This Week in Startups - 2026 Starts with a bang: META AI Drama and Nvidia’s $20B Groq Acquisition | E2230

Episode Date: January 6, 2026

This Week In Startups is made possible by:Crusoe Cloud - https://crusoe.ai/buildUber - http://uber.com/twistEvery.io - http://every.io/Today’s show: Jason and Alex are BACK on TWiST for 2026! This holiday season was anything but calm, with deca-corn acquisitions, massive Polymarket bets, and major new startups breaking from stealth!Jason talks the recent Nvidia-Groq $20B acquisition, a major exit for Chamath as the lead investor back in 2017! Jason delves into how the VC fund math shapes out for pre-seed VC funds vs. Series A VC funds.Jason and Alex delve into drama swirling META’s AI team. Yann LeCun, META’s former Chief AI Scientist, announced that he would be leaving META to become Executive Chairman at AMI Labs. LeCun left the META team in the new year, calling the new Chief AI Scientist, Alexandr Wang, inexperienced. LeCun now looks to move AI beyond the era of LLM at AMI Labs.PLUS Jason and Alex talk about the new social media app Tangle, from Biz Stone, co-founder of Twitter, and Evan Sharp, co-founder of Pinterest. Their Startup, West Co, launched tangle, which seeks to become an “intentional living” app. The two look to improve how humans interact with modern tech. Jason points out that very few news products have worked, but is eager to see how two industry veterans build in the space. Timestamps:(00:00) Why Restaurants are OVER — Peptides and other self medications(06:41) Nvidia Acqui-Hires Groq for $20 BILLION(9:48) Crusoe Cloud: Crusoe is the AI factory company. Reliable infrastructure and expert support. Visit https://crusoe.ai/build to reserve your capacity for the latest GPUs today.(11:00) The VC fund math between seed vs. Series A funds(15:00) META buys TWiST 500 Company, Manus! Why it matters.(20:20) Uber AI Solutions: Your trusted partner to get AI to work in the real world. Book a demo with them TODAY at http://uber.com/twist(21:24) Why Yann LeCun left META, and what could be behind it(25:27) Producer Claude on the Gondola Crash in Zurich(29:13) Jason’s Request for Augmented human intelligence(30:11) Every.io - For all of your incorporation, banking, payroll, benefits, accounting, taxes or other back-office administration needs, visit http://every.io/(32:04) How one Trader made $436.8k on one bet on polymarket!(36:05) Jason’s Predictions for 2026 IPOs(40:01) Is news broken? How Tangle is tackling it.(45:53) How much should startup incur in legal expenses? Should founders try to use AI to avoid costs?(50:59) Why Google should let NotebookLM cook, make it a standalone brand! *Subscribe to the TWiST500 newsletter: https://ticker.thisweekinstartups.com/Check out the TWIST500: https://twist500.comSubscribe to This Week in Startups on Apple: https://rb.gy/v19fcp*Follow Lon:X: https://x.com/lons*Follow Alex:X: https://x.com/alexLinkedIn: https://www.linkedin.com/in/alexwilhelm/*Follow Jason:X: https://twitter.com/JasonLinkedIn: https://www.linkedin.com/in/jasoncalacanis/*Thank you to our partners:(9:48) Crusoe Cloud: Crusoe is the AI factory company. Reliable infrastructure and expert support. Visit https://crusoe.ai/build to reserve your capacity for the latest GPUs today.(20:20) Uber AI Solutions: Your trusted partner to get AI to work in the real world. Book a demo with them TODAY at http://uber.com/twist(30:11) Every.io - For all of your incorporation, banking, payroll, benefits, accounting, taxes or other back-office administration needs, visit http://every.io/

It seems that he collapsed $50,000 worth of legal work into essentially a couple of queries and a free-to-use tool if you already have G-speed. For founders out there who don't want to pay insane amounts of legal expenses, does this sort of like involved intelligent approach using modern tools provide a pathway to spend less money on things that are non-product critical? We had a company that had no lawyer, no accountant. They were doing it all themselves. And they had gotten to hundreds of thousands of revenue.

I know you don't have an attorney, but this will become a blocker for future rounds. if you're going to get to the Series A, the seed round, you start raising over a million dollars, like doing it on convertible notes. I'm like, no, no, I can do it all myself. I can do it with an LLM. So founders find a way. And so there is a product opportunity here. There's always more legal work to be done.

There's going to be 10 times as many startups as young people can't get jobs. This weekend startups is brought to you by every.io. For all of your incorporation, banking, payroll benefits, accounting, taxes, or other backoff administration needs. Visit every. i.o. Uber AI Solutions. your trusted partner to get AI to work in the real world.

Book a demo with them today at uber.com slash twist. Crusoe Cloud. Crusoe is the AI factory company, reliable infrastructure and expert support. Visit crusoe.a.ai slash build to reserve your capacity for the latest GPUs today. All right, everybody, welcome back to Twist. It's 2026. Twist 2026.

We're ready to go. First episode of the year. January 5th. Don't call it this week and sort of. Just say twist. Twist 20, 25. Just twist. It's simpler. It's cleaner. It's faster. It's better. Just twist. No more long names. Yeah, like a little twist. You have a little twist in it. Amen to that. In your case, a mocktail. It's all good. You know what I love? Jason is paying $12 for a mocktail. That always seems like such a good deal to me. Just a little fruit. Try having three daughters and all three want a mocktail. And you get hit for 36 bucks. And then they're like, can we have a second one? And I say absolutely not.

You may have a diet cook at most. I think I got to pump the brakes on this moktail thing. I'm going to give them the choice. Mocktail or dessert. Yes. Because it's just too expensive. It's way too expensive on the moktel thing. But I understand, you know, there's a big thing going on.

Speaking of startups, restaurants, I was talking to like pretty high level, very world-famous chef. I wouldn't say who. Okay. And I said, how's business? He says, it's over. I said, say again? He said, it's over. This is L.A.-based.

Okay. You know, you know his first name. Anyway, says, listen, it's over. And so please unpack. He said, the kids are not coming out. And when they do, there are, you know, people are on Ozempic, so they're ordering half as much food. Nobody drinks.

