PREDICTIONS ON THE RECORD
Dated and falsifiable — full record at /api/episode/<id>.
Gavin Baker · by end of 2026 · 2:54
Anthropic will end 2026 with over $100 billion in revenue and be very profitable, and would trade at roughly $3 trillion if it went public now.
“he thinks they're going to end 2026 with over $100 billion in revenue and very profitable.”
Brad Gerstner · next 6-9 months · 7:09
Anthropic and OpenAI are very likely to complete their IPOs within the next six to nine months (this year or early next), absent a black-swan event.
“Yeah, I think it's very high.”
Brad Gerstner · next year · 19:57
If Anthropic and OpenAI end the year over $100 billion in revenue, they could 3-5x revenue again next year (toward $300-400 billion).
“If these guys end the year over $100 billion, I think that they're on a revenue trajectory that they could 3 to 5x again next year.”
Brad Gerstner · the next many years · 12:22
All of the major AI companies will compound revenue at well over 30% a year for the next many years.
“all of these companies are going to compound revenue at well over 30% for the next many years.”
Chamath Palihapitiya · unspecified · 14:40
At some point companies will be forced to show an AI ROI above the risk-free rate of return or face angry investors.
“at some point, you're going to have to show an ROI that's above the risk-free rate of return. Otherwise, you're going to have some angry investors on your hands.”
Brad Gerstner · the next two to three years · 47:22
Intelligence may not be converging at all; over the next two to three years, recursive superintelligence could extend the frontier labs' lead over everyone else rather than close it.
“the distance between the frontier and everybody else doesn't converge, it actually extends, we shall see.”
Brad Gerstner · first 12 months · 1:08:39
The Trump accounts philanthropic platform can raise $100 billion in its first 12 months.
“We told the president we think we can raise a hundred billion dollars in the first 12 months.”
Brad Gerstner · next 90 days · 1:11:51
All 50 to 70 million eligible kids' Trump accounts will be auto-created within the next 90 days using existing government data.
“Our intention is to get all 50 to 70 million accounts created over the course of the next 90 days using all of this data.”
Brad Gerstner · next 15 years · 1:20:20
Over the next 15 years, between $2 and $4 trillion could be added to the accounts of families and kids who would otherwise have had zero, reaching over 100 million accounts.
“over the course of the next 15 years, you could have somewhere between $2 and $4 trillion added to the accounts of families and kids who would have otherwise had zero.”
David Sacks · by age 28 · 1:22:09
A maxed-out Trump account earning the standard historical market rate of return will make a kid a millionaire by age 28.
“by age 28, that kid will be a millionaire.”
WHO CLAIMED WHAT
Every position taken in the episode, attributed to its holder, with conviction (0–1) and the verbatim receipt.
Chamath Palihapitiya — 9 claims
opinionconviction 0.75:43
AI model improvement has effectively asymptoted — each next increment of quality now requires far more tokens, so token costs are exploding while downstream productivity stays roughly flat.
“you need to use a lot more tokens to get to this next iteration of improvement because we've effectively already asymptoted.”
opinionconviction 0.75:43
The frontier labs should IPO now, before the coming AI-ROI reckoning seeps into market sentiment, to exit at a high price and raise the most money.
“if you can get out now, you should get out now before all of that starts to seep into the water table.”
opinionconviction 0.714:40
Enterprise AI revenue is more brittle than consumer because enterprise buyers are fewer and more demanding and will eventually insist on ROI, whereas tens of millions of consumers at low price points inoculate a provider from that scrutiny.
“Enterprise is probably a little bit more brittle because there are fewer buyers and they're more demanding.”
opinionconviction 0.729:07
A 'good enough' plateau is coming for AI models — like keeping an old iPhone — where the specific model stops mattering as long as the answer is reasonable.
“it shouldn't really matter what model I'm using if I get an answer that I think is reasonable”
assertionconviction 0.829:07
Every country is pursuing its own sovereign-AI strategy and does not believe relying on a closed-source American model is the answer.
“there is not a single country in the world that is not trying to figure out its own sovereign AI strategy. And I don't think they believe using a close source American model is the answer.”
opinionconviction 0.729:07
Countries will accept open-source models that are 95-99% as good and stand up their own sovereign stacks rather than accept technical dependence on US frontier labs.
“if the models are 99% as good or 95% as good, there's going to be a claim that some countries make, which is, it's just good enough.”
opinionconviction 0.629:07
Anthropic's Fable 5 is over-restricted ('nerfed') and refuses legitimate research queries such as health questions.
“when I use Fable 5, the problem is that it's nerfed on a bunch of things that I would normally research.”
opinionconviction 0.731:34
Most companies lack the nerve and capability to undertake painful cost carve-outs and will simply let AI-cost problems compound until forced (as Zuck had to be pressured into acting).
“And most companies, you know this, they just don't do it.”
opinionconviction 0.81:09:20
Refusing a Trump account for your child because of the 'Trump' branding forgoes an irresponsible amount of lifetime compounding value and is patently insane.
“that is an irresponsible amount of money to not give a kid because you don't like the fact that it's called a Trump account.”
Brad Gerstner — 9 claims
assertionconviction 0.87:09
Anthropic's and OpenAI's IPOs would be blockbusters, following the textbook blueprint SpaceX established on pricing, lockups, and index inclusion.
“based on the SpaceX success, I think it would be a blockbuster IPO.”
opinionconviction 0.712:22
Trillion-dollar-plus AI companies won't deliver durable 50-100% IPO pops, but they will be compounders growing revenue well over 30% a year for many years.
