the world's greatest danger and investor needs the world's greatest solution for managing his fund. So I want to tell you about PaperOS and why we use it here at our own companies launch in the syndicate. PaperOS has helped over 10,000 founders and investors, including may automate their workflows.

Their tools will help take you through handling the operational details I don't want to worry about because I'm busy meeting founders and looking for great companies. PaperOS has made the onboarding process so seamless for limited partners. Their investor intake forms, automated subscriptions, it just takes all the busy work out of managing our syndicate. And it automates capital calls, investor accreditation tax filings. These are the chores of running a fund or a syndicate. And they have saved me and my team hundreds. actually now it's probably thousands of hours. Our friends at PaperOS want to give you a complimentary $10,000 credit.

Just go to PaperOS.com slash twist. And if you prefer price discrimination, you can go to PaperOS.com slash disgracian. What do you have to pay Oliver to get access to this? What does Suno charge you? Yeah, so this is, you have to have those Premier Plan. It is $30 per month. You get 10,000 credits.

I've been playing with Suno basically all week. and I'm only 1,500 credits through. So you really have to be abusing Suno. So it's basically free. Let's call it what it is. Yes, exactly. It's free. A dollar a day is free.

If it costs, let me just explain to people, if it costs more than sweet greens, it's free. Of course, less than sweet greens. It's also known as free. Anybody can, an Uber driver, a door dasher who's delivering sweet greens, can afford to eat at sweet greens. If something costs less than your average sweet greens order, which I think is about $25 bucks if you get a limit,

Okay, so let's hear this guitar. Oh my God, I did it right. I was what I was looking for. Either I'm a master prompter or this is really smart. So you can see... It was more raspy too. Wait, did it do the raspy? Did it update the vocals?

It is a bit more raspy. Perfect. Sounds more autotuned actually. Yeah. Anyway, whatever this is, it's super fantastic. Can you upload a person's voice and say, I want it to be in this person's voice without like telling it. So if we like I was talking about in the previous example for IP with Higgs field. Higgs field. One word. Higgs field. So with Hagsfield, you know, if you do it, it's like not their problem. You're just using the tool. So with this tool,

could I upload Sultons of Swing, you know, and a Mark Knopfler track, especially if I found an isolated one and say, make it sound like this guy? So from what I've experienced, if you upload a version of yourself singing and then you kind of mess with it in Suno. It'll have the same style, but it won't sound exactly like you yet. I think that will be a feature that is definitely going to come. So one thing I can show you here is a project that I kind of been working on and kind of give you a little more details on ways that you can use Suno. So one thing that's really cool is I, this is the track I was working on. It's kind of more EDM. this is the one I use.

Let's get into it. Alex, we just drops some MDMA on the day. Go ahead, sorry. I'm trying to do the, you know what Dan's this is. I know what that is. That's the Donald Trump. Totally inappropriate. I know, but I'm imitating to the dozen.

Get your mind out of the gutter, Mr. Calcanus. I'm just saying it's a little bit graphic that dead, but keep going here. So I used one prompt and it gave me these two different tracks. And so I used one. of them but the other one was still I still like some components of it so what I did is I extracted the stems of the other one that I didn't use and I really like the drums so I actually brought the drums into the other song and I was able to take this part and like you like working in a dog is providing value like working in a dot I was able to kind of just chop it up and

And one thing that I think a lot of people don't realize is music is super repetitive. So I can just click Command D and I think that that will kind of do what I wanted to do. So I wanted to... You're looping it essentially. Looping it. So originally there was no drums in the intro. And I kind of wanted to get a little more energy in the beginning of the song. Wow.

It's a banger. What about key changes and time signature and tempo changes? Like how nerdy can we get in music theory terms? So we can get pretty nerdy. I think that in terms of the beats per minute and kind of staying on the same time grid throughout the whole song, you do have an option to select a track

and control the BPMs down here. So for those of you aren't familiar with music, every song is at a certain kind of beats for a minute. So it usually follows the kick. So doom, do, do. And this song, most house songs are in 120 to 130 BPM. So Suno understood that this is an EDM track and they made it at this tempo and every single stem.

And when I added the stems from the other song, it always lines up, which is a great feature to have here. That's why I was asking, because what if one was in 7, 8 and the other one was in, I don't know, 3-9, right? I don't think that things that- I'm assuming that Sunoz kind of stick with 4-4 time signature, which is kind of, you know, 1, 2, 3, 4, 1, 2, 3,

which is very the common one. 3, 4 and other time signatures are way less common. especially in kind of pop music. So I think Suno is definitely focused on that originally. And one thing that I do want to point out here is I was actually able to create a section using my voice. So we can give this a listen. So this is a guitar idea I had.