Everybody's smoking weed and doing ketamine or just abstaining. Again, some of them are on, you know, GLPs, which also tend to. So they come in and they spend $25 a person. We're losing money. because they order a salad or an appetizer or like a side of potatoes, this business is over. It's done. People like you, J. Cal, who come in or Tremoth and orders a bottle of wine or you come in and you want to try a lot of things, you don't mind bring a couple of doggy bags.

I always, you know, or you leave a big tip. He said this is like the end of it. It's over. And so it's really going to be hard, I think. They're losing all the mid-tier restaurants in Los Angeles. And then the high, high-end, you know, $300 a person, $200 a person. That will exist. And then sweet greens or shake shack.

But this middle tier of restaurant where you go spend, you know, 75 bucks a person for dinner. And four people go out, it's a $300 bill. You put a $100 tip in, whatever you do, $400. That's over. He said, people just don't order as much food. People don't want to go out to dinner because they're not hungry because they're on GLP's. It's a disaster. I got to get me on some of this GLP business because I really do feel like I'm the last

person not on them. And I'm kind of curious what it would rub off my personality. Like what burrs would it kind of clean away from me? Reda Trutide, if I'm pronouncing correct, is the one that all my friends are talking about. It's still very much in testing or whatever they call it trials. But because of the compounding pharmacies, there are compounding pharmacies making it. It is all the rage in Silicon Valley in my circles. People are doing low doses of it who are already incredibly muscular, et cetera.

Retta Trutide is the one if you wanted to double click on. and look at. I don't recommend going and doing anything that's not yet FDA approved or whatever, but I can tell you everybody's doing it. I'm not saying that you shouldn't. I'm just saying everyone else is. Well, and then the other thing that occurs is there are testing sites for peptides. So there was a peptide article, I think, in the New York Times this weekend. Everybody was talking about it. So there are testing sites now. So these compounding pharmacies in China are making these in bulk. then they ship them to US. Then these biohackers, Austin, the Valley, etc., are setting up testing sites.

They send it to an independent lab by batch. So you got this batch of Red of Trutide, this batch of BPC, this batch of whatever. Then they, and this is going to become, I think, a business, these are for testing purposes only. So they can't ban them. And if you use them yourself in your body, well, that's on you. But the testing sites are now saying, hey, we take. tested this, this is, you know, 96% pure, 98% pure. And then they keep a track record of the providers and the sellers online who are selling the stuff and their track record. So there's going to be

a whole cottage industry here. I would say it's very similar to what we saw happen in maybe cannabis, where it was kind of legally unknown or it was kind of avant-garde in Canada. They made a national referendum and then people started, you know, all these subsequent businesses around it. I think this will become one of the great businesses of the next two years. If that analogy holds, though, we're going to have like cask strength peptides because if there's one thing the cannabis industry has done, it's make weeds stronger. Too strong. Yeah.

I'm old enough now to agree. Yes. Too strong. I mean, it just they are making these shards or resins that are like a hundred times more powerful than, you know, the street joint that, you know, the Lucy that somebody might have had in the 90s. So be careful out there, folks. It really is terrible for kids because my understanding is it causes psychosis. And then that becomes a bridge to bigger things. So, you know, part of living in a free society, right? You know. All right. Let's talk about some startup news. We've been off live for a

couple of days, Jason. I grab some big stories that I think we should just touch on here. First one is the big Nvidia grok with a cue, grok with a cue. Uh, They're going to be licensing that technology for about $20 billion. GROC is a company that we've known here on the show for a long time. Your friend Soni Madra used to host some AI stuff with us, and he's going to be part of the deal. Going over to InVidia about $20 billion. The idea is that they will have a non-exclusive licensing agreement for GROC's inference technology, which is a fancy way to say that they're going to get to use what GROC has built to handle AI inference, Jason,

which is not training. Instead, it's what AI does when you ask it a question. Yeah. So it's a huge deal. And I said, we're going to start seeing some very big deals in 2026. And here is the first of the very big deals. This is a $20 billion deal. You know, anything above $10 billion is a, you know, significant M&A deal. Anything under $10 billion is, you know, a single or double. But this is the home run category of exits. I was an investor in Sunni Madras Company, Sondi Madras Company, definitive, which was acquired.

He became, I think, essentially the president and head of developer relations over. at Grock. They did fabulously well since he, you know, joined. He was a real accelerant. And now they're going to be part of this. It's one of these licensing deals, which makes it go pretty quick. The existing company will still be there. You license it. We've seen this over and over again. Character AI. Scale AI and Inflection AI. Inflection was the one bought by Microsoft. Scale was bought by Facebook meta. If I have all my memory correct. So this is a new way of doing deals. What's great about these deals? When you do a little bit of things, when you do a lot of licensing deal, you get to sell the company, and it just happens immediately. You know, you stroke the

pen and you're off to the races. So this is going to become one of the preferred reasons to do it. Now, there were people who speculated here on this program that maybe this was a way to get around the FTC and regulatory approval, which the FTC themselves, I believe, put a note out that, hey, under Lina Khan, we're looking at this. We kind of see what you're doing over there. You know, today, I think the whole concept is, don't work. about it. This is just a way to go faster. Now, there are tax implications because you're not buying a share in the company. And so then you have to pay a different tax treatment, I believe. Not a long-term capital gain. It would be more like ordinary income, sounds like. Probably is how this works out. I don't

know that definitively in this case or the others, but I will eventually know that because we were investors at my firm in definitive. And so, you know, I don't know if we'll get a 10x, a 20x, or something like that. But it'll be a nice little bink for us. It's not going to be a fund returner because, again, definitive was bought by GROC and then GROC had this incredible accelerant. But in a short period of time, you know, you'll take a 20X in a three year, if that's what it winds up being for us or a 10X. You know, we'll take it. And remember I talked about on previous programs, add infinitum, if I'm using that correctly, until I was blue in the face, basically. We need these singles and doubles to keep the venture space vibrant. And this will be great for me as a GP to send a

distribution to our LPs. Launching a new company is all about finding your first customers and then just learning how to solve their problems. And that is going to put you on a relentless pace. And that means you're going to be releasing new products and features, hopefully at a brisk pace or better. But my lord, the complexities of working with AI, it's going to slow your developers down. You're going to have to install new GPU drivers. They're going to be provisioning clusters. All of this detail-oriented labor-intensive work can distract you from working on the thing that you're building, your product. But now there's Crusoe. They are the AI factory that worries about your backend so you can focus on your code. And their hardware has been designed specifically for AI development,

so they're going to make sure that your models run with unparalleled performance every time. Head to crusoe.com.com. to apply and receive $100,000 in credits for virtualized Nvidia, GB-200 NVL-7-2 on Crusoe Cloud pending availability, okay? That's C-R-U-S-O-E dot AI slash build to apply and get those $100,000 in credits. This is not a double though, Jason, it's $20 billion. And, you know, everyone from social capital leading the series A and B, we had, you know, D-1 and Tiger and the B-1 and C, addition to alumni ventures, you.

