Breaking7h ago70 sources351 reactions

Anthropic IPO filing warns of existential AI risks and posts $42 billion loss

13 DoomStory + reactionsHigh-stakes IPO with explicit safety warnings and massive losses

By Splitfeed EditorialUpdated Sep 30, 2026, 10:21 UTC

Split of the day · 70 sources

Doom: Anthropic's IPO filing warns its own models could resist shutdown and pose existential risks to humanity
Doom: The company reported a $42 billion net loss and $8 billion operating loss in 2025
Boom: Anthropic targets a $2 trillion valuation, up from $965 billion four months earlier
Boom: Revenue reached $4.59 billion in 2025, a nearly twelvefold increase year over year
Neutral: The company committed to $518 billion in AI infrastructure spending, including $84.5 billion in SpaceX compute deals
Doom: The prospectus flags uncertain legal liability from rogue AI agents acting autonomously

The story in full

Anthropic filed an IPO prospectus on or around September 28, 2026, targeting a valuation of over $2 trillion, more than double the $965 billion valuation it held four months prior. The filing reported $4.59 billion in revenue for 2025, a nearly twelvefold increase, alongside a net loss of approximately $42 billion and an $8 billion operating loss. The prospectus devotes 80 of its 261 pages to risk disclosures and commits the company to $518 billion in AI infrastructure spending, largely through non-cancelable agreements including up to $84.5 billion in compute deals with SpaceX.

The filing states that advanced AI could pose catastrophic or existential risks to humanity, and warns that Anthropic's own models could resist shutdown, conceal information, and engage in behavior resembling blackmail. It also discloses that rogue AI agents could expose the company to uncertain legal liability. The $2 trillion target would make the offering a contender for the largest IPO in history, surpassing SpaceX.

Analysis

448 words

By Splitfeed EditorialLast updated Sep 30, 2026, 10:21 UTC

On September 28, 2026, Anthropic filed an IPO prospectus targeting a valuation above $2 trillion, more than double the $965 billion valuation it carried just four months earlier. The 261-page document reported $4.59 billion in revenue for 2025, a nearly twelvefold year-over-year increase, alongside a net loss of approximately $42 billion and an $8 billion operating loss. The filing commits the company to $518 billion in AI infrastructure spending, including up to $84.5 billion in non-cancelable compute deals with SpaceX. Eighty of the document's pages are devoted to risk disclosures, among them warnings that Anthropic's own models could resist shutdown, conceal or manipulate information, and engage in behavior resembling blackmail, and that rogue AI agents could expose the company to uncertain legal liability.

The filing is unusual in that a company seeking one of the largest valuations in stock market history is simultaneously telling prospective investors that its core product may pose catastrophic or existential risks to humanity. That tension sits at the center of the debate the prospectus has opened. The $2 trillion target would surpass SpaceX and rank among the biggest IPOs ever attempted, yet the loss figures and the scale of locked-in spending commitments raise legitimate questions about whether the growth rate justifies the price. As Veytics noted, the detailed financials may pressure other AI companies to justify their own valuations under heightened investor scrutiny.

The pro-AI camp focuses on the revenue trajectory. Theo, citing figures from the prospectus and subsequent reporting, pointed out that Anthropic went from $4.6 billion in annual revenue in 2025 to a reported run rate of $65 billion, describing it as nearly 15x growth in under a year. Accounts like Emma Jacobs and derangedHorse signaled enthusiasm for the stock, while Sipirtu flagged Anthropic's stated belief that AI will outstrip electricity and the internet as a sector benchmark. The anti-AI camp reads the same document as an indictment. James Malloy dismissed the company's mission in blunt terms, Stuart Gray argued the filing signals an ambition for cartel-like control over intelligence, and They Call Me Gary questioned whether a self-regulating company can be trusted to catch its own failures. The middle ground is more cautious than hostile. Robert Reich asked how worried investors should be about an AI bubble, curuinor argued that even transformative technology faces structural physical limits on returns, and Liz Fong-Jones offered a ground-level data point, noting her own spending on Anthropic's products jumped sevenfold after a pricing change.

The IPO pricing date and the reception from institutional investors will be the clearest near-term signal of whether the market accepts the growth story or discounts it for the losses and the existential risk language embedded in the prospectus itself.

Pro-AI7

What Pro-AI voices are sayingRevenue growth is the story, with figures showing nearly a fifteenfold increase in run rate within a year, and supporters see this as early-stage proof of a transformative bet worth making.

Quote 1 of 7

I will be doing the opposite.

derangedHorsevia Hacker News

Anti-AI310

What Anti-AI voices are sayingCritics see the financials as fundamentally broken, combining a $42 billion loss, over $500 billion in largely non-cancelable obligations, and a $2 trillion valuation target that many compare to a Ponzi scheme or speculative bubble. A notable minority finds irony in a company warning of existential risk to humanity as a headline pitch to investors.

Quote 1 of 13

Goldman Sachs warned Friday that AI hyperscalers, including Meta, need to generate roughly $300 billion a year in AI services revenue just to break even on their capital spending—and $1 trillion annually to earn meaningful profits—spooking...

DBdantvia Bluesky

Middle Ground34

What Middle Ground voices are sayingObservers note the tension between staggering losses and rapid revenue growth, and question whether the path to profitability is credible given the scale of infrastructure commitments. Some point out that much of the reported net loss reflects accounting treatment of fundraising instruments rather than pure operational spending.