And I kind of had this empty section in the middle of the song. You can hear it here. Oh, that's cool. So I was able to prompt my voice to become a guitar. And this is a little more complicated than it seems. You can't just kind of say, hey, make my voice a guitar in this section. You kind of have to do a little more digging, which you can see, which I do on my demo, which you can watch later today.

And but yeah, you can see it really did a great job, kind of copying what I envisioned. All right, so Oliver, given your music background, given that you've been in bands like I have been, what do you think about this? How impressed are you going to scale with one to 10 and are you going to keep paying for it even after you're done testing it out for Twist? I think I'll definitely keep paying for it. I think that this is just V1.00 beta of this tool. So I do think that they will start incorporating other aspects of a tool like Ableton or Logic where you have more ability to really play with the track. And I think that something to point out here is for non-musicians,

the tool like this really allows, opens up the world of creating music to millions of people. And that's why we've seen crazy ARR numbers in revenue from Suno. And I know they're raising it a higher valuation. So I think that not only is it great for non-musicians to play with, to be able to create music for the first time, producers can also use this to help them with their ideas. And now I can kind of take ideas and make them into full songs and show my friends. hey, do you like this idea? I'm thinking about working on this more, and maybe they'll say, yeah, go keep going.

But I do think that there's definitely some things that are lacking. So I think that I'm going to rate this tool a seven at the moment. I think that there's still a lot of room for improvement. But I think this is really the first great tool that I've messed around with in terms of AI and music. I think something to think about is where will the biggest impact be at first? I think obviously it's kind of the fun factor. that I think if you think about score composers, that's a huge market. Buying a song for your TV show or commercial, that process is usually takes, you know, a couple weeks, even a couple months. You're going back and forth with the, usually it's like a freelance artist who you hire one time for that gig.

You'd go back and forth. Yeah, this clearly is the huge win. If you need stock music, you want to make a song for somebody's wedding, like a parody song, or you're doing music for your event and you don't want to have to pay for live event, music, license it, you know, like if you do a conference and you were to use somebody else's song in your conference, they might ask you to pay a fee for it. So yeah, it's just absolutely incredible. Great job, producer Oliver.

Okay, go back to work. Get out of here. Get that kid back to work. We got a little bit more on the docket we need to get there. So let's rocket docket. It's rocket docket time. We were talking about cloud earnings this week. curious if the AI world is spending too much money on compute. The answer, Jason, is at least not

yet. Three big numbers for everybody. Growth at Azure for Microsoft was 40% in the most recent quarter. Google Cloud, 34%. Amazon, 20.2%, the best in 11 quarters over at Amazon. Each of the companies also said they're going to spend even more money than they thought. For example, Google said, we now expect CAPEX to be in the range of $91 to $93 billion up from our previous estimate of $85. That's for this year. So in other words, people are still spending, they're still growing, and everyone said we are compute constrained. So Jason, I'm curious, does this mean the party is going to keep going for longer? Absolutely. If these firms can make the investment from cash flow without taking loans or taking additional investing and diluting

their cap table, what a great use of... of cash on the balance sheet to build up your infrastructure and have customers willing to take the excess, which is to say when we had Elon on All In, which will come out at the same time you're listening to this. You can jump from this week in startups and All In and get tons of J-Cal. We were talking about GROCAPedia. Now, GROCAPDia, my GROCAPDia page, if you pull it up, is five times longer than my Wikipedia page.

It's also more accurate. It's also more up to date. My Grogopedia in point one, I would say, is five times or ten times better than my Wikipedia page. And I'll have editorial director, Lon, come in and talk about this a bit. One of the things you'll notice, if you go to the bottom of my page, is the number of citations. So how many citations do I have on that page? And then tell me how many I have on my Wikipedia page. I think it's like triple.

It's 177 on the longer Grogapedia. Jason Calcane is on Wikipedia, as I fantastically type as. we do this live. Under 50. Yeah. 54. Okay. And then what happens is on my Wikipedia page, anybody who I fired, anybody who I competed

with in the marketplace, anybody who's mentally ill, they go crazy on my page. If you're part of the Jeffrey Epstein, like, you know, cult of people, the Pizagate people, you're obsessed with me because I appear in Jeffrey Epstein's black book. I never went to the island. I met him at the TED Commerce. I've talked about it a thousand times. I give him my business card. He puts it in his black book along with 5,000 other people.

So now those people come to my page and we have to make this a big part of Jason Calacanus Square. I met the guy like a half dozen times. He was donating to Marvin Minsky and Daniel Dennett. I don't know who. He was like literally surrounded by scientists and giving them donations. This is long before all this craziness. I'm talking about like the late 90s, early 2000s before he was ever, you know, a key

of anything or went to jail. Anyway, the point is, they're relentlessly focused on these weird things that happened in my life, not like the balance of my life and what's important. So, Grockapedia, point one, powered by GROC, any LLM could rebuild GROC people. But he was saying that I think he said that it thinks that they're used, they use 10,000 H-100s to like bill this or something. And then he said they were using a similar number to do. the new GROC algorithm.

