Heat. Heat. N. Heat. Heat. Heat. Heat. N. Heat. Heat. Heat. Heat. N. Heat. Heat. N. Hey. Hey. Hey. We came to this world to reach the stars. We came to this world to see a future. We came to this world to reach the stars to shape our future. You feel the new heat.
I see multiple journalists on the horizon. Standby ready for deploy. Friendly reindeer in Boom.
Well done, brother. I'll see you in the new year.
You're watching TVPN. Today is Thursday, December 18th, 2025. We are live from the TBPN Ultra Dome, the temple of technology, the fortress of fin.com. Count is money, save both, easy corporate cards, bill pay, accounting, a whole lot more all in one place. We got a video here that has been promised to me. I saved this for the show. To this reaction. I haven't seen the video, but I've seen the metrics around the video. It's got 15,000 likes just on a quote tweet of it. I think it's going to be very funny. Tyler was saying, 'You got to watch this.' So, we're going to pull this up. Pegasus is quote tweeting. It says, 'This is OpenAI showing the public its financials once it goes public.' Okay, looks pretty good. Getting a haircut. Looks good. Oh, no. You know, I've been watching these haircut videos and they're actually incredibly good content because I saw one. I mean, obviously that one's very silly, but
Dude, this video is going viral. This guy's going to be full Giga Chad mode within 6 months. I guarantee it.
Guarantee it. Guarantee it. But he just did that. He knew what was going to happen. He did it to inspire himself, I guess, to look. For sure. For sure.
To start smashing. I looked at a haircut video where it's incredibly sticky content because you're watching the guy describe what he wants and then at the very end they show you a montage of the photos and when it works out obviously the joke is that it's a downgrade. But when it works out it's really satisfying. The payoff comes right at the end which is great for retention and average view duration. So it's a growing format. It does make me wonder, can every different vertical, every different type of content become high retention or are there some things that are just more naturally payoff based?
Anyway, let me tell you about Adio, the AI native CRM. Adio builds, scales and grows your company to the next level. We are in Arena Mag. This is breaking news or the latest edition of Arena. We had three martini lunch. This is issue number six, Q4 2025. Three martini lunch and we are in here with a wonderful spread. Oh, there's a bunch of great people in here. Look at this. We got Daniel in here. I don't know if he's doxed in this photo actually.
I think so. But there's a profile on some archer stuff, some V tall stuff. Very nice. We got California Forever in here. Apparently, we are having a YouTube issue, folks. You'll have to refresh the app if you want to avoid any type of lag. Okay. Oh, they did a profile on or a review of Dwaresh Patel's 'The Scaling Era'. What is going on here? The Christmas spirit. That camera had one too many eggnogs. Tyler also has 'The Scaling Era', an oral history of AI 2019 to 2025 by Doresh Patel. There's a whole review of the book. And then, of course, you get the beautiful TBP spread. Look at this.
Isn't that great? I like that they kind of brought their own aesthetic on top of some photos that we took here in the Ultra Dome. Very, very fun. They respected the brand.
Yeah, it looks great. And here's a photo of me and Jordy having breakfast, showing you what the day in the life is like. And then here's me reading the Wall Street Journal before the show starts. Life of Kugan.
Kugan Mode. Very fun. I could take you through it, but you've heard the story. And you can go subscribe to Arena Mag. You can read it online. Three Martini Lunch. Interesting. I did want to comment on this. The art direction for the cover art. I like three martini lunch. It feels like a pretty big departure from the previous very collage-based covers. So this is a much more simple design. Still very
At the same time it's still very nostalgic, right?
Yes. I mean look at the back
Yeah. The back says 'The business of America is business.' And it has a pager. That's something we haven't gotten into. A cigarette and an ashtray and a Casio watch which is pretty cool, pretty classic. Anyway, back on the timeline. Public.com, investing for those that take it seriously. Multi-asset investing trusted by millions. What's back on the
Katie Roof, one of the greatest scoop athletes of our time. A future scoop hall of famer has a scoop opening. She really has someone at OpenAI who loves her right now. She's been on a scoop tear. She says, 'Scoop, OpenAI could soon be worth 750 billion in a financing round.' They've had early talks. Let's give it up for early talks. We prefer advanced talk, but early talks.
All right, you got to start somewhere.
About raising tens of billions of dollars or even a hundred billion. Company was last valued at 500 billion a few months ago. Tay Kim chimes in, says public markets are crushing OpenAI exposed stocks while private investors with visibility into OpenAI's metrics and internal numbers are piling in essentially.
I trust Katie Roof's reporting so he's saying that it's bullish for OpenAI, that's his take from this.
Yeah, maybe some of the public stocks have been oversold. That said, I'll be interested to see how this round comes together. Again, remember, it's a good time to be fundraising if you need tens of billions of dollars.
If Warner Brothers ends up going with Netflix, right? Because that was a huge if you're raising tens of billions of dollars.
Yeah. And you're not doing it from strategics like the hyperscalers. It pretty much has to be sovereign wealth funds.
Yeah. Right. And potentially if they pass on Warner Brothers, they have loose change in the couch cushions to toss OpenAI's way to the tune of $10 billion checks potentially. But I mean it does feel like OpenAI is at a bit of a trough of disillusionment potentially going into a plateau of productivity. It feels like they're turning it around. It feels like the new models and the products
I haven't seen many people say I love 5.2 too.
But then again, outside in the real world, clearly, ask people what AI app they're using. That'll give you a lot of signal. I thought this
It just feels like, okay, maybe we're not in Baja Blast territory, but we know that every code red is followed by an equal
Adult mode is coming in January.
Yeah, maybe. But it does seem like a lot of the negativity has kind of worked its way through. The market's traded down. It feels like we're finding a bottom. Like we're finding the footing of, okay, reset the narratives, rebuild back up. Can they catch up in this? Can they catch up in that? If you see a couple good charts, couple growth charts, then you're sort of back in business because
And again, people got so negative. Like the point I'm making is a couple months ago it was like it's Enron, it's going to blow up. It's zero. It's not going to work at all. And it does feel like we're turning the corner on that.