this has to be like a home run, right? I mean, $20 billion. Oh, yeah. Grand Slam. Okay. Okay, got it. Got it. Yeah, I would say Grant Slam.

Anything over 10, anything in the billions is a home run typically, depending on when you got on the train. If you were the last round, no. But for everybody else should be a home run. Anything above 10 billions, the Grand Slam category for me. Now, for us, it'll wind up being a 20x for a seed fund. You know, for us, that's a home run, not a grand slam. because in our funds we will have 300 names, 300 companies in our fund, whereas a typical venture

fund that's doing Series A has 30 names in it. So each name represents 3% of the fund. In ours, each name represents, you know, 30 or 50 basis points of the fund, which means we need a 200x to have a fund returner if you just do the back of the envelope math. So a fund returner for a seed fund or a venture fund that has 30 names in it, 40 names in it, tends to be a 30 or, you know, a 40x, something like that 50x, you can return the whole fund. For us, we have a higher benchmark. So we spread more bets. We get an earlier, lower valuations and hope for the best.

But congratulations to the team over there. And I think it's just the start. I'm going to stick with my prediction, $100 billion will occur in 2026. A $100 billion deal will occur. Oh, a hard agree. For folks who are there curious why NVIDIA would want to buy part of the GROC technology stack. I think, Jason, the answer is just TPUs from Google. Jonathan Ross, who was one of the founders of GROC, was a former Google TPU engineer.

And if you recall, companies like Anthropic are turning to Google to use their technology. I think essentially, invidias like, look, GPUs are great, but we don't want to lose a trick on inference. So I think this is essentially derisking for Jensen and company. But that's just a little bit of context for folks out there who are curious. Well, the company's worth $3 trillion now, so this is not even 1%. Right? 1% would be $30 billion if it's a, or is it a $4 trillion company now? Anyway, it's like 50 basis points or something.

So 4.6. We're both wrong. Wow. Oh, my God. I just, you know, I keep losing track because things keep getting so big. So anyway, you know, if it's under 50 basis points of ownership dilution for the shareholders, if it doesn't work, who cares? If it does work, oh boy.

And this is where, you know, Nvidia should be buying another. Invita should be buying conservatively 10 companies a year, maybe 20, given the market cap of the company. If they can use their shares, they should be buying 20 companies a year, two a month, one every other week. They should be on a explosive M&A because what if they work? Just like Palo Alto networks, just like Microsoft or Google, you know, during peak periods, you want to put that high market cap to work.

So this is going to be great for founders. you don't want to build your company to be bought, but you do want to have that kind of in the back of your mind. If this company can't grow revenues and get to $100 billion, I get to $100 million, $200 million in revenue, which is then you start just thinking about IPOs or you're just going to be massively sustainable. You know, then you're, you want to start thinking about who would buy this company and why. The obvious answer here of why to buy it is now they have a complete package. You have the inference. They take it off the market.

if the inference winds up, you know, being the more important part of the puzzle and people like Claude and Anthropic are saying, hey, we're going to do less, we're going to do more with less. We're going to use less compute to make our models. And that becomes the trend. Well, that's not good for Nvidia because maybe people overbuilt and now they don't need more. So very complicated chessboard. Congratulations to them. What else happened over the break? So Butterfly Effect was a Chinese company, eventually moved to Singapore. for and it's best known for its manis agent. This took flight last year.

People were obsessed with it. It was essentially an englomeration of other AI tools kind of duct tape together and they turned it into a really cool product. It's now aimed today, Jason, at SMB. So if you're an SMB that wants kind of a gentic AI automation, well, this is for you. It's grown to $100 million in ARR and a run rate of $125 million and was thinking about raising more capital.

Meta took it off the chessboard, $2.5 billion, inclusive of, a $500 million employee retention pool, if you will. But the real thing about this that I think is interesting is it shows that there is a way to get companies out of China and put into a new package essentially that allows them to both accept investment from global investors, including the U.S., and also to find a way to sell themselves. This is super notable because benchmark had invested $75 million in the company. They got a lot of pushback because, wait, you can't invest in Chinese companies anymore. We have all kinds of regulations around this and it just feels un-American to invest in China now with this great robbery going on. Company moved to Singapore, you know, but then again,

TikTok's in Singapore. So this really doesn't mean anything. But again, to your point, if American companies can extract a great company from China or half from China or originating in China, you know, listen, we can get into the semantics of this. There is something very notable here that Zuckerberg was able to extract this company, get them into the democracy bucket. So we need to be thoughtful as a country about winning the AI race. We've had this conversation a million times. And if we can have some of these models and companies, the founders want a base in Singapore, you know, or get out of Hong Kong. You know, it used to be Hong Kong is where you would put these things. But now it's Singapore. Great. And this is why we launched Found a University in the Middle East, specifically in Riyadh and Saudi Arabia and also in Japan, is because we think there's going to be a lot of international M&A. There's going to be a lot of international companies. Founders from around the world are choosing because they can't get into America to domicile in the Middle East, to domicile in Canada and to domicile in Japan, because those are all really dope places to live.