And he said, the GROC algorithm on Twitter now, the new one, is reading 10% of like the 100 million tweets every day. And then making sense of them in real time with like 10,000 or 20,000 of these things. That's something that like Twitter would never have done before AI. What that means is you're going to be able to do a very natural language search on Twitter soon. If you're Google, if you're Apple, if you're Amazon, Oracle, any of these companies, and you got cash laying around, why not build an ad hoc? competitor or co-existor. I think the power move for Elon, for Claude, for OpenAI, for Apple, and for

meta would be to launch a cloud service so that they could offer it to their developers. Apple having an AWS competitor would be mind-blowing because they have all these app developers. What if they said if you're an app developer, you get $10,000 a month in free credits at Apple Web Services, AWS? Is that That's legal, right? You could say it would cause confusion, so they would not do that. They would say Apple API Services or something, AA. Yeah, they'd have to. Apple Cloud Services, ACS.

The acronym being the same would not be a big deal unless they marketed the acronym. That would be the big deal. But if they just market. I meant offering free credits for their in-house service to compete with ADWS because, well, they've gotten some trouble with the regulations there. Yeah, if you were to look at. antitrust, they would have 1% of the cloud market. Therefore, you know, are they in some way using their dominance and smartphones to do it? No. If they said you could only have apps in the

app store if you were a customer and spend at least $10,000 with Apple Cloud Services, ACS, yes. So it's a great question, though. I'll be very good to see. But I think meta is the obvious candidate, Jason, because meta is spending all this money. And then offering that much. And Elon, look what he's doing with Colossus. I mean, he knows how to build a lot of these things. A crazy performance from Amazon. And I think Amazon is my pick for the stock of the next five years.

Because of the headcount reductions, reacceleration of growth at Azure and the ads business, I presume. Well, Azure is Microsoft. I'm saying, AWS, yeah. Friday Brain, everyone. Don't worry about it. I just want to make sure we're talking about the same company. So with Amazon, I think they have the most to gain. from AI that they have not yet realized.

Ah, which is to say they're behind. So the growth hasn't shown up now. They're the worst performing of the Mac 7. So sometimes you're behind and you got to like catch up, dropping 10% of the head count in a quarter where you smashed it. And then they need to have their own language model. I think they need to try to buy XAI or Claude or Mistral. They need to buy one of those teams, perplexity or just, you know, maybe

become the open source champion of the world. Yeah. And they have work to do because people will have their services on AWS, but they'll use a language model directly with XAI, with quad, you know, an anthropic. So you can be multi-cloud and do your AI one place, but they have to catch up on that. It's going to, I think if we're sitting here,

I think Amazon will be the number one performer in the, it will be the number, one, two, or three, performer in the Mag 7 over the next five years. Next up on the docket. Speaking of AI models, do you know who's making them, Jason? Startups, surprisingly enough. Yeah. So this week, Cursor launched an update to their experience, also including a new model

called Composer, that they say is roughly performant with existing state-of-the-air models, but is a lot faster. Then Canva launched their Canva design model this week, what they claim is, quote, the first AI model trained to understand the full complexity of design. And then Winsurf, which of course is now part of cognition, launched an update to their first model called SWE1, now SWE1.5. And these are all pretty cool for two reasons.

One, it's fun to see startups not want to pay Anthropics margins for them. But also the thought here, and this is also confirmed by DD-D-DAS over at Minlo, is that these are basically reskinned and retrained Chinese open-source models, perhaps from Minimax, perhaps from whoever made the G-LM family. But in a world where you can kind of roll your own model, Jason, pretty quickly, that's pretty damn near close to the state of the art. Why would you go pay, you know, XAIs, Anthropics, or Open AIs gross margins for them? Well, especially if, you know, because Curser was an anthropic partner for the first

couple of years, I believe. And so this is the natural tension between platforms and the application layer. If you built a 99-cent flashlight in the app store in the first couple of years or a calculator, you woke up one day and Apple had released a calculator or flashlight. And yours was, yeah, there it is. It's not necessary. So that's what's going to happen in the LLM space. They're going to say, you know what, we should really have the cursor business. We should really have the co-pilot business. We're going to just make it one of our business lines at open AI at Anthropic, et cetera. So there you haven't.

They have no choice but to do it. It's existential. Yes. And if they have an open source model, well, then they could be making progress. And for consumers, there's no difference between paying $10, 20, or $30 for business consumers for these products. In fact, if they gave you but 1% more efficiency as a developer, you'd buy all three and spend $300 a month.

Who cares? $100 each. because developers time is so valuable. And so it is the absolute right thing to do for Cursor. They're going to accrue a lot of value. And what Cursor should actually be doing is they should be selling this to anybody who wants it. So they should have an API. And they should allow anybody building developer tools to use their Cursor API.