Maybe muted the people that are still saying that because they're still saying it.
They're still saying it. Well, it just got old.
This was a funny post. They say Sam says, 'Sell me this pen.' Jensen says, 'I'll lend you the money to buy it.'
Of course, Jensen is investing money.
Yeah. But anyway, a very good Linear. Meet the system for modern software development. Linear streamlines work across the entire development cycle from roadmap to release. Warner CEO David Zaslav and Netflix co-CEOs Greg Peters and Ted Sarandos on the Warner Brothers lot. They took a game day photo. They went out deliberately to just strut and just flex and they clearly knew what they were doing. Hey. Hey.
Well, this is their new lot.
You point. You look over here, guys. Can we look over this direction? Okay. Yeah, that's perfect. Snap. Snap. Snap. Okay. Yeah. You know that they know what they're doing. This looks like no hostile offer can get between us. We're bros. We're hanging out. This is a sign of strength.
Sun's out. There's palm trees.
Yeah. This is them sending the signal. Hey, we're excited to work together. We're excited to be in partnership collaborating. There was one person who quoted this and was unhappy about it. Where is
They were saying that they weren't well-dressed. Yeah, they were saying that they weren't well dressed and he was like, I'm kind of in that boat. It's not
They threw on their blazers. Give them something. The Hollywood executive style that's sort of become the norm. Maybe we go back to the three martini lunches, guys. Maybe we throw on suits now that you guys are in business.
But this is also a good way for people that are maybe more on the tech side to start dressing up. You don't have to go from jeans straight into a suit. You can throw on a blazer, right?
Matt Levine is back. Says, 'I miss Christmas Eve, New York, New Year's, and the Fourth of July holidays for newsletters a lot smaller than this. This is as good as it gets for newsletter writers.' And we can jump into the article. He says, 'Warner doesn't trust Paramount. Revocable trust, ESG side letters, IPO lockups, Destiny, and doing deals over the holidays. Trust. If you're a normal person, most of your wealth is probably in your own name. But if you're one of the richest people in the world, you probably have a lot more complicated estate and tax planning, which probably means that a lot of your wealth is in trusts, legal arrangements that your lawyer set up to hold assets for you. Most simply, you might have a revocable trust, where your assets technically belong to the trust, but you have control over the trust assets, investment decisions, beneficiaries, etc. This can be useful for estate planning, but for most purposes, owning assets in a revocable trust is pretty much like owning them yourself. I regularly say that Mark Zuckerberg owns a lot of shares of Meta or that Elon owns a lot of shares of Tesla even though neither statement is exactly true. Their trusts own the shares but for most practical purposes you can think of they own them. This creates a small dumb problem. Let's say you build big yachts and Mark Zuckerberg comes to you and asks to buy a billion dollar yacht. Naturally he will pay you the billion dollars on completion of the yacht. You might ask him, 'Well, do you have a billion dollars?' And he might say, 'No, actually I don't. But the Mark Zuckerberg Trust owns like a bajillion dollars worth of Meta stock and I control that, so I'm good.' You find that persuasive. So you pull up the yacht contract. Who do you put in as the buyer? If you put Mark Zuckerberg, he has no money. If you put the Mark Zuckerberg Trust, it has plenty of money now. But if it is a revocable trust, he can just take all the money or shares out, whatever he wants. It's revocable. If he changes his mind about the yacht, he can clean out the trust. You'll send the bill to the trust, but the trust will have no money and you'll be stuck with the yacht. Not that bad of a situation to be stuck with a billion dollar yacht. But
If you're in the business of selling them, you certainly want to be able to complete the transaction.
I get the point. Honestly, this is a pretty easy problem to solve. You put Mark Zuckerberg in the contract. He doesn't own the assets directly, but he controls the trust. So, if he owes you a billion dollars, you can make him take it out of the trust. But maybe you'll get confused or he'll get confused and you'll end up signing a contract with the trust. Then if he changes his mind, he can zero out the trust and stiff you. I cannot imagine that this comes up a lot with yacht builders or anyone else. Elon Musk once did sign a contract to buy Twitter for 44 billion, then changed his mind and tried to get out of it. The fact that his wealth is mostly in a trust did not at any stage of the proceedings trouble anyone. He had signed his equity commitment in his own name and everyone understood that if he was sued and lost, he'd have to pay with his trust money if necessary. And Twitter did sue him and he did pay. Fine, fine, fine. On the other hand, Paramount is trying to buy Warner Brothers Discovery, breaking up Warner's deal to sell itself to Netflix, Inc. And Paramount does not have nearly enough money to pay for Warner. Instead, much of the money behind the bid comes from Paramount chief executive officer's dad, Larry Ellison, as of this writing the fifth richest person in the world, or rather it comes from his trust. And this would be one of the greatest tricks in the history of mergers and acquisitions. One major sticking point is Warner Brothers' concern about the financing proposed by Paramount which is led by David Ellison. A big part of the equity is backstopped by a trust that manages the wealth of his father Larry Ellison. Because it's a revocable trust, assets can be taken out of it at any time and Warner Brothers may have no recourse if that happens, the people said. And Matt says, 'Aha, sure. The risk here is Warner throws over Netflix, pays it a $2.8 billion breakup fee, and signs a deal to sell itself to Paramount for something like $8.4 billion in cash. For some reason, changing market conditions, regulatory difficulties, and a change of heart by the Ellisons or a change of heart by their co-investors, the Ellisons decide they don't want to close the deal. Quietly, in the comfort of his own home, without saying anything to Warner, Larry Ellison takes all of his stock out of the trust, leaving it empty. Yoink, he whispers to himself. Oops, never mind. Paramount tells Warner. Warner sues Paramount for specific performance, seeking to make it pay $108.4 billion and close the deal. Lol. We don't have that kind of money, says Paramount, which has a market capitalization of about $15 billion and about $3 billion of cash. When we signed the deal, you knew that the money was coming from the Ellison Trust. So, Warner sued the Ellison Trust for specific performance, seeking to make it pay $108 billion and close a deal. Lol. We don't have that kind of money, says the Ellison Trust. We did, sure, but the trust got revoked. Yoink, adds Larry Ellison more loudly this time. So Warner sues Larry Ellison for the money. He says, 'Who me?' says Larry Ellison. You have no deal with me. I don't even know what you're talking about. This is not my problem. This does not strike me as especially likely and it would obviously be bad for Paramount and the Ellison, but I suppose it is possible and it would be a pretty fun trick. I talked about this
A little bit yesterday on the show. This kind of situation. Today, Warner officially rejected Paramount's bid, advising shareholders not to sell their shares in Paramount's tender offer. Here is how Warner describes Larry Ellison's commitment or lack thereof. And this was the exact quote that I read yesterday, which was, 'Despite Paramount's headline claims, there is no equity commitment or backstop from the Ellison family for the offer.' And I'll skip over the rest of that quote, but Matt continues, 'As far as I can tell, and somewhat bafflingly, this is correct. Paramount's own tender offer says that the Ellison equity commitment and guarantee will be from the Lawrence Ellison revocable trust, not from Larry Ellison himself. The offer points out that the trust is rich. The Ellison Trust has financial resources well in excess of what would be required to meet its financial obligations under the equity commitment letter, including many other assets and financial resources available to it. Record and beneficial ownership of approximately 1.16 billion
Shares of Oracle stock with a market value of approximately $252 billion as of the date of this offer to purchase, but as the name says, it's revocable. This seems extremely fixable. Have Larry Ellison sign the commitments personally? Yep.