and talent wants to be there, and they're very fluid with their immigration laws, as opposed to the United States, which is, you know, we basically lock this place off as pretty rigid. Okay. So congratulations there. What's next? One other thing that really kind of caught my eye speaking about companies from China that are doing interesting things in the capital markets, Moonshot, the company behind the Kimmy family of AI models. People might know Kimmy K2 Thinking, a very popular model at the moment, raised $500 million at a $4.3 billion valuation, including capital from Alibaba and T. Tencent. And I bring this up, Jason, because we're seeing a number of Chinese tigers, their AI labs that are challenging Anthropic and Open AI go public. And so we recently got data from z.a.i or zipu AI, which builds the GLM family of AI models. They filed to go public. And Minimax, another of the AI Tigers from China, has also filed to go public. So we're seeing a lot of capital availability and some IPOs. But it just seems like every time I read one of these announcements from China, it's missing a zero. Like $500 million for moonshot, fantastic. But in the

US wouldn't that have been five billion? And so I'm not sure if they're doing more with less and therefore they're going to win or if they lack resources. It just seemed like this, because it's an open source model, the business opportunity will be different than a proprietary close source model. Right. And so Kimmy is open source. And that means any startup can take it, fork it, do whatever they want with it. And Moonshot gets nothing. They have to make money off people using, they, are APIs on their hosted services. So much like the WordPress project or MongoDB, I guess, is also open source. You know, when you have an open source project, you're hoping to capture 10% or 20% of the value, not 100%. Whereas chat GPT is a Gemini, Grock, and Claude, because they're

not open source, they just capture 100% of the value. So that's the difference in the valuations. Do you want to see the Minimax income statement from their IPO filing? I know you love these things. So, yeah, what do you glean from it? Thank you for humoring me, Jason. I appreciate that. All right. So this is from the translation of their IPO filing in Hong Kong into both U.S.

dollars and English. And what struck me here is, well, they were gross margin negative in 2023. So first of all, start there. But also, the numbers are small. So this is Minimax. They're doing, you know, people use their models all the time. they had revenue of just $54, $53 million through Q3 of this year, which just isn't that much. And so I think your point about not capturing all the value is super pertinent here.

And maybe that's why the evaluations are lower and the fundraisers are smaller. But it just surprised me that we're only talking about. And it was a $4 billion evaluation, you said? For Moonshot, yes. They raise private. This is for Minimax who's going public. We don't know their IPO evaluation yet, but it will be single-digit millions. So billions, I'm sorry.

your AI is only as good as the data it's learning from every huge leap we're seeing an AI development is based on refining better data sets and guess who came up with the ideal solution for your company that's right my pal's at Uber did you all know as an early investor in Uber maybe you heard third or fourth and I never talk about it but it's true and now Uber AI Solutions works with enterprises all over the world helping them source label evaluate and scale real world high quality data for every industry. When you think about it, this makes total sense. No company understands how to maintain quality while scaling exponentially like Uber. Their own ability to refine and process data sets is why they currently power millions of trips per hour. Now they're bringing that insane level of

insight and expertise to your startup or enterprise. It's pretty exciting. Book a demo today by going to uber.com slash twist. That's uber.com slash twist. Yeah, so they're probably going to to go for 40 times revenue or something like that. So there you go. All right, Jan Lacoon versus Alexander Wang. Jason, you mentioned scale AI earlier being quasi purchased by Meta as part of that deal. They got then scale CEO Alexander Wang to move over and kind of take over the super intelligence team over at Meta. And as we discussed earlier on the show last year at some point in time, that meant that Jan Lecun would report into Mr. Wang. Now, Yon Lecun is a attorney award winner, kind of one of the godfathers of AI, very well known, critical of LLNs. He is a lot of. He is

is leaving to go, found his own company. But in an interview he did with The Financial Times, there was quite a lot of interesting tippets. I wanted to bring those to everyone out there. Do you recall the launch of Lama 4 from Meta, which was kind of a flop, if you will? Well, Yon went ahead and admitted that the results were fudged a little bit. So there we go.

Confirmation. Well, I think they did like a fat, like the different variations of Lama 4. And then they picked the ones that had the best benchmarks. And so they weren't really being up front with it. But as a result of that fiasco, essentially he said that Mark Zuckerberg was kind of over their existing team, just kind of sidelined them. So that does kind of put into context of the change with Alexander Wang and so forth.

And the rest of this that I wanted to bring up is just that he's very LLM negative in a way that I was even surprised coming from, from Yon. Well, yeah. I mean, the idea that guess the next word will allow you to understand the complexity of planet earth of the human brain for him as a researcher who's 65 years old who's like top two or three in the world at this who's been at it for decades who created that category the fact that you had him report into a 20-something who does the training data now those training things are important but that would be the equivalent of putting like the michelin starred celebrated chef who defined

the category, Thomas Keller, and you put him in charge of, you know, the pastry chef is now their boss, right? Or the matri-D is their boss. Like, okay, like the matri-D or the pastry chef are important. They're obviously components of this. You know, somebody might even say, like, the dishwasher. I wouldn't say, like, training data is, like, that low on the totem pole. But there was obviously, like, a lot of disrespect here he got. And, you know, you have to wonder what Zuckerberg is thinking here to lose somebody of this caliber by doing something so divisive as having the 65-year-old research god report into the 27-year-old, you know, with a couple of years' experience doing this training data.

That's no dig to Wang, but, you know, he's a bit of a live wire, according to reports. He's a bit precocious maybe. So you just got to be thoughtful here. Maybe a team of rivals would be what I would have done. I would have said, you know, hey, we're going to have two groups competing to get these problems solved. But I would not have done it this way. Probably Zuckerberg regrets it. But, you know, management.

Who knows? Maybe Zuckerberg just thought, I got to get this guy out of the company. So how do I get him to quit? I'll have him report into the 27-year-old. Either he starts shipping product faster or he just gets a. out and he quits and I don't have to pay, you know, some seven something. So there's all kinds of like game of throne stuff that happens at this level. But AMI is going to be a company to watch because they're going straight to world building.

And that's going to be the discussion a lot of people are going to be having with self-driving cars, with robotics, with just trying to solve and provide more value. you know, like even if you want a salesperson to be an agentic salesperson, if they can understand the entirety of the world they're operating in, well, that's going to be, you know, sort of the next phase. And guess the next word and the literary nature of human existence is going to be complemented with the physicality and the dexterity of human existence. So if you're Andrew, you're making, you know, bombs and, you know, weapons technology. And, you know, uh, weapons technology. do you need to guess the next word?