And then they would be monetizing it on top of, you know, that. In other words, let their competitors use it too. Who cares? So to be clear, what we have here is state subsidized data centers in China being used by venture-back Chinese startups to make open-source models for American tech companies to retrain to offer to other developers to offer to their customers. This is the modern economy. Jason, just to back up what you're saying about the risk of being taken over by OpenAI or a similar company, here's a news item that we're not going to talk about in detail from yesterday. Introducing ArtVARC, OpenAI's Agentic Security researcher. That's the future.

They're coming for the app layer, so maybe you don't want to be dependent on them. If I was a developer of any kind, I would never work with Sam Altman and Open AI. Let me say that. This is a warning. People can clip this. This is a warning for anybody dumb enough to use Sam Altman's OpenAI API. They are studying you. Remember, Sam Haltman's been around the block.

I've known him since Luke. He's an incredibly savvy person, and he wants every bit of revenue from the ecosystem. He isn't taking no prisoners. He's going to study how you're using the API, which he has the right to do. Sam Altman comes from the Zuckerberg School of Business, which is give dumb people, access to your tools, study them, and like the Borg, steal every innovation they have. And Zuckerberg got it from Bill Gates and Microsoft. Microsoft had a platform and operating system.

They let people build Lotus 1, 2, 3. Then they did Microsoft Excel. They let people build a product called Word Perfect and WordStar, and then they built Microsoft Word. RIP. They were more than happy to have everybody at their developer conferences showing off their work, giving them awards. I'm talking about Microsoft now.

Zuckerberg did the same thing at Facebook with their platform. Oh, Zingas, their big partner. And then they shivd them. I guarantee you if you're using and you're trusting Sam Altman from OpenAI and their API, they're studying your data. They're studying your usage patterns. I guarantee it. And they're allowed to do it.

Now, they're not going to study your data, but they're going to see how much of the API you use. That's all they need to know. You remember they gave an award at that developer conference, and they gave everybody a plaque for how much they spent with Open AI? Yes. Somebody can pull up the image here and show them giving out. All these people, these idiots were like, oh my God, I got an award from Sam Altman for giving him my money. Here's one from McKenzie, the consulting company, Jason. Okay, well, that's a slightly different situation. McKenzie should be using it. 100 million tokens. Look at that. Okay, great. So, but now if you were building cursor and you're

like, I bought a billion tokens, and they know the list of, who's using the most tokens. As somebody moves up the list, that's their roadmap. So if cursor was number one and I don't know, what's another tool that could be using it? Suno is like in the top 20. I guarantee you if Suno is feeding their information in there and it goes into the top 100, there's a business development product manager saying, oh, we should do a music platform.

And then they will just go step by step and steal, you know, I don't want to say steal, copy, liberate every innovation you have. they have the platform so they're going to roll over you. Do not use their API. It's a trap. It's a trap. Insert into the post here, Admiral Ackbar, saying, it's a trap. And then take me in the facial thing and make me into Admiral Adbar saying, it's a trap and make a meme out of it, producer Marcus, and release it over the weekend. That's your weekend homework. First of all, I mean, it's Akbar. And I'll just say that it was going to be very interesting to see who wins between open AI and Anthropic and Similers, eating more and more of the app layer. And on the other hand, open source

Chinese model. Anthropics not trying to take the whole app later. They're very cognizant of that. They want to, yeah, they're very rarely dipping down to compete with their customers. So it's just something to think about. Here's a chart of Open AI customers that have used more than a trillion tokens. Sorry for the quality here, folks. It's not the best. But Duolingo, open router, indeed, Salesforce, Code Rabbit, I Solutions AI, no idea what that is. Tiger Analytics, ramp, a bridge.

Let's see who else we know here. Shopify Notion, Whoop, HubSpot. Of course, there's JetBrains, T-Mobiles, Indesk, Harvey, Canva, Cognition, etc. Okay, here's what's going to happen. I guarantee you, Duolingo will be built into opening a. Oh, guaranteed. 100%. Indeed.

And job search, 100%. They'll do open AI. They'll do chat GPT jobs. Boom. He's going to do every single business. Why? He wants to spend $1.4 trillion. In order to spend $1.4 trillion, you can't just be providing tokens in an API that go down 90%

a year. It's not possible. So how is he going to do it? He has to own your business. If you're using them, you're enabling your own demise. Do you feel the same way about all companies, like XAI, Anthropic as well? They straw and so forth? No, because if you just look at the amount of product on the application layer that OpenAI is producing,

they want to produce a device and compete with Apple. They want to produce, they have a browser competing with Chrome. They have a code thing. They released SORA to compete with Instagram and TikTok. They're going after every single vertical. Wake up people. you're going to see them go after every single vertical. They'll go after Craigslist.

