Anthropic investors floated $1.5 trillion, not $2 trillion, as the FTC sent subpoenas

Anthropic told prospective IPO investors on October 14 that third-quarter revenue rose 45% to $16.8 billion and that it made money on an adjusted basis. By the next evening, Reuters reported that some of those investors were talking about a $1.5 trillion valuation instead of the roughly $2 trillion floated for a November listing. That same day Axios reported that the FTC had subpoenaed Anthropic and OpenAI. Over the next two days the people who price this deal said what they thought on camera. Ben Horowitz says the IPO will be fine. Steve Eisman says he would not buy it. Chamath Palihapitiya wants a lower price and more cash. My read is that the $500 billion gap is about one question. Can Anthropic hold its prices once cheap open-weight models get good enough?

The number that moved the price

The revenue itself is enormous. The problem is the slope. Anthropic booked $11.6 billion in the second quarter, up from $4.7 billion in the first, according to attendees quoted by Calcalist. That is roughly 147% sequential growth, and it fell to 45% one quarter later. A company asking for 30 times sales gets judged on the slope.

Horowitz, whose firm owns both Anthropic and OpenAI, named the slowdown before he defended the deal on the All-In podcast, taped October 15:

growth has slowed from Q2 to Q3 by a lot in terms of revenue growth, and then we don't know what the ARR or the recurring revenue number is, and so people have questions about how fast the thing is growing.

Ben Horowitz, October 15, 2026, source

Then he gave his range: "So, I think the IPO will be fine. Like, whether it's, you know, you want to be in at two trillion or 1.9 trillion or 2.3 trillion, I think that's what a lot of the debate is about." Note where he put the floor. The investors Reuters heard from were already $400 billion below it.

Eisman says the token binge is over

Eisman has the clearest bear case, and he gave it on The Compound the same day. His argument is about price, not demand. Through the spring, companies told staff to run AI around the clock and paid whatever it cost. He thinks that phase has ended, and that cheaper models now set the price.

So token maxing is done. You've got open-weight models that are a tenth the price. They're not as good as the models, but you know, you don't need a Mercedes-Benz for every road trip, right?

Steve Eisman, October 15, 2026, source

He also explains why this matters past Anthropic's own shareholders. In his telling, Nvidia sells mostly to a few hyperscalers, and most of their AI revenue comes from the two labs. "What I'm most focused on is this business is so concentrated from Anthropic and OpenAI that if there's a price war, none of that matters," he said. "The whole thing will correct."

Asked whether Anthropic should list this year, he said it has to, and soon:

I think they actually need to go public right now because the whole token-max thing is going to make their numbers start to look really bad.

Steve Eisman, October 15, 2026, source

Would he buy the shares? "No way." His case is thin in one place. He offered no evidence of a price war beyond "I'm already hearing" that Claude is cutting prices. Still, the 45% print is the kind of number his thesis predicts.

Chamath would take $1 trillion and the cash

Palihapitiya called the revenue extraordinary on the same All-In episode. He still thinks the company should sell more stock at a lower price. His math is about gigawatts. In his view $200 billion "barely gets them a few gigawatts," and the company should raise it now, while the market is open.

If Anthropic is going to go through and survive an eventual correction, I think it's much better for them to have 200 to 250 billion today than to try to ring the bell at 1.9 or 2 trillion for 80 billion

Chamath Palihapitiya, October 15, 2026, source

He then put a number on it: "I don't think you can raise a quarter trillion at two trillion. I think you can raise a quarter trillion at a trillion." That is the most bearish valuation anyone gave in the window, and it came from someone who likes the company. I read it as a funding argument more than a valuation call. A lab that rents compute on five-year terms needs cash that lasts through a bad year, and the CoreWeave bond scare covered in our October 15 piece showed how fast that money can get expensive.

Sacks likes the business and blames the lobbying

David Sacks was the most bullish voice on the panel about the IPO and the angriest about the company. He said the investor-day leaks could not be trusted because attendees signed NDAs, but he had heard the mood was good. Then he turned:

wearing my hat as an investor, I think it's a very impressive company, which is all the more reason why I find their attempts to kneecap open source through their incessant fear-mongering campaigns and regulatory capture to be so distasteful

David Sacks, October 15, 2026, source

His evidence was commercial. He cited a Coinbase test in which a fine-tuned 9 billion parameter open model cut fraudulent transactions by about 30% and ran cheaper than Claude. "Small models are taking share," he said, and Anthropic's warning about AI worms on smart-home devices came, "coincidentally," right after. Sacks and Eisman agree on the economics and disagree on everything else. Both think enterprises are moving cheap workloads to small open models. Eisman calls that a price war, while Sacks calls it productivity.

The open-weight vendors say the same thing. Misha Laskin, whose Reflection AI released its Beam model this month, told Fox Business on October 16 that demand had turned within months: "from six months ago, not much demand for open models, now the majority of tokens on surfaces like OpenRouter are actually coming from them." That is his claim about one routing service, not the whole market, but it points the same way.

Washington sent subpoenas the same week

The FTC's civil subpoenas seek records on security incidents, safeguards, marketing materials and privacy policies, and an agency spokesperson confirmed them, Techstrong reported. FTC chair Andrew Ferguson tried to lower the temperature on CNBC on October 16: "This is an ordinary consumer protection investigation of the type that we do all the time."

Jay Clayton, who chairs the White House superintelligence task force, did not lower it. Asked the same morning whether the probe could hold back new model releases, he said "yes, absolutely," and added:

If there are unfair practices that would fall under the FTC, if there are, frankly, violations of criminal law that would fall under the DOJ, we are going to enforce them vigorously.

Jay Clayton, October 16, 2026, source

The same interview had a detail Sacks did not mention. Clayton said the task force asked PCAST, "led by David Sacks," to run its AI risk study. The man who calls Anthropic's warnings fear-mongering now runs the government's review of those risks. An IPO prospectus has to list regulatory risk, and that line just got longer.

Is $1.5 trillion the floor or the opening bid?

I think $1.5 trillion is closer to the floor than the ceiling, but the reason matters. Nobody in this window doubted the revenue. Horowitz, Sacks and Palihapitiya all called the company exceptional, and even Eisman's bear case depends on a price war he cannot yet prove. The discount comes from the growth slope and the cost of cash. If the fourth quarter shows growth back above 45% with prices holding, Horowitz's range wins. If it shows Anthropic cutting prices to keep share from open models, Eisman wins, and Palihapitiya's trillion starts to look generous. The FTC case is unlikely to decide the price on its own. It does give buyers one more reason to ask for a discount, and on these numbers they are already asking for one.