If you're optimist, do you need to guess the next word? In some cases, perhaps. But you also primarily need to understand, like, okay, this drone is flying over this set of, I don't know, the gondola behind me. And, you know, the gondolas are, you know, in Zurich or whatever or in China. Like, you don't want to hit the gondola or the cable because it's going to kill the people on the gondola, which has happened, by the way. I don't know if you ever heard that story. One of the craziest stories in the history of gondolas, these two American fighter pilots were deciding they would, you know,

have a little bit of fun and fly under the gondolas. One of them clipped the cable and killed everybody. And they, and this happened. They were flying fighter jets in Europe somewhere. And they just decided to make a really bad decision. That's a career. Oh, yeah, no, I think these guys went to jail. Yeah.

Since we're here, let's look it up and we'll pull up the page. Ask producer Claude about this instance that's in my memory somewhere, but producer Claude will find much faster. And we'll just throw producer Claude up on the screen. The reason this is important as an example is like this is an edge case in the world. This thin little cable flying between mountains in Zurich or Germany, Switzerland, Italy, the Dolomites. Why on earth? Or, you know, how on earth does somebody account for that? There's only, you know, there's less than a thousand gondels in the world.

And this is where world building is going to, you know, as opposed to guess the next word or let's try to understand the question better. What are the subsequent questions? Let's try to do deep research. All that is fantastic when we're on our laptops. Pells in comparison to understanding the nuances of the real world. Go ahead, show it. Let's go ahead and talk about this. So we're talking about the Cavalese cable car crash.

This is back in February 3rd of 1998. near the Italian town, Cavalese. And what happened? Well, a U.S. Marine Corps, EA 6B prowler aircraft, which I think I can summon my head, did cut a cable in a gondola, killing 20 people.

The gondola fell more than 250 feet. Seven Germans, five Belgians, three Italians, two poles, two Austrians, and one Dutch person along with the operator. What happened to the people who did this? Well, the pilot, Captain Richard Ashby, was on what was supposed to be a low-altitude training mission. Fair enough.

and then he violated regulations by flying at 621 miles per hour above the limit of 517 struck the cable 365 feet above the ground supposed to be 1,000 feet up. Okay, that's what I was expecting. He was not supposed to be flying that low. That was already breaking protocol. Can you go to jail? Let's see.

He was tried involuntary manslaughter, negligent homicide, convicted of obstruction of justice as well. And they were dismissed from the Marine Corps, the acquittal string. They didn't go to jail, though. Wow. They should have gone to jail. It's amazing. Like you have these weird, this is one of the incredible things about human memory.

Like I have this memory of this, but I, you know, and I have different pieces of it. But if I had actually had to go research that prior to the internet existing, it would be impossible. Like, how would you find that out? You'd have to get on the phone and call sources or go look at like philo facts or whatever that was called back in the day. I didn't think. And the internet comes out. There's a Wikipedia page. I just go to Wikipedia. I find it or Google search. And now, with LLMs, you just summarize it, bang. And this is this augmented human intelligence. What I really want as a product is something I can add to my desktop. If somebody could make this, I would be thrilled. I want to be able to add to my desktop computer, the ability to listen in on everything I'm doing, not recorded and transcribe it. But just as I mention things, give me little one-liners. Oh, that was the Zurich instance here. Would you like more? That was this. Would you like more? And when I get my heads-up glasses, that would be the next unlock is to have your heads-up glasses showing you that. And that's that augmenting human intelligence that for some reason, somebody hasn't created that. If that exists in the world, producer Oliver, who's producing this week in AI, which is well underway here. We're going to be launching it formally, I think, in February, second in the string of new shows that we will be producing here in 2026. Yeah, anyway, I think that's the software I want.

If anybody has an idea about that, just email Jason at calicanus.com. I'm a big enough deal now that I can afford to hire my own admin team. Look at me. They handle all the details of running my company. But if you're a startup, you need to spend your time obsessing about your product, not filling out paperwork and doing all these books. That's why I want to tell you about every. They've worked with over 1,000 startups from first-time founders to VC-funded teams and more.

And they've got the experience to help your company navigate whatever's coming around the bend. Maybe it's time to incorporate your startup so investors take you seriously. Every's going to take care of those filings for free without any unnecessary delays or legal fees. Hey, maybe you just got some new funding and it's time for you to scale up. With every, you'll get 3% cash back on every dollar you spend on your company's card. Hey, maybe it's time to say goodbye to a member of your team. It happens. Well, they've got an employee offboarding checklist and you know I love a checklist. It's going to help you ensure a smooth transition while protecting your business, both legally and financially. Plus, you'll get a $2,000 bonus when you move $250K or more into your every account.

So for your incorporation, banking, payroll, benefits, accounting, taxes, and other back office administrative needs, visit every.io. That's eV-E-R-Y. Do you remember Windows Vista from way back in the day? Which one? Windows Vista. They had these things called widgets. They would go over here and they were kind of transparent, kind of like liquid gloss today in iOS. If there was like a liquid glass version of this that did what you said. I would absolutely use that day and day out, because it would just think for me and I could just grab it when I wanted it to. And so it would be like give me a third arm, but never when I didn't need it, which would be ideal. I want to invest in Jan Lacoon's company. Can somebody on my team reach out to him and ask him to come on the program?

I'd like to interview him. And then I want to invest in this company because I think there's an opportunity here. I'd like to place a little bit. Let's keep moving. All right. Let's talk about prediction markets. Oh, speaking of placing a bet. Speaking of a bet, nothing happened over the weekend whatsoever, and there was nothing to talk about, and there was nothing to wager on over on Pauly Market. No, I'm kidding. Traders on Pauly Market seemed to make an early, accurate, and incredibly lucrative wager on the capture of Venezuela's Maduro, and this picked up coverage from Axios the Journal. People started to bid on this happening Friday night, hours and hours before it actually occurred. And one trader,

because this is all on the blockchain, turned 30K into a net profit of 440K, which was, I think, a very impressive 10, 11x bagger. No, no, I'm sorry, way more than that, like 15X in just a couple of hours. So this has led to people once again saying, hey, maybe there's a lot of insider information going on to trading on polymarket and Kauci and so forth. We talked about this. Which is why they exist, is that people can make smart bets. I'm getting to that.

