More Trillion Dollar IPOs, Anthropic $3T, Zuck's Price War, China Ends Open Source?, Trump Accounts

They should report the EPS gains attributable to AI.

Brad Gerstner thinks it is very likely that both Anthropic and OpenAI will go public within the next six to nine months barring a black swan event

when you wake up in the morning and like 14 jobs have been done, you're like, wait a second, this is completely different.

everybody will get to a point they'll get to it at different times where they just say you know what like it shouldn't really matter what model I using if I get an answer that I think is reasonable and I can kind of go about my day

The ability to create verticalized models is getting easier and easier every six months or so, and major eight and nine figure customers of frontier labs are moving toward their own proprietary verticalized models.

China restricting its open source models would hurt China more than the US, making it more likely chess-playing than a genuine strategic threat

Trump Accounts are officially named Trump accounts but the enabling legislation is called the Invest America Act, and many Democrats call them Invest America accounts

That's way better than social security.

Open source as a share of enterprise spending is actually decreasing.

SpaceX raised $75 billion at $1.75 trillion valuation in its IPO

SpaceX is up 25% from its IPO price, currently trading at roughly $2 trillion on approximately $35 billion of forward revenue

Anthropic confidentially filed on June 1st

Polymarket says 65% chance Anthropic's IPO will happen this year on light volume of 360k

Gavin Baker thinks Anthropic will end 2026 with over $100 billion in revenue and be very profitable, and would trade at $3 trillion if it went public now

Anthropic is rumored to be trending over $100 billion in revenue compared to SpaceX's $35 billion

OpenAI's cash burn is still quite high due to the diffuse nature of their business and more reliance on consumer than enterprise

Anthropic may actually be accidentally profitable

Brad Gerstner's CTO reported that token costs are doubling every 45 days while downstream productivity improvement is 5% max

More tokens are needed to reach the next iteration of AI improvement because performance has effectively already asymptoted

SpaceX's IPO featured a staged lockup release tied to milestones and early inclusion in indexes, raising $75 billion

The peak-to-trough drawdown in the six months post-IPO for SpaceX was 50%

Anthropic kind of passed OpenAI on a revenue trajectory.

OpenAI has kind of got its swagger and mojo back.

GPT-6 coming out within the next 30 days.

OpenAI rumored revenue is around $70 billion per year.

$70 billion is still twice where the revenue of SpaceX is at.

Anthropic rumored revenue is over $100 billion.

OpenAI has more complexity associated with the corporate restructuring they have to go through.

Altimeter would be a buyer at scale and at size in both of those IPOs.

Accredited investor laws are insane that we have in this country and keeps people from participating in these things.

I don't expect that they're going to be priced in a way where you're going to get a 50 to 100% durable bounce out of the IPOs.

All of these companies are going to compound revenue at well over 30% for the next many years.

The EPS growth of the S&P 493 attributable to AI was 9%, with the overwhelming majority from pricing power sitting on top of inflation and another 3% from buybacks, leaving actual AI ROI somewhere between zero and 2%.

Today, 99% of Uber's engineers use AI tools.

More than 70% of pull requests are attributed to local or cloud agents at Uber.

Uber's engineers have built 2,500 agentic skills.

Chumath is right. The only question is on what timeframe. There's no doubt that there is a lot of money being spent today that is in the experimental bucket, right? Where I think there probably isn't direct ROI, Chumath, to your point, but I think we're so early, nobody cares.

The total addressable market here is every single small, medium, large company on the planet.

We've never seen revenue growth like this because we've never seen a TAM like this.

There are millions of customers independently, economically making the decision that is rational for them every day that it makes sense, like Praveen at Uber.

Jensen Huang has talked many times that all of his design work all of his design work now at nvidia is using ai to design the next generation chip the machine is building the machine

If these guys end the year over $100 billion, I think that they're on a revenue trajectory that they could three to five X again next year.