They'll go after the travel sites Expedia. They're going to go after Apple. They're building a social network with Sora. They're going after everyone. Wake up. Wake up. If you use their API, you're educating them as to how many tokens you use. And while you might, you know, while you might have it set to not train,

Their algorithm, you know, this was all the arguments that people made when Facebook double-crossed everybody and took away the social graph. Remember that? Oh, vividly. They were going to give everybody access to the social graph. Everybody built all these amazing tools. And then Zuckerberg was like, we need to make more money. F this. And they got rid of the social graph.

You couldn't use a social graph. You couldn't post to people's pages. You couldn't post your latest harvest from, you know, whatever mafia game you were playing or your hand from poker. And they just took the entire business away from me. everybody. And if you wanted to do anything like Zynga, you had to buy Zuckerberg dollars. Zuckerbucks. Zuckerbucks. And literally use their currency, which then cost you 30%, which then killed your margins. And Zinga literally, after having a deep partnership with Zuckerberg,

got stabbed in the back. No, I think they got stabbed in the front. Sure. Yeah. And opening AI is going to go public. That's so clear what's happening here. If they're allowed to, because even though they flipped, and I guess this is the next story on the doctor. even though they flipped, remember the Elon lawsuit is still hanging out there. That was carved out of the settlement. So they still have to get that resolved. I don't have any insight into it, but we talked about it on all ends. You can go after this episode we talked about it over there. I thought we could do a little bet, a little bet for fun. We love to do bets here at Twist. And so Reuters reports that Open AI is talking to bankers, thinking about maybe a late 26, early 27

IPO. I found a polymarket, 46% of folks over the there think the company will go public before the end of next year. So less than half. I think it's 26 IPO, Jason. So for fun, I thought you and I could do a little bet on when they go public. Well, I think we both agree it's going to happen before the end of 26. Am I right? Dang it. Okay. I was hoping you were on the other side of that. So now we're just down to what month. So would you like to propose a month and we go over under? Or would you like to propose a valuation bet over under? What would you like to propose now? Let's do evaluation things. I think we're too aligned on on time to get in too much trouble there.

So Reuters, again, reports that they're thinking about a trillion. You pick a number, I pick over under. A trillion. Okay. And so this is the price it goes out at, not the end of day trading. Yes, this is their IPO price based on their per share on IPO price. And you say a trillion. They're trillion to keep it simple.

Okay. And so you're setting the line and I'm picking the over under and you'll take the other one. I'll take the other one for fun. Yeah. It's a normal $100 back. Right. But I mean, this is why you set the line. You don't have to set the line at just, you know, in terms of like being a sharp here and like really getting into wagering.

You can set whatever line you want. Oh, okay. Let's do, um, think it through. Let's do, uh, one point two five trillion. I'll take the under. I'll take the over. Okay. Got it.

Okay. For $100. And so here's what I want you to do, Marcus. I want you to make a calendar item for 30 days out. And for that date or a calendar item with a 30 day out notice with me, you and Alex on it and Lon on it. So make a calendar item and then put it on this week in startups.com slash bets. And I think we have a calendar there. If we don't, let's make a Google calendar so people can add the bets to the calendar.

We see it on our calendars. I'm always building systems. I'm always thinking in systems. When we make a bet live on the air, we make a calendar item immediately. Great bet. I'm going to say it happens in September, October. Exactly. Not over the summer.

You get the summer. They're going out on the road in the summer for sure. And then they just have this like right after Labor Day. So yeah, it's Memorial Day than Labor Day. Yeah. So I would say, you know, September, October is the right timeline for that IPO. The question is just price. Can we just do one tiny thing just to wrap us up today about an IPO that I think is pretty important?

Okay. So Navon, formerly known as Trip Actions, Price is IPO at $25 a share right in the middle. Between $6 and $7 billion evaluation, depending on how you count shares. there. So a little bit down from its peak valuation, but most people that backed the company did very, very well. And then it kind of flopped. It lost 20% on its first day. And the question I just wanted to ask you, and I wanted to get this in before we go is how bad is this? Because you and I both wanted to see more IPOs, more liquidity, more deals. But to see Navant, a well-known company with 30% growth, falling losses, improving gross margins, really stumble out the gate surprised me. So how bad is it?

Well, usually when this happens, it's because either people fundamentally are not excited about the core business. I think it's being dumped on retail or it was mispriced, right? So is there hair on the dog is the first thing we have to ask? And then the second thing is, did they just price it wrong? So we can go in reverse order. In terms of the price, what's the price sales ratio here? What's the market cap? What's the price sales ratio?

It's valued at $6, $7 billion. right? And they went out at $25 a share. And what's their revenue right now? So if we get their revenue for the year, then we can divide the revenue into the valuation. So the revenue, Jason, in the most recent six-month periods, we can kind of multiply that by two, was a total of $330 million, let's say. So call it $660 run rate. All right. So we'll go to $700. If it's at a $7 billion dollar valuation, it's only 10 times, nine times price to sales ratio. Okay, so that's not insane. It's 30% growth.