But you remember the Google search end of year thing that led to someone making a million dollar bet and making a lot of money on that. We talked about it a month or two ago. Well, now people are trying to set some rules into place. Okay. Explain the rules. Representative Richie Torres, a Democrat from New York, has introduced the Public Integrity in Financial

Prediction Markets Act of 2006. Okay. I read it. It's two pages. And what it does is pretty simple. If you work for the government, if you are a political appointee, if you are a member of Congress, you are not allowed to trade on insider information. And I'm torn about this because on one hand, Seems fair. On the other hand, well, then what are we doing? So I kind of want to bring this to you as a, what's the Jason smart path?

Well, these things are already exist. You know, if you work for a company, you're not allowed to do things like those expert networks where an expert network is where you hire somebody for $5,000 an hour. They say this person works in search. It's like, okay, so they work at Bing or Google or duck, go. And then you can hire them for $5,000 an hour. just talk to them and asking questions. Those, that's not illegal, but it might be against the employment agreement of that person. So it is on these companies and these individuals to make sure that they don't trade on and start information. Just like if you were on a basketball team and you knew LeBron James sprained his ankle in practice and you placed a bet on, you know, one of these prop betting sites, they're educating people in the NBA now, don't do that because there were people under investigation.

remember that DOJ investigation earlier this year. So this is really just more about reminding people of their fiduciary duty and their employment agreements than an issue with the prediction markets. Now, a clever person might also see the buildup of ships around a certain country and say, why are we building up those ships? They might hear the rhetoric of people talking. Now, if you don't work for the government and you happen to be in a conversation and somebody who's adjacent, to the administration or the basketball team, and they say, yeah, you know, I think that there's something going on here, my intuition and all this stuff adds up. I'm going to place a bet. I'm going to place a wager. That's kind of what we want in these markets. We don't want the people who are the

primaries doing it. Get used to it, folks. This is how the world has always worked. The world has always worked this way. People were making trades on inside information or adjacent information. I'm going to just coin the term adjacent information or cocktail party discussions. People are always making trades based on that, not illegal, but the person who's, you know, inside of Google and makes a bet like that or is inside an administration that makes a bet like that, they do need to be reminded. It's fine for it to be legislated. But, you know, use your brains, folks. And the one I'm interested in is the IPOs. I love polymarket. Polymark's amazing.

We spend a lot of time in Polymarket. We're big fans. I found a great market for us. We have a great partnership with them too. What's this one about IPOs? There's a lot of volume on this. This is going to become a big, I think, market. And I would like to, I'm online to be part of Polymarket.

IPOs before 2027. This is interesting. Yes. So I wanted to find a good forward looking polymarket to kick off the year. And I love this because it doesn't just rank three companies that had a whole bunch of names with people setting different odds about who's going to list. So, Jason, what I thought we'd do is play a little game. I'll tell you the name of the company and what odds polymarky

gives it for an IPO this year. And you tell me if it's too high or too low. Okay, sure. All right. Cerebrus, the kind of GROC competitor, 80% chance of an IPO this year. Your take. Yeah, that's accurate. I say that's appropriately priced.

Can I say that? I think that's very fair, yes. I think that one's appropriately priced. Inside your house. So if you want it to decline in a comment, I understand. But 74% chance according to polymarket. that Space X goes out this year. Not a SpaceX holder, except maybe through a venture firm I'm in.

And I would say I have no inside information. I think that that is, I would take the, I think that's under. I would say it's a 90% chance. So I think there's some money to be made there. I agree with that. Discord, 50% chance and declining slightly. I would also take that that's under price. I think Discord is going to go public.

I think it's probably a 60% chance. So it's probably a little bit of money to me there. All right. 50% chance for Anthropic on its way up. That's totally mispriced. I would say it's 80% chance. Anthropic will go out. All right.

Let's do one or two more here just for fun. Andrew, 49% chance. From my pal. From my palmer lucky. You're a favorite man. Yeah. I think that is also.

I think that's a 60, 70% chance of it going public. Yeah. Because he said it. He said it over and over again. He wants to go public. He's got a ton of contracts. You know, very, he's in very tightly with this administration.

they're going to keep dropping contracts. So yeah, of course. I think it's 70% chance, yeah. Let's grab a fun one. Let's see. How about 12% chance for RAMP to go out this year? One in eight. I would say, hmm, hmm.

Yeah, I don't think they're going out this year. So I think it's probably appropriately priced. I do think Open AI is the one to look at here. Open AI. Will they go out this year? will they go out this year? Hmm. I think they have to.

Because they need the money? Yeah. I think that the funding, they, the Masayoshi-san, $40 billion got put in, right? He pulled that off. So what that says to me is

they do have a decent amount of cash, but not enough to hit their targets. Revenue's booming. I think it's more like a 50% chance they go out. I would say it's 100% chance they go out by 20, by the end of 2027, so before 20, 28. So I think Open AI is probably 50% chance. But this is a great little game here. And this is, you know, one I would like to get involved in. So I hope that I become, uh, approved. Because I'm going

to just start making, I'm going to start making these trades. I'm going to just start making these trades on the show. I'm going to put like a hundred grand in my thing. And we'll just keep track of my ability to predict. And we can even chart into a micro website. It'd be a lot of fun. Absolutely. I'll be J-trading on Polly Market. There's, um, something going to on with my guy, Biz Stone. I've known Bis Stone for 15, 20 years now. He's brilliant. Great product guy. What's he up to? Well, are you familiar with the supergroup in music? Yeah, supergroup. Yeah, that's when you, like, traveling willberries would be my favorite supergroup or Mark Knopfler did the nodding hillbillies. It's when four or five really famous names form a group for a limited

run to do an album or a tour. Yeah, essentially you take some of the greats, this guitar player from that band, that bass player from that band, you put them together and see what they come up with, and fans will often show up just because they, oh, I already know that guy. So, Bistone, formerly a co-founder of Twitter, and Evan Sharp, one of the guys behind Pinterest, had teamed up to build a company called West Coe with a dot. And I mentioned super good because this is kind of what this feels like to me in product sense. It's a bunch of people who've done things, who are well-known, who can raise capital and know how to build, build teams and so forth, and they're trying something new. So they are trying to

built, and they've kind of just started to talk about this, a different form of social media. Their first app is called Tangle. It's invite only, Jason, and is designed to help get people to, like, follow through with their intentions. I have pulled a screenshot from the app store here for you, and the idea is kind of like an anti-social media. So instead of scrolling, instead of swiping, you're imagining, growing, reflecting, supporting, and learning, I guess.