William Cohan reports that the whole situation is trains passing in the night and quote, 'The Ellisons believe they can still be sued for specific performance to fund the deal.' He quotes a person familiar with their thinking saying that there is no financing condition in the deal and Paramount and the Ellison/Redbird along with our lenders are legally obligated to close regardless of future financial or business performance at Paramount. That is not an impossible thought to convey in merger papers and yet so far they apparently haven't. And then
Back just well I mean the ESG side is a different story I'm sure. I don't think it has anything to do with the okay. Matt Levine writes like multiple little blurbs.
But the real trick is obviously create a revocable trust and name it the irrevocable trust.
One simple trick. EA lawyers hate this one simple trick. Yeah. I am wondering if this is very common. I mean Matt Levine talks about this, but like why did Elon not do this during the Twitter buyout?
That is a good question because it says Elon did sign the contract in his name, which he could have just done to the trust and then he could have backed out when he wanted to.
Well, I mean, one thing is it feels like Elon just rips checks and was very much just like, 'Yeah, I'm good for it. I'm going for it. I don't want to play this game at all. Because I'm all in. You have the full faith and credit of the bank of Elon Musk essentially.'
Yeah. I mean, also, do we even know that this is legally feasible that you can just revoke the trust? Is that just 100% totally fine? Has that happened before?
Yeah, that's a good question. I don't know if it's happened before. But it certainly seems like something that you are going to. We talked about with like that when you make an offer for a big company there's the expected value of the close and saying hey we have this backstop is something that increases the probability that you close, that the financing condition is met. But then saying, 'Okay, we're doing it in a trust or revocable trust,' takes that down a little bit. But again, there's a world where all of the different capital providers lined up and you're not even talking about trust, revocable, or irrevocable because you never get there because all the different funds are saying, 'I want to get in. I have to get my stake. I want 10% of this. Here's $10 billion. I'm ready to go.' And so you don't even care about the backstop or you don't care about the trust because everyone's lining up. The only reason that maybe they're in this scenario is because it feels like Warner Brothers did actually sort of kick the tires on the Paramount deal and get to the end state of evaluating it and saying, 'Oh, well, like we've heard, you mentioned Jared Kushner's in. Is Affinity Partners really in?' And when they pushed them, Affinity Partners seemed like they might be out. I don't know exactly what happened, but there's a variety of stakeholders that came in and the level of it's not just that they were thumbs up, thumbs down. There were some that people said that they were thumbs up, they wound up being thumbs down. Some of them were here and they wound up being there. A little edgy depending on a lot of different folks where it's like putting together any other financing round. Ideally every investor that goes into a seed-stage company or Series A company is like, 'Yes, I'm good regardless of who else is in the round.' Every once in a while you get VCs who say, 'Yeah, I'll sign a term sheet, but term sheets are non-binding, and we'll see how the round pencils out. I'll sit on this signature. I'm not going to sign this. I'm not going to wire.'
Or they would give a verbal thinking that Sequoia is going to come in, Sequoia doesn't, and then they're like, 'Oh,'
Or vice versa. They give the verbal, they sit on it, they're not wiring, they're maybe leaning back, and then Sequoia comes in or Founders Fund or Andreessen or someone comes in and then they're like, 'Wait, wait, wait. You said that I could get 2% of this company for $100K.' Like, well, give me my terms. And you're like, 'You didn't sign that doc. You were very much leaning out.' I've been in that situation. It's never super fun conversation. But yeah, you know it ain't over till the bank.
A little bit more from Matt Levine. He's talking about lockups. Specifically SpaceX, we all know they're trying to go public at a $1.5 trillion valuation. They want to raise $30 billion via the IPO. And Matt says the raise is not that big relative to the valuation. But then he says the problem is not the $30 billion, the problem is the other $1.47 trillion. And he goes on to say what happens if $500 billion of SpaceX stock is unlocked at a single moment and people are like, 'I've been locked up for, I have had limited liquidity for'
Let's say you're an employee who's been there 10, 15 years. There could be a kind of a rush to get liquidity.