So if it grows 30%, you know, that price of sales ratio would come down a bit. But it's not cheap. So it was priced high. Okay. Or the growth. So that's part of it. So people aren't excited about the price. Now let's look at the core business.

Are they losing money or making money? Are there earnings or are they losing money is the next thing we have to guess? because I am not familiar with Navvon. Okay, I'll just pull up the income statement here for everybody. The gist is they're still unprofitable, but as you can see right here, Jason, their operating losses have fallen by about 50% over the last year. So I would say pretty positive direction there.

Elsewhere in this S1A filing, their gross margins are improving. Elsewhere in this filing, they talk about using AI to control their customer support costs, keeping their G&A costs, growth slow. And I thought that was a pretty compelling story. Reasonable growth for an IPO over the $500 million run rate mark, improving profitability, improving revenue quality, and an AI story. It seemed like something that would at least stay flat at its IPO price.

So with six or $700, yeah. So if you're losing $180 million on $600 million and it's going down, okay, they only lost $100 million in the first half. Can they get that to go down even more, hopefully? But I think it's just not as attractive as other options. So then it becomes, what else can I buy? Where else can I put my money? And if you see Google, which was supposed to, remember a Euroguard was telling us like, oh, my God, the search franchise is over.

And I was like, no, I think the pie gets bigger. The pie got bigger, right? Like search was up. Again, the number of searches is going up. Why? People are searching and then they're seeing the AI summary. and they're like, oh, this is dope. Let me do more searches.

Yes. I tried to explain this to people. If the product gets better, consumption goes up. And so are people going to just automatically forget about Google or change it from Google to chat GPT in their browser settings? I mean, I might, you might, but we're the Vanguard. Yeah, we're the Vanguard. We're the top 10%. Search and research is going to grow 10%.

I'm sorry, 10x in the coming year. Why is it going to go 10X the number of searches? Because the results are so much better. They're so much more helpful. If you could do a search and figure out what's wrong with your dishwasher, I always love to use a dishwasher one and not have to call the plumber, you're like, this is an incredible experience. You know what?

People haven't had that experience yet. But this winter, when people's HVACs go off or their pipes freeze and they take a picture of it where they do a search on Google and it says, oh, you have this HVAC unit. it's known to do this. Here's how you reset it and relight the pilot and you don't have to call the person and you fixed it or, you know, your housekeeper was able to clear the dishwasher thing because suddenly they know how to, you know, take a picture of it and put the model number and say, what is ERA 72? And it's like ERA 72 is something's blocking and take out this thing and clear the blockage. How many times have you had somebody come to your house and they were like

fixed it in five minutes and you paid $150 or $300? More times that I wanted to admit, including some very embarrassing situations, but I just didn't understand something. They came and flipped switch. I literally had this happen. They called, I come into the kitchen. I'm in my robe, and there's a guy there, and he's like, yeah, what's wrong with the refrigerator?

I'm like, I don't know. I just live here, man. I just live here and pay the bills. I go over and he's like, yeah, you have to press this button to turn on your freezer. Your freezer was off. And I'm like, oh, I was like, how much is that going to cost me? He's like, oh, it's $150.

bucks. I'm like, oh, it took your five minutes. Can I pay you 50 in cash? And the guy was like, I'm sorry, I work for a business. That's 150 is like my time to drive over. It took me half an hour a drive and I got another half hour to drive back to the office. I was like, all right. That's a little bit 20 in his hand out the way at the door. That's what I was trying to do. I was like, hey, if I give you a 50 in cash, he's like, it wasn't his business. Yeah, I understand. I understand. So I, but I tried. So anyway, it's all going to go up. So why would I put my money into this company when I could put it into Amazon, which I just said is going to be tremendous. You know Amazon and Google are going to make crazy runs. Now, are they fully

valued? Yes, they're fully valued for a reason. I think the optimism you see in Palantir and Tesla is the optimism we should see in Uber and Google, which are starting to in Amazon. There are other companies that are leveraging massively AI and they're going to have an incredible outcome. But the sentiment has to change over time. And the sentiment takes a little time to change. The sentiment in Amazon changed. The sentiment in Google just changed. Sentiment is going the other way with Facebook, which is, hey, maybe you're spending too much here and not enough products. So they just got and they just made all those purchases. So it's going to be wild. For founders, you know, the lesson in all of this is there are so much money to be raised right now. I am seeing people. And I talked about this

on Twitter. I think it's a good place for us to end for the weekend. And it'll be a little stump speech here. But somebody was complaining that there were three startups they had heard of that had raised money, kept the money on convertible notes, and the founders didn't build anything. They just took the money. So show the original tweet, not mine. I am, okay, here's the original tweet.