200 billion of incremental revenue is incomprehensible in the history of Silicon Valley

intelligence is the largest TAM we've ever seen in the history of the world

if the average salary is $100,000, $150,000 at this organization, it's only an incremental 3%, 4%, 5% on top of their salary

did it make that person 3%, 4%, 5% times more effective at their job? And I think the answer is yes.

somebody who is creating a subnet that is putting glm 5.2 and other models available at really cheap prices so i all of a sudden experienced because they gave me an api key having my token costs go down 95 percent

when you have unlimited tokens as an exercise which is going to come to everybody eventually everybody's going to learn how to drop the price by 95 percent

we've seen 90% reductions in the price of tokens for each of the last two and a half years.

for 18 months since the deep seek moment, right? When the deep seek moment happened, the markets fell 40%.

Many started arguing that the frontier models were screwed, that open source was going to kill them, that they were closing the intelligence gap, that model routing was going to make it easier and easier to route these tasks to cheap tokens.

despite all of those arguments, the facts on the field are just the opposite. The share of economic value. Right. 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, you know, to the other guys.

Coinbase and DoorDash built a token routing system, a layer of middleware that allows them to send frontier tasks to frontier models and non-frontier tasks to more mundane models.

when the corporate CFO gets involved that be an entirely different conversation altogether

There is not a single country in the world that is not trying to figure out its own sovereign AI strategy.

Countries do not believe using a closed source American model is the answer to their sovereign AI strategy.

Certain countries are willing to spend the money to take an open source model like NVIDIA's and stand up their own stack soup to nuts for their own people and companies inside their own country.

Implementing open source AI is very hard compared to just firing up Claude and having Claude already approved in your organization.

It took Jason Calacanis hours to configure his setup to get onto the BitTensor network and get OpenRouter going.

DoorDash CTO Andy Fang reported that with their internal coding benchmarks they are able to confidently introduce open weight models into their AI code review without degrading code quality, using frontier models like Anthropic for the hardest work and delegating lower level work to Kimi 2.6.

The preponderance of tokens today are already shifting toward cheaper, lower-tier lagging models out of OpenAI or lagging models out of Anthropic.

Meta's new model Llama or related release was described as offering the same quality at like one one-hundredth of the cost of competitors.

The difference between spending three dollars on a cheap model versus fifteen dollars on an expensive model to replace a two-hundred-dollar-per-hour consultant is irrelevant if the expensive model delivers bulletproof results.

There is no evidence on the field today that the intelligence gap between commodity and frontier models is collapsing to the point that people won't pay for frontier models.

The UAE has its own Abu Dhabi Technology Innovation Institute shipping the Falcon model.

Saudi Arabia has Humane and is doing their own Arabic LLM models.

Japan is investing $6 billion in a consortium called Neotera and is skipping ahead to physical AI, i.e. robotics.

Open source went from 19% last year to 11% this year as a share of enterprise spending.

Once enterprises figure out what the workflow and workload are going to be and exactly what they're trying to accomplish, then they can use a small, highly trained model.

For customer support, your model doesn't need to know physics.

Nikesh Arora said that enterprises would like model fungibility — the ability to hot swap models — but no one has figured out how to make context, memory, and history fully portable.

Ali, the founder of Databricks, found that for the same model, the choice of harness can significantly save costs by about 2x.

With GLM 5.2, tasks are literally getting cut in half using the same model but with a different harness.

When I optimized it, it was like 80% less token use.

Every single hyperscaler in the world is going to provide tools that allow you to achieve some level of model fungibility.

The non-consensus argument might be that intelligence is not converging at all — that as superintelligence becomes fully self-recursive, the smarter your model gets the more revenue you get, the more compute you can buy, and the better the model you can build.

Over the course of the next two to three years, as we take on much more complex agentic tasks, the distance between the frontier and everybody else doesn't converge, it actually extends.

Lovable went from $100 to $600 million in revenue over the last two years, going from $0 to $350 million in the first two years.

Both the CEO of Lovable and the CEO of Eleven Labs confirmed they are spending tens of millions of dollars with frontier models and are working essentially on their own models.