And so he goes on to say that in the conversations that Wall Street has been having with SpaceX, Anthropic and OpenAI, there is some conversations that would create a scenario with, I'll just read it specifically. Investment bankers are starting to prepare discussing proposed solution to help win deals. At least two large banks largely ruled out a standard IPO lockup period of either 90 to 180 days and are discussing how to design a staggered lockup release for companies like the ones I mentioned. And then he says if you're a shareholder of a trillion dollar private company you're used to getting limited periodic liquidity events where you can sell shares. Why would that change when the company goes public? So
The dance of taking SpaceX public. It's going to be an interesting year for IPOs. There's a bunch of great companies that are finally mature enough to potentially go out. Let me tell you about Vanta. Automate compliance and security. Vanta's AI powers everything from evidence collection and continuous monitoring to security reviews and vendor risk. There's an article in the Wall Street Journal about a massive IPO. Wall Street gets early taste of hot year expected for IPOs. Shares of a company called Medline began trading in the biggest new stock listing since 2021. Wall Street is getting a glimpse of what could be the biggest year ever for IPOs in the US after four years of choppiness in the market. He didn't even let me get to the number. The medical supply company raised $6.3 billion in its stock offering. The stock closed at $41 a share. That's up 41% from its $29 IPO price. So massive offering. It's the biggest IPO since Rivian actually in 2021. And the ticker is MDLN. 'We're the largest company no one's ever heard of,' said the chief executive Jim Bole. He said he and his team have spent the past year and a half educating the investment community ahead of the stock offering. You got to tell everyone what you do before you ask for their money because if you've been operating behind the scenes, no one's going to know to invest or be excited. So, the offering could help set the stage for some of the most highly anticipated IPOs with rocket maker SpaceX, AI company Anthropic, and mortgage giants Fannie Mae and Freddie Mac among the companies looking at listings in 2026. OpenAI also rumored. There's a couple other companies that might go out.
Well, there's a bunch of tech darlings that look kind of like infants in comparison to
But still might be of maturity to go out.
Very notable like companies that are basically at Figma scale.
Sure. Yeah. Or even beyond. I mean for a long time, if you're a $10 billion company in the tens of billions, the decacorns, any decacorn should be comfortable going out potentially as long as they're not in some crazy R&D cycle and they actually have the core financials humming at this point. Bankers say their teams have been busy in recent weeks pitching companies to be lead adviser on their IPOs, marking a pickup in activity. Many startups had assumed going public in recent years. 'I'm not sure that's entirely true anymore,' said Magdalena Henrik, head of US technology equity capital markets at Bank of America. Bankers are hoping Medline, which is backed by Blackstone, Carlyle, and Hellman & Friedman, helps to close out a positive year for IPOs. Well, speaking of Blackstone, we have a little Christmas present under the tree. Why don't we open it?
What do we have from Blackstone? The good folks over at Blackstone sent us a present and we will open it live on the stream here. So have you heard that the CEO of Blackstone, Schwarzman, is becoming a DJ? Goldman Sachs CEO David Solomon gave him a stamp of approval and they sent over this very cool tape recorder. I don't know if this is a real tape recorder. I don't know if I have this queued up, but you can maybe hear it. Let's see.
They went super viral for this because they made an entire like they basically clowned every startup vibe reel. Oh, you have a cinematic vibe reel.
We had our whole team sing a jingle and filmed it and sent it out on a physical thing. This is very cool. Portable cassette.
Is that made to look like a cassette tape?
I think it's actually a cassette. Yeah. This is crazy.
Wow. I didn't even know we knew how to make cassettes. This is a very good drop. It's hilarious. It also feels like them leaning in because I know that the response to that Blackstone video was not overwhelmingly positive. People were like, 'What is this?' But clearly it's an inside joke and the team finds it funny and they were doing it tongue in cheek and they were having fun with it and they had a lot of 1980s throwback cameos there.
That's great. Anyway, thank you to the good folks over at Blackstone for sending this over. And go listen to the full song and watch the video if you haven't already. Numeral compliance handled, numeral worries about sales tax and VAT compliance so you can focus on growth.
In our newsletter today, we did a little review of the model wars. Of course our newsletter, you can get it at tbp.com. Tech analysis and news. We have a daily op-ed, top headlines, and more. Sign up at tvpn.com. So, what a year. Just in terms of model releases, August 7th, OpenAI releases GPT-5. September 29th, Anthropic releases Claude Sonnet 4.5. Then the day after, OpenAI releases Sora 2. September, the same day, Meta releases the Meta Ray-Bans displays. November 3rd, OpenAI and Amazon announce a partnership valued at $38 billion. November 12th, 5.1 drops from OpenAI. Yep.
November 17th, xAI releases Grok 4.1. November 18th, Google releases Gemini 3. Massive response to that. It's crazy that that was just one month ago. November 20th, Nano Banana Pro comes out. Then November 24th, Anthropic releases Claude Opus 4.5. We talked to Schulte about that. And then December 1st, DeepSeek releases DeepSeek Version 3.2. And of course, before the end of the year, OpenAI had to fire back with GPT 5.2 on December 11th. And then December 17th, Google releases Gemini 3 Flash.
Also, today I think OpenAI released 5.2 codex.
So, the coding model version.
They're starting to stagger them out a little bit more.
Which I think is smart, right? I was wondering about this with Logan, like do you think that we will have a period where we get on annual release cadences and it's just like, 'Oh, the model is like GPT 2026, I'm running the 26 model,' or do you think it'll be random?
We're not quite at the plateau where I think that would make sense. Right now every company is fighting for market share and if you can release tomorrow that's a lot better than releasing
But there is a world where OpenAI takes 5.1 and 5.2 and says, 'You know what, that's GPT-6, that's GPT-7,' right? They could and people would be like, 'Whoa, this is really incremental,' so they use the 0.1 but they also didn't do it's 5.0.1 0.1
I think the reason there's so much pressure to launch is there's pressure to be at the top of the benchmarks. In other product categories it's not like there's this very specific way that you
What are you talking about? In cars, Nürburgring.