It's from, I was from my friend Jason Limkin. I love Jason. Jason's great. One of the many Jason's in my life, one of four. All right. Here is our guy. So far, I've watched three VCC deals this year where the founders just decided to keep the money for themselves and not build the product. All were on safes and they took the position they owed investors nothing and that, in fact, they were not even investors, just safe holders.

What eventually returned cash after keeping a large chunk for themselves? Another hired an attorney who said plainly nothing was owed to the investors and the founders could pay themselves whatever they wanted. I did not personally invest in these deals, but I might have. They had a lot going for them or so it seemed. This is going to happen in the age of AI. it's the flip side of so many things also exploding so quickly and growing at insane rates. It's just a risk to be more aware of, perhaps, than in the past. I didn't know you could do that.

I thought when someone gave you money, you had to be responsible with it, Jason. What's this? So we live in a trust-based system. Venture capital and startups are a trust-based system. There are protective provisions in the documents. Hey, you've got to send us information. We have information rights. You've got to let us know if you're going to raise more money and dilute us.

We have these protective provisions against dilution of corporate governance, of board seats. But in YC's infinite wisdom, they decided to create this safe device, which was quick to help people fund. And in fact, a lot of founders just sign these and investors signed it without ever talking to attorney without ever changing them with like an expiration date or interest, like a convertible note. Because we went off the convertible note, which had more protective provisions, the safe. is so in favor of founders that, you know, with no expiration date on it, it can just go on forever. And I have seen myself founders sign a term sheet with a firm at, you know, ex-valuation. Which is a contract, right? It is. So you would have some honor and you'd have some ethics around that.

Then get another offer within 10 days. Then say the other contract's not valid. I'll get. you your money back, lie to the, lie to the investors, say we're shutting the company down, all kinds of hand gestures. Oh, then they would go to the investors say, you're not being founder friendly. These three can claim that. Now, if you were Jason Lemkin and one of these founders did it to you, you're like, okay, I put 150K and I put 250K and they screwed me. Am I going to sue them for it? Okay, that's going to cost a million dollars. Or am I going to write a legal letter? Then they're going to say, I'm suing a founder. So founders now know this. And there are bad actors in the system. Every system has bad actors. In a peak market, you see bad behavior go up.

Whether it's VCs on a board ousting the CEO, which was the pendulum the old way, now the pendulum can go the new way, which is the founders can abuse the trust in the system. This is a message to those founders who are doing this. And I've had countless messages to the VCs and investors. I'll save that for another time. But just for founders, play the long game. If somebody believed in you and they were your first investor, I've seen this situation, I've been in this situation, you're the first investor in the startup, you help them get it off the ground, they use your name, the fact that you were an investor in Uber and Robin Hood, they put your face in the deck, they ask you for introductions to 50 people, they get 20 meetings, they get more

investor, then they try to take away your rights, they try to take away your board seat, and then I have to have this discussion with the founders, hey, by the way, if you're new investors want to screw your earliest investors, then they will keep trying to do that. But once you've screwed your early investors, the next people to get screwed and the only people left are the founders. So what you want to do is you want to cherish your early investors. You want to say, oh, my God, Paul Graham and, you know, Gary Tan backed me first at YC or I backed you first or Angelus, Naval backed you. Wow, those are the really, those are the good guys and gals, Siam Bannister. Okay, I'm doing another round. These new investors who come in,

three years later when the business is on fire. And they say you should screw your existing investors and get rid of their prorata and tell them if we invest in your company, they have to, well, give up their prorata in their winner. But they weren't here for the last three years and helped you. And they didn't write you a check when nobody believed in you. And then you're willing to screw them because they said to you, oh, we're not going to do this deal. We'll give you a great valuation. We'll let you take some secondary. But you have to screw those people and take away their board say it. Take away their information rides. Take away their prorata. or whatever it is.

But, ethics. Okay. Well, what happens is the investors who come later try to like,

what's the character in Lord of the Rings worm ear? Worm tongue. Worm tongue? Grima worm tongue. Long for all future Lord of the Rings references,

I got them. Okay. But this character was speaking into the king's ear poison, right? And he was making bad decisions.

And the Rohan riders were like the most ethical and greatest people, and they poison the king's brain. This happens to founders, actually. Some new investors like, F, your early investors, they got too good of a deal. We're willing to pay you $50 million for this company. YC paid $2 million. Yeah, YC paid $2 million when it was just two people and an idea. When it was literally worth zero, by the way. When it was worth zero. Yes. And you loved it and you loved them for it. And you signed the deal and you were well informed And they got 7% so who cares?

This new person is going to get 15. And there's still plenty left. Cherish the early investors, act with honor. And then you will be in a situation like I am with Sierraou, where I got to invest in his first company, reportive. And we got to collaborate on superhuman or with Jonathan from Thumbtack, where I got to participate. And, you know, he valued my contribution with Athena. Go to Athenawow.com and get like two to four weeks for free because I'm a spokesperson for that company. I use my platform to build Athena Wow.