Decagon said that 90% of their usage now is being sent to open models

those are open models that they've had the opportunity to post-train on and do a huge amount of customization based on all of their learnings and all of the data that they've gotten

Anthropic is at 60-something billion ARR and OpenAI at 40-something billion ARR, and no other AI company registers meaningfully in token revenue market share

A year ago it seemed like we had five major labs; now it seems like there's a top two and then everybody else

CCP officials are reportedly considering restricting overseas access to China's top AI models, with Chinese regulators meeting with Alibaba, ByteDance, and Zhipu to discuss limiting access

China is making any theft or leaks of AI research a national security offense

Manus was a Chinese company that tried to go to Singapore and the CCP pulled those employees from Singapore back to China

ByteDance's model is the number one model in China and has always been closed source

Alibaba's Qwen was open and is now going closed source

Zhipu's GLM 5.2 seemed to be catching up to what was then commercially available as the American frontier at certain tasks

GLM 5.2 has watermarks from Mythos all over it, indicating Chinese labs were distilling from American frontier models

Brad Gerstner spent time in DC talking with the White House and Treasury and found absolute agreement on doing everything to stay ahead of China on AI, with the president personally interested in how far ahead the US is

Frontier labs told Brad Gerstner they aren't making open source models because there's not a lot of demand for it

If China starts cracking down on their AI labs in the same way that doomers want to do in the US, that would be the best thing that could ever happen to the US

The president's big AI policy speech about one year ago declared that the US was in an AI race and America had to win it

Congress is more responsive to the Doomer community and that community is creating a lot of political pressure right now

The throttle on AI progress might not be software or chips—it might be energy

Between now and 2050, the US is about three entire Californias worth of energy short, assuming only regular consumption of devices and cars, fridges, televisions, and computers

Taiwan runs on LNG and has only two or three weeks of it, meaning a Chinese blockade would cause Taiwan to run out of energy immediately

The Trump Accounts app became the number one app in the world by top downloads

The Invest America Act was passed into law about one year ago as part of the reconciliation bill

On July 4th of this year, the Trump Accounts app went live, millions of accounts were created and funded

Starting with $1,000 and matching contributions while saving $10 a week results in $50,000 at age 18

Over a million and a half accounts were created in the first 24 hours after the launch, with over a billion dollars of deposits

The president suggested auto-creating accounts for all 50 million to upwards of 70 million kids under the age of 18

Michael and Susan Dell anchored philanthropic contributions with over $6 billion—$250 for each of 25 million children, primarily lower and middle income kids

SpaceX's president Gwen Shotwell contributed $350 million in SpaceX shares for children of lower income communities

Micron put in $250 million, up to $1,000 per employee

Brad Gerstner contributed $100 million personally to Trump Accounts

We think we can raise a hundred billion dollars in the first 12 months

Parents signing up for Trump Accounts are across the income spectrum, economic spectrum, and political spectrum

Cory Booker, Gavin Newsom, Governor Wes Moore, John Fetterman, Senator from Pennsylvania have come out and supported these accounts

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

you can donate up to $5,000 a year to your kid, as long as they're under 18

it's not just you, it's any friends and family or others can contribute as well, which is new

they get tax-free compounding until they're 18

your employer can contribute up to $2,500 tax-free

you'll be able to convert the Trump account very cheaply into a Roth IRA. And now they're going to have $200,000 to $300,000 potentially in that account

if you start with two to three hundred thousand dollars at age 18 you'll be at 10 million dollars plus by age 60 if you just let it compound

it's got to be voluntary. And number two, it should go into citizen accounts, right? Privately held in citizen accounts, a compound for their life.

we are on a trajectory now that we're going to have over a hundred million of these counts set up over the next decade

you're going to add 3.7 million a year so you're going to be at a hundred million private individual accounts that are compounding for people's lives

every kid owns a little bit of Nvidia, a little bit of Microsoft, a little bit of Apple

over the course of the next 15 years, you could have somewhere between two and four trillion dollars added to the accounts of families and kids who would have otherwise had zero

when Gwynne Shotwell contributes 2 million shares of SpaceX to 2 million kids, each kid's getting a share of stock that's worth $150

they take 40% from their offices and their salaries

if a Trump account had been maxed out and you have the standard market rate of return that we've had for, say, the past 30 years, then by age 28, that kid will be a millionaire

The $50,000 and the $200,000, it doesn't assume maxing. That just assumes people are adding $50 a month

we have 529 accounts that already helped the top 10%

it's the largest change to our social contract since 1935 and social security

AI token costs doubling every 45 days with only ~5% productivity upside creates an unsustainable cost-benefit dynamic that will eventually force a reckoning across all companies

Early index inclusion of a newly public company risks forcing passive investors to buy at peak price before a potential 30-50% drawdown

Once a company is valued at over a trillion dollars, like the get rich quick schemes are over.