Yet another news app. I've been here, folks. None of these news apps have ever worked. I started one inside. Smarter news in Japan did work. And obviously, people get their news from Twitter. And Ev was a co-founder of Twitter famously. He's really smart. I like people taking a swing at a category that's never worked. Pulse, you remember that one. I was an investor in Circa. That was very elegantly done. Beautiful app. You know, the problem is people like getting their news in their Facebook, Instagram, TikTok, and X, formerly known as Twitter feeds, and going news first may or may not work. But somebody might figure it out. And there is an issue right now that news is broken or there's a perception that news is broken. When something breaks, there's an opportunity. I'll give the number one example of it.

Investigative journalism is super important in the world. It's super expensive. it comes with high risk and it comes with high reward 60 minutes front line i was thinking about buying front line but it can't buy it it's part of like a non-profit kind of situation but i've always in another life wanted to run an investigative journalism group it's incredibly expensive the business model doesn't really work but we saw this past weekend with you know fox nine and we talked about it on all and we had the person who's been doing these investigative journalism stuff on the ground in Minnesota, knocking on doors of all this fraud occurring in Minnesota.

This is a 15-year story almost, and these stories take multiple decades. There is no business model for it, except if you're smart and you're a solo running gunner and you start trending on YouTube, and YouTube now allows you to monetize these, The monetization program previously left this stuff out. Now my understanding is it includes it. I think X includes it. Obviously, substack is a great way to do it. There's an amazing opportunity to investigative journalism.

I would love to talk to somebody in the investigative journalism space about some startups to do this, crowdsourced, which was supposed to do anything, data-driven. There's so many opportunities here. When things break and there is chaos, that is the ultimate opportunity for a startup to insert themselves. And most will fail, but one will figure it out. And I think investigative journalism, not generic. So this is where a tangle, I'm like, you know, is it enough for this to, is there enough here for people to pay for it or advertisers?

I don't know. But I do know there's something in the need for investigative journalism to exist in the world. And I hope that more people try it. Well, we used to have classified ads that subsidized all sorts of things, you know. that went away. We had online ads. That kind of worked out. Subscriptions have worked out for like three publications. I think individuals, you're probably right, will have more impact, but they don't have the structure. Legal teams, backup, editing. It's just, philanthropy only gets this so far. It's tough. So the key here then is to take what we saw. What's the individual's name who was on All-In?

Gosh, I'm sorry. Nick Shirley. Nick Shirley, who I don't, did you see the interview on All-in by No, my chance, Alex. No, I was children. Oh, good. You know, I interviewed him and I was asking him questions about his process, et cetera. There's no process. He's got an LLC, but he doesn't have apparently errors in omission insurance. He doesn't have fact checking or these kind of things. It's gonzo run and gun. That has a place in the world, but it's hard to scale, hard to get investors for. If you had investors, they would insist on errors in our mission. You'd have to slow down the process and add, you know, all those features to, which I believe he will do over time, out of necessity. Because if you knock on a door and let's say you get something wrong, well, people

will file lawsuits. And you can just look at his 60 Minutes has been in front line. They've faced all kinds of lawsuits and threats of lawsuits for everything from the Catholic Church scandal highlighted in the film Spotlight by the Boston Globes investigative team 60 Minutes versus the tobacco companies. The movie The Insider came out of that. There's so much important work to be done here, but you do need to add that infrastructure. All that infrastructure then could become a package like Substack does. So what if Substack added errors in emission insurance to their offering? Wow, that would be interesting.

And then you could, you know, pay 10% of your revenue to them for the platform. But if you want to turn on investigative journalism and have errors in emission, you've got to pay an extra 10% more. Yeah, 10% more, whatever it is to cover you. All of this is just business opportunity. So for entrepreneurs out there, when you see something going away in society, but people still respond to it, that means opportunity. So get curious. And I would love to have these two gentlemen on.

So producers, get to work. Let's get my friend Biz on the program. And let's have him tell us and show us the new product. All right, Jason, I want to do a combination founder queue and idea for you because I found something really interesting online. So there was a great question over on the startup subreddit. We're not going to talk about it in particular. But it was all about what does it cost to get legal representation for your startup? When do you need it?

And people were talking back and forth about that, in the back of my head. Then Fred Wilson did a very interesting post about how he was dealing with the closing of a startup investment. Normally, he would take the draft agreement, take it to his lawyers and they would do a lot of diligence on it. And he said he did that. And in this case, the lawyers wanted $50,000. To do a series A, that would be typical. Series A. That's typical. Okay.

So what he did instead was he took all of their prior deals they'd done. put them into one notebook on notebook LM, and then he took all the information from the startup and put that, essentially their legal data room, into a different notebook, and then interrogated them to find what was different about this contract, and also what could he learn about the startup's legal setup? A couple hours of his time,

ended up with one question he couldn't answer, which he then went out to get legal help with, but it seems that he collapsed $50,000 worth of legal work into essentially a couple of queries and a free-to-use tool if you already have G-Spoon. So, first of all, are we, underestimating the amount of work that's going to get collapsed by AI because, holy crap. But two, for founders out there who don't want to pay insane amounts of legal expenses,

does this sort of like involved intelligent approach using modern tools provide a pathway to spend less money on things that are non-product critical? Yeah, this is already happening. I have met a number of founders in our accelerator, launch accelerator, launch.com, slash apply if you want to come and spend 12 weeks with me and get 125K. or Founder University. We had a company that had no lawyer, no accountant. They were doing it all themselves.