Sure, sure. I'm talking in enterprise software, typically it's not like somebody launches a product and then somebody else launches a product and investors can look at it and be like, 'Well okay, I think this one is just objectively better.'
And so I think that given the goal to sit at the top of the benchmarks, there's going to be a mass
The benchmark is like ARR numbers that you might give to the press or fundraising which traditionally is done every 12 to 18 months but sometimes can happen four times in a year. Yeah.
You're a certain company. Also, I just want to say Brandon who wrote the piece, he said at the end he said it seems like the doomers, people who believe AI will turn us all into paper clips, have been defanged, which is good. I disagree that that's true. It seems like AI is like I think there's this whole narrative where like, 'Oh AI is stagnating.'
It's like bro, look at the benchmarks. It's like
Look at the benchmarks. They're saturating. It's incredible.
Exactly. If the benchmarks are saturating, what does that tell you? Benchmarks are saturating. We don't even have good benchmarks anymore because models are way too good. It says benchmarks are just as useful as a used piece of toilet paper.
Exactly. That's the whole point. If models were bad, benchmarks would be useful.
So, I mean the risk is like there was a moment where just multiplying two really big numbers together on a computer, like you needed more memory to do a thousand digits times a thousand digits. Your basic calculator had a limit and then eventually it got to the point where you have enough RAM that this computer can calculate the number of digits in pi to 10,000 digits, probably pretty quickly. Progress along that line is not necessarily the same line of progress towards human labor replacement. We got calculators that can calculate any math perfectly in fractions of a second
And it was useful in a million ways but no one's calling that AGI or ASI that can do anything.
Yeah, because that's specific intelligence.
Exactly, specific intelligence. And so we've created a bunch more lanes of specific intelligence,
But people still feel dissatisfied with the fact that yes, the spikes are getting longer, but it doesn't feel like the intelligence is getting any less spikier. That's what Brandon's feeling. There's nothing about the benchmarks. When you show me a new model card, I'm just like, cool, spiky. The spikes got spikier. Awesome. There's nothing where it's like, oh, okay, it's no longer spiky.
Sure. I mean, look, when I use coding models, they're way better.
Yeah, that's one of the spikes.
That's one of the spikes.
Once you get enough spikes, then what's the difference between a ton of spikes and just like a circle, right?
Yeah, this is the diagram. You've seen the diagram, right? Where it's like you have the circle with the little tiny spikes where the
Better. There are benchmarks like the Anthropic vend where Claude runs a vending machine. At the beginning of the year they released the first paper, I don't know when it was, it was like a couple months ago they released the first part of the paper and Claude was horrible, it lost a ton of money running the vending machine. Then I think today they released a new one where Claude is profitable.
Yeah, we need to get you a massive winter coat to get through this AI winter because the AI winter is happening, dude. I'm telling you.
There is. And it's the age of research. Okay. The age of research.
Imagine it's June next year. It's hot in the Ultradome. Tyler, just keep the coat on, buddy. You can't take it off till it's over.
Till it's over. No, I mean the progress is like people want AI to stagnate so bad. So many people want that.
And it's not happening, bro.
It's like the models are getting better.
Literally, everyone without bags is saying it's stagnating. Well, no. Because all those people are like, 'Oh, I missed Nvidia or whatever.' You think Karpathy's like, 'I missed Nvidia, so I got to go trash AI.'
Well, Karpathy, I think he's very excited about what AI is like right now.
Yeah. Well, I think we can all agree, Tyler, that it's the age of research, but it's also the age of deals.
We got to talk about Trump Media and Technology. This morning I woke up and a lot of people that do mental health podcasts say look at your phone within 5 seconds of opening your eyes, immediately connect to the internet and just start marinating in all the notifications that you maybe got over the last eight hours. That's what I did this morning. I opened my eyes, immediately grabbed my phone, saw this push notification from Bloomberg that Truth Social parent to merge with nuclear fusion firm in a $6 billion deal. And I thought I was dreaming still because it just seemed insane. Trump's been so active this year on the deal making side. Who would have thought he had another multi-billion dollar deal to do this year? So, of course, Trump Media and Technology Group, the social media and crypto company part owned by President Trump, said it would develop a utility scale fusion power plant.
Correct. Okay. President Trump's social media company, which recently expanded into streaming and cryptocurrency, is now entering its fourth act. Fusion power, a promising but still unproven source of alternative energy. DJT and TAE Technologies said Thursday they had agreed to an all stock merger that the company valued at more than $6 billion. The deal would be a metamorphosis for Trump Media, the money-losing parent company of Truth Social, a social media platform that has struggled to gain market share beyond serving as the main online megaphone for President Trump. Mr. Trump is the company's largest shareholder with a large stake worth more than $1 billion that is held in a trust. We're trust trust maxing, of course managed by Don Jr. who is a Trump Media board member. The company based in Sarasota, Florida has only a few dozen full-time employees and has recorded tens of millions of losses in recent years. The merger with TAE would create one of the first publicly traded nuclear fusion companies. According to the release, they intend to begin construction on the first utility scale fusion power plant next year with plans to build additional plants in the future. So the stock is up pretty dramatically today. The CEO over at Mitchell Bendenbower says, 'We've got the tools and the tech and the engineering. What we were lacking was the capital.' So, I went over to Gemini and got a little background on TAE Technologies.
Interesting company. TAE stands for Tri Alpha Energy, which is a pretty cool name.
It's also good advice if you're a beta. Just try being an alpha or just rebrand to an alpha.
Try. Try or T R I, not T R Y. Yeah.
Try Sigma. Try whatever works.
Sigma Energy would be a great name for a small modular reactor company.
I mean there's a trading firm Two Sigma and it's two guys.
You know that they had a founder breakup because they were literally two sigmas.
Well, no. Because one of them was Two Sigma. Oh, T.
No, I think the nominative determinism is that you put two lone wolves together and it didn't work and they were fighting constantly. So, the Two Sigma guys broke up.