OpenAI got very distracted with Sora and the Disney relationship going consumer, then realized the revenue seems to be in enterprise first.

OpenAI kind of gave the Google position, the high growth position, to Claude and Anthropic and took the Yahoo position.

Enterprise is probably a little bit more brittle because there are fewer buyers and they're more demanding.

At some point, a million dollars a year on tokens will require showing an ROI above the risk-free rate of return.

Well it less about being dismissive that way It more that regulators and other people won necessarily allow to use it the way you want Okay so finance HIPAA yeah there HR data You're not allowed to put that to work just yet.

when i use fable five the problem is that it's nerfed on a bunch of things that i would normally research you know i was with somebody this weekend and he was telling me about some health thing and i put it into fable and it's like it won't answer you

If open-source sovereign models are 95–99% as good as frontier models, some countries will claim they are just good enough, posing a competitive risk to frontier AI providers.

Companies will not have the earnings growth to justify AI costs without going on some long protracted carve out of cost, and most companies don't have the nerve to do it.

In a moment of earnings misses, people will find it very difficult to lay off other people and will instead cut AI costs, making the ROI case for frontier AI models more fragile.

Enterprise CTOs are seeing their compute or token costs skyrocketing and are trying to figure out how to put the brakes on this and ensure ROI.

Enterprise CTOs would like to shift their token consumption to cheaper models for efficiency, but face the AI sovereignty issue of worrying about giving up secrets.

Enterprises are giving away their sauce or the alpha in their business to a frontier lab that may one day be competing with them.

No one has really figured out a way to abstract memory, context, and history away from the model yet, which prevents true model fungibility.

Open source model usage represents 'dark tokens' that don't appear as revenue, so token utilization at companies like DoorDash using open source is invisible in market share metrics

Lower-level bureaucrats could make ham-fisted decisions—like banning something without truly understanding all the implications—that are counterproductive to winning the AI race against China

Some people with TDS are refusing to enroll their children in Trump Accounts due to the political branding, potentially denying them compounding investment returns

this is a tremendous new philanthropic platform and that's really important especially in this time of growing anger and backlash and populism against billionaires and people questioning whether the system is rigged and whether they can be successful in america

I don't like this idea of shaking down our companies, taking their shares and then putting them in some government slush fund that perhaps Bernie or AOC or somebody is going to control in the future.

this is so much more efficient and so much better than the whole NGO industrial complex

SpaceX's IPO is considered a template for Anthropic and OpenAI on total raise size, pricing, liquidity structure, index inclusion timing, and lockup staging

ChatGPT took the consumer brand lane for large language models while Claude took the corporate/enterprise lane.

Consumer becomes an incredible safe harbor because you have tens of millions of buyers at a much smaller price point, inoculating you from an ROI discussion.

Every single person in every single organization is playing with these tools. So if everybody's playing with it, everybody's trying to apply it all at the same time

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.

inference is being impacted like three or four different ways the software is getting better open source at the same time you're going to have distributed networks like tau and you're going to have better you know chipsets from grok and cerebrus etc all that's happening at the same time

We've talked a lot about Jevons paradox, which I think you're referencing here, which is you're going to use a hell of a lot more when it happens.

when the iPhone was a novelty, everybody would keep upgrading because you expected that the new price was worth it. And then at some point, there's a moment, you can debate when it happened, where people said, you know what, I'm just going to keep the old phone because it's good enough and I just don't see the difference.

Three trends shaping AI adoption: geographic penetration of humans, transition from experimentation to ongoing repeatable usage, and integration into existing regulatory infrastructure.