Wow. Athena. Wow. Even Christopher Walken wants to use it. It's not a celebrity endorsement. But I'm doing endorsements for Athena because I love Jonathan and the team so much. I want to see them. And he's loyal to me.

I'm loyal to him. I'm an investor in Cloud Kitchens. I was an investor in Uber. And I'm going to help Travis with his XG Gammon because we both. of backgammon and I've been crushing him in backgammon games head heads up and we're going to collaborate on doing a backgammon. This is the best part of life people is when you can build a 10, 20, 30 year friendship where you just collaborate on everything and you win and sometimes you lose. But if you lose, you learn something and you just get up. Don't fuck that up by trying to

be petty between rounds and like this person signed and wired and you want to rip that up because the next person offered you more and then they worm eard you. Worm-tonged. Worm-tonged you? And for these people who are worm-tonging, I'm on to you. I think it would be more tongue-worming. I'm on to you. No, literally somebody was like, oh my God, J-Cow is screwing and screwing founders.

Somebody said this and I was like, wait, what? You come to my accelerator. This is a rumor. Somebody said that's to me. I say, that's not, wait, what? You come every, and I called the person up and I was like, do we have a problem or something? And he's like, no, it's not true. And I'm like, okay, that's interesting. But like, we have tremendous relationships with all the other accelerators, all the seed funds,

because we find companies that they are too busy because they like to invest at a $10 million valuation. They don't want to do Founder University $1 million valuation. They don't want to do an accelerator, Y, combinator, $2 million valuation. They don't even want to do the $5 million valuation when you get to $10K a month in revenue. They want to get on the train. when it's $10 million valuation, and you have $500,000 or a million in revenue. And we know that. Well, 50%, I would say. 50%, 50%, yeah.

And I get on when it's 0% de-risk. And then I make a second bet when it's 20% de-risk. They get on a 50% de-risk. In the series A, it's 60% de-risk. The series B is 80% de-risk. It's 90% de-risk after that. Great. But let's not screw up a trust.

based system. And I really applaud Jason, who we've got to have them back on the pod. It's been too long. But Jason's amazing. And he's taking a little risk by putting that information out there. There's bad behavior all over the ecosystem. Just play nice. Words of Michael Dell, win but play nice. And I told this to a founder. I was like, can I be candid with you? the only way for my investment in your company, like if we own 7% of the company, and I get 20% of the gain on that 7% so if that, if our 7% became worth 70 million, our firm gets like 20% of that, it's $14 million. And then I give some of that to my team. And then I pay taxes on it.

It literally would not change my net worth. It would be a couple million bucks. It doesn't do anything for me in 2025. You know, I'm past doing something for me. You would need to be a hundred billion dollar company, a $200 billion company for me to notice it. So all the work I'm doing is because I love founders and I love you and I want to see you succeed. I'm not doing it for money, folks. This is an important thing to understand.

The best investors, when I saw Doug Leone and Michael Moritz at 8 a.m. doing meetings at Sequoia after Google, after Apple, after Cisco, after. Airbnb, after Zappos. And then I came back at the end of the day one time to meet with Ruloff and it was like 435 and they're still in meetings with different founders in the same conference rooms. They're doing back-to-pack meetings trying to help their portfolio companies. Like there are some people who are in it for the right reason. Align yourself with those folks.

And you can tell because they're just, they're working hard. They're working hard to try to make things happen for their partners. Win, but play nice. And you go further together. If you want to go fast, go along. Sure. If you want to go far, go together, folks. It's been another amazing week of this week in startups. He's Alex Willem. Cautious optimism is where you're going to take your $100 and get $1,000 in value. You're going to go read his daily newsletter. Throw him the hundy. He's got kids to feed.

And you've got knowledge you need to gain. He's at Alex on Twitter. I'm at Jason. And Lon Harris is at Lonz and producer Oliver. Maybe in a year, if you do as good as you did today, we'll shout you out and get you couple followers. But for now, heads down, everybody go check out. TWAI.substack.com. TWAI.com. TWA dot substack.com. Everybody go visit TWIAI.a.i.com. This week in A.com. Just five letters. TwIAI. Substack.com. I'm going to put in the notes coming next year. All demos all the time with a roundtable every week. So it'll be kind of like the all in round table or the liquidity roundtable we do here when we have like for investors on a little roundtable. And I'm recruiting AI experts to be part of that.

I'm looking for like, you know, Chumoth Sachs-level people, like really high-end folks to do that roundtable for me. And then we're going to have demos between the weekly episodes. I just like my guy, producer Oliver, that credible demo he did today. Imagine a couple of those in between the roundtables coming to you, coming at you in the new year. Have a great weekend, everybody.

Enjoy your Halloween. Stay safe. Bye, bye.