Anyway, so people are going to talk about this. The company was founded in 1998 and spent over 25 years developing a unique approach to fusion that focuses on aneutronic power energy generation that produces no harmful neutron radiation. They have raised from Google, Chevron, and Goldman Sachs. So,
Great to see them all get some liquidity now or at least shares in a company that is liquid.
So, does that mean Google is on the Truth Social cap table now?
I don't know. I mean, it's possible they got some liquidity along the way, but seems more likely that Google now has a nice slug in DJT
Wait, I didn't even think about that because this is happening at the top co totally at DJT. So, they have exposure to fusion, cryptocurrency, social media, and streaming.
I mean, Google famously never got Google Circles working, their Facebook competitor. They have YouTube obviously a behemoth, but
True social networking, they've struggled. Maybe this is the foothold for them. Yeah.
Roll in the entire thing at some point.
You know, people are going to talk trash about this like, 'Oh, what is a social network doing building energy assets?' But you look at what Amazon's doing building massive data centers for AWS. Amazon owns Twitch. That's a social network, a streaming platform. What else does Amazon own? Well, they own some diesel generators for backup power at the data centers. Fusion theoretically cleaner. Who would have thought that out of the Silicon Valley tech people, they would be using diesel while the big guy is using clean energy to power his social network.
Fascinating stuff. Let me tell you about Fall Generative Media, a platform for developers. Develop and fine-tune models with serverless GPUs and on-demand clusters.
Trey in the chat says, 'This is such a weird deal. TAE is a legit company.'
Maybe they're all legit. Maybe they're all going legit. Maybe it works out. I don't know. I mean, fusion's hard.
What if True Social just puts up monster numbers? I mean, it's kind of funny.
So hard to call legit right now. It really is an interesting test given that when I type Truth Social into my browser, the homepage is effectively just Donald Trump's profile, and they do have a lot of ads on here.
It'll be interesting to understand what this web page is actually worth.
Well, we'll find out. I mean, Truth Social, I think if you went back a year or two ago when Truth Social was started, which I believe Trump was out of office, he was not president when he started it, and you asked anyone what the terminal value of this asset will be, everyone would say it's not going to be over a billion dollars. And then when it went out, it was like, 'Okay, yeah, maybe they expect it, but it's going to trade down a ton.' I think it's still a multi-billion dollar company, right? DJT, what's the market cap?
Four billion still. That's higher than a lot of people thought.
It traded up almost 40% just today.
Well, we were doing a deep dive on the company a year ago, so you know there's alpha here.
What categories is DJT not in?
robotics. Oh wait, no, they're in pharma, right?
Pharma. So I think TAE does have a biotech division, a life sciences division.
So I'm very bullish on them kind of going into pharma.
Yeah. You got to go further out in the sci-fi curve. No one's pitching a time travel company yet. I think that's an interesting area to get into. What happens after you're traveling through space? You must travel through time, the fourth dimension.
I mean, you could break time travel into life sciences, right? Because you can freeze yourself and wake up in 20 years. You can't go back, though.
Dyson sphere. No one has come out and said we're the company that's going to do the Dyson sphere. It's been much more incremental visions.
Truth Social is also getting into the prediction market game. Launching a native prediction market feature in the app.
It's crazy that they're not.
I wonder if it'll be like PredictTrade on what?
Trade on the content of Trump's next post. That would be fascinating.
Anyway, very fascinating. I am interested to see how future presidents approach business. Who was the guy with the peanut farm?
He got into a little hot water because of his peanut farm. I think he didn't fully divest.
No, he did divest, but he still got in hot water.
It's just remarkable because the peanut farm was so small and so insignificant in the US economy and even in his personal balance sheet. The fact that he was saying, 'Look, I don't want any idea of impropriety, so I'm going to divest from a small peanut farm that I'm associated with so that no one can say that I'm in the pocket of big peanut or pro peanut.' Jimmy Carter divested from the peanut farm. He was not in big oil or a railroad baron or associated with banking. He had very little conflict of interest and he reduced it even further. That's always been a sign of what great looks like. It's been a role model for some presidents. Some have looked and said, 'If Jimmy Carter did it this way, I should also do it this way.' Anyway, Gemini 3 Pro, Google's most intelligent model yet, state-of-the-art reasoning, next-level vibe coding, and deep multimodal understanding.
There is a crazy story in the...
Joe Weisenthal, by the way, one more. He says, sharing that DJT is up 35% to a new 30-day high. And Joe says Trump has been in founder mode for a long time, getting all kinds of criticism about the business from people who aren't actually in the arena. It is crazy the degree to which he is a successful technology founder, billions of dollars in market cap.
People can have their issues with the president, but they cannot deny that with a straight face he is not a technology...
He's never founded a unicorn tech company.
He's never taken a social media app public.
You can't say that. He's never merged with a nuclear fusion company backed by Chevron, Google, and Goldman Sachs.
Dave, have you seen this chart, the AI investment cycle visualized? Almost $1 trillion has been invested into AI so far, and that may just be the start. This is such an interesting way to visualize the flow of money. From a $10 billion deal here, a hundred billion dollar deal there. This shows the investment over time, and you can see Nvidia flowing to Amazon and all the different companies. Where's OpenAI? OpenAI is just flowing around. It's just a cool little data visualization. I thought this was fun to look at.
Did you see Brett Adcock has a new startup?
Yeah. Figure CEO Brett Adcock launches a new AI lab with a hundred million in funding.
Wow. Brett Adcock, of course, the CEO of robotics startup Figure AI, who we believe might have some big news soon based on their holiday party. DeadMass has started a new AI lab called HARK, which will be funded by $100 million of his personal capital, according to a copy of a memo Adcock sent to Figure employees and investors on Thursday seen by Stephanie Palazolo. The AI lab will be focused on building human-centric AI which can think proactively, recursively, and care deeply about people. HARK's first cluster of GPUs went online on Monday. The memo said, though it couldn't be learned how large the cluster was. Adcock will remain the CEO of Figure in addition to his new role at HARK. HARK continues the trend of AI startup CEOs founding new companies while remaining at their old one. I remember Richard over at you launched a new lab himself. More broadly, well-known researchers and executives have raised billions of dollars for so-called neolabs in recent months which hope to exploit new approaches to developing models. Figure previously raised nearly two billion in funding from investors including Parkway, Nvidia, Microsoft, etc. So interesting that he is diversifying when Figure's valuation has been going through the roof.