And they had gotten to hundreds of thousands of revenue. And I said, hey, I know you don't have an attorney, but this will become a blocker for future rounds. If you're going to get to the Series A, the seed round, you start raising over a million dollars, like doing it on convertible notes. I'm like, no, no, I can do it all myself. I can do it with an LLM.

So founders find a way. And so there is a product opportunity here. And that product opportunity is going to exist inside, of Fenwick, Wilson Sonsini, Oric, where attorneys will use these tools, and these tools will make them go faster, which then should result in them lowering their hourly for doing these things. So that'll happen from the supply side, the lawyers. And then on the demand side, the startups, they're doing it on their own already. There's always more legal work to be done. There's going to be 10 times as many startups as young people can't get jobs. That's why

I'm all in on Founder University and I'm going to Japan on Friday after my CES. quick 48-hour, 36-hour, you know, jaunt that I'm going to right after we've recorded this episode. Just look at convertible notes. Convertible notes and safes, these were things that founders started doing on their own. The law firms were like, yeah, it's totally fine. Come to us after you have 10 notes stacked. We'll clean it all up. You know, manage your own cap table using all these different cap table tools. Right. Boom. Founders are doing it. So all it's going to do is allow more startups, just like cloud computing did. So if you look at this like cloud computing, instead of setting up your own servers and having your own rack somewhere co-located,

that goes away. You save each startup 250K. That happened 10 years ago. Now, instead of 50K, it goes down to 5K, all good. There'll be more startups to service. So that's what the AI boom is showing me over time is if you can't get a job at Microsoft Google and you're graduating from school, you're going to need to be resilient. So not to make this political, but Mondiott.

Tommy made this comment, like rugged individualism is going to move to collectivism, i.e. socialism. That's like one trend. Sure, the government sees his assets. They provide everything for you in your life. If you want that, move to New York City. That experiment's going on there. And if you want the rugged individualism, come join me in Texas. And you can be a rugged individualist. You're going to need to be a rugged individualist as a young person graduating. And this is part of that trend. And Fred Wilson is awesome and like a legendary VC. If he's on this, and here's his AVC.XYZ, here's his blog post that he posted here. Yeah, go read it. We'll put it in there.

Google Notebook LM is an incredible product that most people don't know about. That's getting better and better. We had Stephen Berlin Johnson on this program. We'll link to that in the notes. We should have Stephen Berlin Johnson back on producer Marcus, maybe, or actually producer Oliver. That would make a really good person to have on the AI roundtable. We're going to be doing weekly with this week in AI. Stephen Berlin Johnson, Brilliant Mind, and friend of mine from the 90s and Freds from the 90s.

Yeah, they did a feed. They did the feed zine. Back in the day, there was a zine called Word, and there was a zine called feed. There was a zine called salon. This was like the zine movement online, predated blogs, just meant like a news feed. And this is the future of it. So awesome. Legal due diligence.

We're using notebook L.M for 10. 10 different projects here at our firm. Great product. Should Google spin that out, Jason? Should they spin out notebook LM into a company? Like, I feel like it's, or just, it's so popular. And people just, everyone talks about the same tone that you just did. I hope it doesn't get lost inside a big company for being too niche. I like the idea that it's part of the Google suite of products. I like the idea that, you know, you upload a bunch of information. The number of sources, I think you can only put in 300 items right now. They should have an uncapped version. So, for example, I want to have all of this week in startups in there.

I had my Athena-a-assistance. Go to Athena-wows.com to get an assistant. I had my Athena assistants take 300 episodes, the first 300 of this week in startups and the last 300. So when we need to search, hey, did Alex and Jason debate, you know, Apple AirPods or, you know, the pin or, you know, what was that pin called, that failed, the AI pin?

The friend. The friend was a pen. No, there was the other one that you, the Apple people did that you clipped to your shirt. You would put your hand there and project onto it. Anyway, all of these AI pendants and, you know, third device that will go with your, you know, laptop and your iPhone. How could we ever figure out when we talked about that and what we said?

The way we do that is we go into the Google notebook and we have to have eight of them because we have 300, we're at 2,300 episodes or so over this podcast. It's just hard. You know, Leo wants to go back to this week in tech and he's at whatever number of episodes over 20 years. Like, how do you even find those discussions? It's too hard. you'd have to put a producer on it.

Now a producer can find that in 15 minutes or five minutes instead of five hours. And so all of this is to say, great product. And yeah, I actually told Sergey when I saw him, by the way, this is a sleeper product. You've got to really give this team more resources. I think it's like the future. Double, triple, quadruple their compute allocation, make everyone happy. Take off that limit. Yeah.

And notebooklm.com, you know, the thing that Google has always strong. with is when to have it have its own domain name and be its own brand or to bundle it. Google Plus, which was their social network, would have done fantastically if they had ripped it out of Google and just called it plus.com. And it would still exist today. But because they made it Google Plus and they tried to goose the number of users by making a part of the suite, nobody ever knew where to find it. Nobody knows where to find Google Notebook. That's why it should be Notebooklm.com, stand alone, let it work, let it cook, right? That's why I made the syndicate.com and took the time to get that IP and make that domain name

so I can have that as a brand. That's why I took Founder University and made Founder. Dot University. If you really care about a product, want it to stand alone, you got to put a GM or a CEO in charge of it, and you've got to do what they did with YouTube.com, which is let it live on its own domain name with its own brand. This product is good enough to do that. That's my best advice to my friends at Google.

Amen to that. All right. Another amazing episode of this week. week in startups, Mondays, markets and startups, Fridays, Lon Harris will be joining us. Wednesdays, I think it's going to be the investment roundtable. This week in AI, we got lots of plans with you for 2026 here on twist. Call it twist, please. Like, you know, put a little twist in your beverage, or take a twist on the wild side with us here. He's Alex Wilhelm. I'm Jason Calacanus. Cautionism.com. Is that your domain or cautiousoptimism.com? What do you got?

Cautiontumnsum.com, if you will. At news. Okay, go to cautious optimism. Dot news and throw Alex a hundee. Fund your local journalism. Find your local journalists. Okay, we'll see you next time on Wednesday on this week's startups. Bye bye.