“these things can be compounders. They're going to compound at the rate they compound revenue. And I think all of these companies are going to compound revenue at well over 30% for the next many years.”
opinionconviction 0.7522:21
It is far too early for AI ROI to matter because intelligence is the largest total addressable market in history, so frontier revenue will keep compounding even amid token optimization.
“intelligence is the largest TAM we've ever seen in the history of the world.”
assertionconviction 0.8527:25
Despite the post-DeepSeek predictions that open source would kill the frontier labs, the frontier labs' share of economic value and wallet is actually rising while commodity tokens shift to the cheaper providers.
“The economic value, the share of wallet, is actually increasing to the frontier labs, while the share of tokens, these commodity tokens, is obviously going up to the other guys.”
opinionconviction 0.837:35
For long, high-value tasks like replacing an engineer or a $200/hour consultant, the inference-cost saving from a cheaper model is irrelevant next to the risk of a 95%-as-good model failing mid-task.
“The difference between spending 3 bucks on a cheap model or 15 bucks on an expensive model to replace a $200 an hour consultant, it's just irrelevance.”
assertionconviction 0.838:19
There is no evidence today that the intelligence gap between commodity and frontier models is collapsing to the point people stop paying for the frontier; it may develop later but is not on the field now.
“There is no evidence of that on the field today.”
opinionconviction 0.758:08
China restricting its open-source models is more chess-play than real threat, since it would hurt China far more than the US, which can spin up its own open-source models (Reflection, NVIDIA).
“I think this is probably more chess playing by China than actual threats because it would hurt them a lot more than it would hurt us.”
assertionconviction 0.858:08
GLM 5.2 carries watermarks from Anthropic's 'mythos' model proving distillation, and the US government should (and will) take steps against distillation.
“GLM 5.2 has watermarks from mythos all over it, right? So we know they were distilling”
opinionconviction 0.81:22:27
Trump accounts are the largest change to the US social contract since Social Security in 1935 and a superior, direct philanthropic platform versus the NGO/charitable-middleman complex.
“It's the largest change to our social contract since 1935 and social security.”
David Sacks — 7 claims
opinionconviction 0.840:28
Enterprises would like to diversify off closed models but most lack the technical ability; only sophisticated firms like Coinbase and DoorDash built the token-routing middleware to do it — the spirit is willing but the flesh is weak.
“I think this is a case of the spirit is willing, but the flesh is weak.”
assertionconviction 0.8540:28
Anyone claiming closed models are losing is ignoring the data: closed-model revenue is skyrocketing while open source fell from 19% to 11% of enterprise AI spending year over year.
“I think that open source went from 19% last year to 11% this year. So open source as a share of enterprise spending is actually decreasing.”
opinionconviction 0.7543:03
For immature, undefined use cases enterprises want the most powerful general model; only once a workflow is well-defined does it make sense to move it to a cheaper post-trained open model.
“But if you don't know exactly what you're going to use them for, you want the most powerful general intelligence that you can get.”
opinionconviction 0.7552:28
The AI model layer has effectively become a two-company duopoly by revenue (Anthropic and OpenAI), collapsing from what looked like five major labs a year ago, consistent with tech's historical monopoly/duopoly pattern.
“A year ago, it seemed like we had five major labs. Now it seems like there's a top two and then everybody else.”
assertionconviction 0.7556:03
Chinese labs (like OpenAI and Meta before them) follow a pattern of staying open to catch the frontier, then going closed to capture the value for themselves.
“the tactic is you stay open until you catch the frontier, you get close to it, and then there's a really compelling incentive to go closed because you want to capture all the value for yourself.”
opinionconviction 0.71:00:32
The real AI-policy risk is not the president — who wants to win the race — but lower-level bureaucrats or Congress making ham-fisted, counterproductive moves like banning something without understanding it.
“The question is whether at lower levels of the bureaucracy, you can get mistakes being made.”
opinionconviction 0.81:40:07
A Trump account is effectively better than an IRA and one of the best wealth/tax-planning vehicles available, thanks to tax-free compounding from birth, employer and philanthropist contributions, and a Roth rollover.
“It's actually better than an IRA.”
Jason Calacanis — 4 claims
opinionconviction 0.723:16
AI is a bottom-up technology that hits every employee in every department at once, which is what explains the unprecedented, hard-to-model revenue ramp.
“when a bottom-up technology hits everybody at the same time, that's what would explain this revenue ramp that we're all having a hard time adjusting to.”
assertionconviction 0.7553:41
Revenue understates open-source usage because self-hosted 'dark tokens' don't register as lab revenue — only as hosting/hardware cost — so the revenue metric misses utilization at firms like DoorDash.
“when you're doing open source, those are dark tokens. Those don't come up as revenue.”
opinionconviction 0.71:02:31
Energy — nuclear, solar, and batteries — not chips or software, may be the real gating factor on AI in the US, and clearing it is a regulatory challenge.
“We have to get nuclear running, more solar running, more batteries, more of everything.”
opinionconviction 0.751:13:27
If Trump accounts succeed, US equity ownership could rise from ~50% to 70-75% of the country, making it a unifying antidote to rising socialism.
“we're going to go from 50% of people owning equities in the country to as much as 70, maybe even 75% of the country having access”
Mark Zuckerberg — 1 claim
assertionconviction 0.735:25voiced by Chamath Palihapitiya
Meta is launching an AI price war, promising the same quality at roughly one one-hundredth of the cost.
“I'm going to give you the same quality at like one one hundredth of the cost.”