He also has a company that does metal detectors at schools, right? Something along those lines. He is a serial entrepreneur. Anyway, let me tell you about Finn.ai, the number one AI agent for customer service. Automate the most complex customer service queries on every channel. Tyler clearly has an alt. Tyler Glale. Tyler clearly put this in the show notes. Tyler, not R. Tyler, but this other Tyler says, 'Can't wait for this AI bubble to pop so we can all go back to normal, just like how the internet completely disappeared after the dot bubble popped.' It's another bullish take for AI. It's not going anywhere.
It is a great take. And 18,000 people agreed it's a great take. I also think it's a great take. But this doesn't seem so fast takeoff build, more internet build, which is slow takeoff. The internet had a slow takeoff. Everyone thought it was going to be the super highway, a thousandx growth. Yahoo was going to be worth a trillion dollars. Then it kind of settled in. Took a while to find use cases and value, but eventually it did transform everything.
But this time it's done. I mean, yeah, this time's different, right? Because this time we have the internet, so it's way faster to do distribution.
I don't want to... You weren't even alive there, Tyler. I don't even want to...
I was racking servers in a data center while you were in the cradle.
Melt says, 'A lot of naval gazing about venture. There are a lot of ways to make a couple hundred million dollars. There are a reasonable number of ways to make a billion. There are very few ways to make tens of billions of dollars. This makes a $1 billion fund with 10x DPI obscenely impressive.' Is this a subweet of someone in particular? So Mike Chan says ain't no $1 billion fund doing 10x DPI, and Meltzer says several have, which is wild.
I'm trying to find in the comments a fund that actually did this, and I can't think of any personally. I'm sure they're out there. But yeah, all of that's impressive.
Okay, more importantly, this Chinese streamer cut the Apple Vision Pro's weight in half using helium.
That's genius. This is innovation. This is crazy innovation. Wow.
Imagine walking into an office and every person there is just locked in with a balloon. This is somehow more cyberpunk than just the default. Put the big screen on your face. I like this.
And people say the Chinese can only copy and can't invent new stuff.
Completely disproven with this Vision Pro helium balloon.
So imagine being able to, if you're class clown coded, which I certainly am, walk around the office with a little needle.
No, no, no, no, no. This doesn't actually hurt them. It's still strapped to their head.
But you're popping their balloon, then they have to go get more helium, and helium is a scarce resource.
True. But I would hope they look... He's got a big canister there. Plenty.
You should not be popping this. And you know why? Because Ty in the chat says this should be a blimp. And this is very blimp coded. So you should allow this.
I want this guy to make a personal blimp. Just a harness you can put on with enough helium to take you up.
I think the MythBusters did that. Yeah.
You just get a bunch of balloons.
Doing that and then skydiving down would be pretty cool.
You should work on that over the winter break.
Well, whatever you watch in your Apple Vision Pro, make sure it's on ReStream. One live stream, 30-plus destinations.
If you want to multistream, go to restream.com.
Oh, this is so funny. Alphaton Capital, did you see this? Alphaton Capital post big news. 'We are making a historic $30 million strategic investment in Anduril Industries.' They tagged them and said, 'We are investing in the future of defense tech and the convergence of decentralized AI with next-gen national security. This unprecedented investment positions Alphaton at the forefront of next generation defense technology infrastructure, marking a pivotal moment in the convergence of public markets and advanced national security capabilities.' NASDAQ: Eton. So they're trying to become an Anduril-like holdco basically, or treasury company. And Palmer just quote tweets them into the stratosphere by saying, 'Alphaton is lying. This is not true. Public heads up to CEO Brittany Kaiser, own your data now, the whistleblower, who is behind Bitcoin Alphaton Capital. You do not have permission to use my trademarks to defraud your investors.' And Brittany says, 'Hey, Palmer, we signed an agreement to purchase economic exposure to Anduril shares through an SPV and to accumulate more exposure over time and offer access via formal secondaries product. Given our network's demand for your innovative tech, I will issue a clarification.' And Palmer says, 'That isn't what you and your company said, though. To be extremely clear, Anduril has to authorize these types of transfers, and I will not authorize it.' Wow. And Sarah is there saying, 'So much nonsense fraud happening in these layered SPVs.' This is such a crazy moment.
What is the history of this company? It went out at $280 per share in December of 2020. It's now sitting at exactly $69.
So... I wonder what they have been up to. But that was a crazy announcement. It looks like an AI bot is just posting, but this is a public company.
It does look like some complete slop scam. You would assume that if you go down this particular funnel, you wind up at something like 'send me some random crypto.' It feels very scammy. Palmer is sort of calling it a scam, but it is a public company, I guess.
Yeah. I'm sure Anduril's legal counsel will really enjoy having a nice calm holiday weekend while they clean this mess up. One of many, but it's a champagne problem. Let's come back to this later because we have our guests in the restream waiting room. While we're bringing them in, let me tell you about Cognition. The team behind the AI software engineer Devin. Crush your backlog with a personal AI engineering team. Merry Christmas to everyone. We have Brian Armstrong from Coinbase and T is back on the show. So many repeat guests. It's so great to see both of you. Congratulations on all the fantastic success. Brian, would you mind kicking us off with a little high-level overview of the news? Good to see you.
Yeah, sure. Good to see you guys again. We had a great product event yesterday in San Francisco. We announced the Everything Exchange. So we've got stock trading on there. We've got prediction markets. We've got millions of tokens now trading through decentralized exchanges. We've got simple derivatives. We announced a ton of stuff yesterday, and one of the big ones was prediction markets, which we're happy to partner with Kalshi on. We love Kalshi. They've been awesome to work with.