Token workload segmentation: summarizing a document may take 20,000 cheap tokens suitable for a lagging or open source model, while replacing a software engineer for two hours may take two million expensive tokens where using a 95%-as-good model carries high consequence.

For mature use cases, enterprises want to go open source; for immature use cases, which are all the new things people are discovering right now, they want to use the most capable general model available.

Not only routing to the right LLM, but routing to the right harness — multiplexing different harnesses and models for different tasks.

If you look at the benchmarks today, models seem to be converging, yet if you look at the revenue distribution, it's not converging at all.

Immature use cases use the most powerful frontier model; once well-defined, workloads migrate to post-trained open models

Chinese labs stay open until they catch or approach the frontier, then go closed to capture value — mirroring what OpenAI did going from open to closed models

The core Trump Account structure: $1,000 for every child at birth in a privately owned investment account invested in the S&P 500, free for the lifetime of the recipient, accessible starting at age 18

On the one hand, you have Bernie and Mondami. They want to take and tax all these corporations. They want to control all that money in Washington and decide who gets it

let's set up a private account for every kid in America. Let's fund them. Let's not make them dependent. Let's make them independent of the government to build wealth on their own, financial literacy on their own, more likely to graduate from high school, start a business, buy a home

when the kid turns 18, they can get access to it and they can do a rollover into an IRA or into a Roth IRA, which is even better because when the Roth IRA matures, you don't pay tax on the money that gets distributed out of it

we can also set up a pooled account, David, where it can be distributed over time. So you distribute it to all the kids subject to the limits that you have today. and then any remainder you can distribute to the three and a half million kids that are going to be born next year

philanthropists can contribute towards that $5,000 per kid, right? So when Gwynne Shotwell contributes 2 million shares of SpaceX to 2 million kids, each kid's getting a share of stock that's worth $150. So now that's counting against their $5,000 limit.

if your employer takes 2500 hundred bucks out of your salary and puts it in your kids Trump account then you're it's reducing your taxable income so it's a no-brainer if you're a profitable company your employees are gonna love you

If you can get out via IPO now, you should get out now before token cost economics seep into the water table and compress valuations

Uber decided to use forward-deployed engineers placed into departments to work with department heads and build agentic pods to deploy AI beyond engineering.

i went to open router i got my own keys i've been playing with glm

once i got down to 95 cheaper i started setting my my agents, instead of doing daily runs to doing hourly runs. Then I took my agents from doing one task and I broke them up into three agents and have them doing three different things on the hour.

I have a trend spotting agent running every hour informing me of the top three or four trends

The US government is going to take steps against distillation of American frontier models by Chinese labs

We need nuclear, more solar, more batteries, and more of everything to address the US energy deficit

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 but you know listen we got to work through treasury the white house social security etc

the best way to do this is wait till your kid is actually not a dependent anymore. Like, so maybe they're in college or they just graduated from college and they're in like the 0% tax bracket because they're not making any money and then do the conversion

You should convince OpenAI and Anthropic to give the equity of those companies. If this is going to be as big as you say, $100 billion, $300 billion, zillion, trillion, put it into the accounts of every kid.

every employer should be signed up to be an employer that can contribute because again you could take that 2500 and hopefully look it's additive

“our token costs are doubling every 45 days”

“When you see Mark Benioff in Action Man, this guy is a master. He is the impresario of impresarios.”

“the actual ROI was somewhere between zero and 2%.”

“cheaper, pretty damn good. 90% is good enough to do all these tasks”

“today we're releasing MuseSpark 1.1, a strong agentic encoding model at a very low price. It's available through our new meta model API and in meta AI.”

“anyone who's saying that these closed models are going to lose or are somehow losing, you're just not seeing it in the data.”

“the spirit is willing, but the flesh is weak”

“Chinese authorities are deeply worried about the potential for mythos to exploit software vulnerabilities and that Washington might deploy a model against Chinese interests”

“The absolute best thing that could happen for America in terms of winning the AI race against China is if China somehow sprouted their own Doomer community”

“How do we get more people auto enrolled in this faster? I want every kid to have the shot to have this.”

“the antidote to more socialism is more capitalism. And as I told the President, this is more capitalism.”

“Yeah, they're just grifting.”