Prediction markets are huge. They're an important asset class for people who are trading them, but it's also a way for people to get information about what's going to happen in the world. So it's kind of an alternative to traditional media. It's entertainment. It's everything.
Talk to me about how the Kalshi integration actually works. Is it more like a B2B deal? Will a Coinbase customer actually see the Kalshi logo? Will they need two accounts, or are the accounts linked? How are you thinking about the kind of integration taking shape over the next year?
Yeah, it's a white-labeled integration, and Tariq can talk more about his strategy of being a provider to lots of different companies. It made it really easy to integrate. It's not an exclusive arrangement, so we can integrate with others or launch our own. But for right now, we're very happy building on top of Kalshi. It's allowed us to get to market quickly. From a Coinbase point of view, TradFi and crypto are intersecting, or another way to say it is crypto is just eating all financial services. Coinbase is the leader in the space with the most trusted brand, the most deep crypto expertise. We're storing more crypto than any other company in the world by far, about $500 billion of assets. If customers can trade everything where their assets reside, it's our key to winning.
Tariq, take me through how this plays into your growth strategy and trajectory for the next year. Is it a new top of funnel? Building up more liquidity? Opening up different markets? Will we see different Kalshi markets because of this partnership? How do you think things will evolve over the next year?
Let me answer at a personal level. I don't know if you saw Wanna Street yesterday, which is funny. The first time we actually both interfaced with Coinbase was when we got the $100 of free crypto in our freshman year at MIT. We got the $100, and a decade later, we're partnering with Coinbase. Just an amazing personal moment. But from a company perspective, our goal has been to build the first-party product, Kalshi.com, but at the same time, each product requires liquidity. Partnering with brokerages or large platforms with huge customer bases and deposits brings more flow to the exchange, which improves liquidity for everybody, including Coinbase. We're very excited about this partnership. There's going to be a long range of markets available on Coinbase that people can participate in. We're going to be adding increasingly more markets. This is just the start.
How do you guys spend time talking to users? You've both gone viral for different things related to prediction markets over the last couple weeks. Everybody online will form their opinion on why prediction markets are good or bad. But how do consumers feel on your side, Brian? I assume you and the team spend time talking to people to understand what users want out of prediction markets on the Coinbase app and what they're most excited about.
Obviously, sports is a big category driving a lot of demand right now. That's an important use case we see from customers. There are open questions about how that will be treated in society and go through the courts. We think ultimately the CFTC will have federal preemption here. Some states are grumpy about that, but that's fine. Prediction markets are much broader than just sports. It's people betting on economic indicators, figuring out if the Suez Canal will be open next month, political outcomes. The number of markets could be dramatically higher. 1% of people use it as an asset class to trade, but 99% of people looking at these are using it to figure out what's going to happen in the world next. It's an alternative to traditional media. Tariq told me a really powerful idea that policymakers could use this to determine the best policies to implement. If you're trying to build more housing in San Francisco or reduce unemployment, you could put that on a prediction market, get real signal from the market about what would happen, and use that to choose the best policy. Prediction markets are a really powerful tool.
That example in particular, who are the market participants in that situation? I think a lot of people's big concern is that you end up with, like, if I'm a big housing developer, I could potentially influence policy through trading that market. I think part of why prediction markets keep going viral is that in any of these situations, it's easy to imagine some negative externalities. But that's the case for all new technology.
Yeah. Tariq should talk more about it, but the game theory is interesting. There are parties on both sides who stand to gain or lose. If you believe the housing market would be better for your business, you can invest in that contract. But if you're wrong, you lose money. It's the best signal with real skin in the game. Compare that to pollsters running surveys. There are revealed preferences versus stated preferences. Sometimes people state a preference, but the revealed preference is different because of real skin in the game. Prediction markets have much stronger signal than traditional polling. Tariq, you're the expert here. What do you think?
We've dealt with this a lot. Last year when we were fighting the government to legalize election markets and won that lawsuit, the criticism was that people could use money to influence the odds. The best argument is that if you wanted to influence the perception of who will win an election or whether a policy is good or bad, you would influence the polls. There's no cost. You can sample in a certain way, bias it. Or you can go on TV and say things, which is how most of us get informed today. The elegant thing about prediction markets is there's a strong monetary incentive to be truthful and a strong disincentive to lie. People don't lie when money's involved. That's why over time, prediction markets tend to be a more accurate gauge of what will happen. In this specific example about what policies to adopt, it's the same principle.
The debate around which markets are valuable will continue. Is there a meaningful distinction between positive-sum markets versus zero-sum markets? It feels like some of the vibe shift is that if you create a financial product allowing people to invest in the American stock market, it's historically positive-sum. But there can only be one winner of the Super Bowl, which is zero-sum. Is that a reasonable framework to think about where the activity is shifting?
I'd say I don't think so. Derivatives markets like options and futures have historically been zero-sum.
When you buy, right? But they're still very useful. The main criticism is that they're speculators speculating on an outcome. That's okay because without speculators, you don't get the benefits of price and forecasting. If people need to hedge against something like Brexit or an election outcome, they need liquidity. The healthy balance is between people who drive usefulness from the markets and speculators, which is how traditional markets operate.
Yeah, it's an entertaining experience scrolling and trying to understand what could become future headlines. I appreciate the public benefit created through people choosing to trade these markets.
Brian, you've done a great job running a firm that services institutional investors as well as consumers. There's been viral chatter about prediction markets potentially replacing or augmenting insurance products. The initial reaction was, 'Wait, instead of fire insurance, I have to trade a contract every day?' But I was thinking maybe there's a new insurance firm using a prediction market under the hood. Have you thought about the role of institutions in a world where prediction market liquidity is increasing? How do institutions build products around prediction markets that have value to a different consumer?