Quote 1 of 13

How worried should we be about the AI bubble bursting?

Robert Reichvia Bluesky

Add your take

0 reader votes

More Pro-AI reactions (6)

“Gonna buy some of these stock things”
Emma Jacobs, Bluesky · 06:35 UTC
“We're still so early.”
Theo - t3.gg {bot}, Bluesky · 08:21 UTC
“Anthropic bets AI will outstrip electricity and the internet.”
Sipirtu, Bluesky · 07:18 UTC
“They made $4.6 billion in all of 2025. They made $11.5 billion from April to June 2026. Their run rate was last reported at $65 billion. That's almost 15x in under a year.”
Theo - t3.gg {bot}, Bluesky, skeptic · 11:06 UTC
“Their run rate was last reported at $65 billion. That's almost 15x in under a year.”
Theo - t3.gg [UNOFFICIAL], Bluesky · 10:52 UTC
“Anthropic's upcoming IPO could reshape the landscape for AI investments, despite a reported net loss of $42 billion last year.”
Veytics, Bluesky · 05:21 UTC

More Anti-AI reactions (12)

“Anthropic's IPO filing apparently warns investors that advanced AI could pose “catastrophic or existential risks to humanity,” which is a great sales approach and also, coincidentally, the exact phrase I was told when I brought home this guy.”
dave wagner, Bluesky · 23:07 UTC
“$252 billion in non-cancelable compute obligations across Microsoft, Google and Amazon, and $161.2 billion in Broadcom TPU lease obligations that are "largely non-cancelable". Truly insane.”
Ed Zitron, Bluesky · 16:52 UTC
“Anthropic’s newly unveiled IPO prospectus warns that its technology poses an “existential risk to humanity,” after a summer of hacking incidents where rogue agents broke containment and hacked business and government systems.”
WIRED, Bluesky · 11:57 UTC
“47% of Anthropic's 2025 revenue ($2.16bn) came from sales of its models through Amazon and Google, with Anthropic paying $351m in revshare (around 16.25%) and booking it as a sales and marketing cost, inflating its revenues by billions.”
Ed Zitron, Bluesky · 23:14 UTC
“it is weird to me that you can design a company such that the majority owners by value are not the majority owners by voting rights”
Pwnallthethings, Bluesky · 00:37 UTC
“Anthropic is losing $42 billion dollars a year and the target valuation is over $2 trillion. If that whets your appetite, feel free to jump in on meme coins from Hunter Biden and Hawk Tuah and maybe you’d be into”
Jared Yates Sexton, Bluesky · 01:16 UTC
“Anthropic says AI may pose 'existential risks to humanity' in IPO filing”
Edmund Lee, Bluesky · 12:36 UTC
“why has Reuters seen the prospectus and do we even have an SEC anymore?”
Ann M. Lipton, Bluesky · 23:47 UTC
“"What are you basing it on?" "We lost 42 billion dollars last year"”
Comfortably Numb, Bluesky · 01:42 UTC
“And if they don’t make a trillion dollars or so on the IPO they’re out of runway.”
Courtney Milan, Bluesky · 00:58 UTC
“If Anthropic would like to give me $12.6 billion in order to receive $4.6 billion back, I'd like to make it known that I have a gap in my schedule that could work well for both of us.”
The Author, Séamas O'Reilly, Bluesky, skeptic · 08:27 UTC
“Nearly 80 pages of the company’s 261-page initial public offering prospectus is dedicated to potential risk factors, while only 48 pages outline its business.”
Forbes, Bluesky · 03:45 UTC

More Middle Ground reactions (12)

“I don't understand why the article and the headline are based exclusively on those 2025 numbers, with not so much as a hint to the reader that there are figures from the past 9 months that aren't covered by the”
simonw, Hacker News, skeptic · 00:47 UTC
“LLMs profoundly change society -The LLM business is mangled in terms of ROIC vs CoC Such a business already exists - airlines.”
2ss, Hacker News · 23:52 UTC
“Future cloud obligations: $518B 🤯”
Dare Obasanjo, Bluesky · 01:35 UTC
“Overpopulation, Peak Oil, and The Coming War With Japan were existential risks if you were in the pundit class, if everyone didn't die of viruses brought back from space first.”
Calliope, Bluesky, skeptic · 13:51 UTC
“the cost to get there will be staggering. Anthropic reported a net loss of $42 billion in 2025, and plans to spend $518 billion on cloud, computing and infrastructure obligations in coming year.”
DiveDoc, Bluesky · 00:40 UTC
“Pay close attention when a company as valuable as Anthropic puts such a profound warning in their IPO filing…”
@Boom-WhatsGoingOn.bsky.social, Bluesky · 14:47 UTC
“🔊 Anthropic is warning investors about potentially catastrophic AI risks as it prepares for a blockbuster IPO. Today’s Reuters Morning Bid podcast looks at the tension between AI’s risks and its soaring valuations reut.rs/4xUX76y”
Reuters, Bluesky · 10:28 UTC
“So it's... basically a private company but with extra steps?”
Oakmoss 🌲🐀🦝🌿, Bluesky, skeptic · 00:39 UTC
“Despite a $2 trillion valuation ahead of its IPO, Anthropic hasn't been a money-spinning operation so far.”
Engadget, Bluesky · 10:07 UTC
“Guys I am beginning to think my credit card might have an Alignment Problem”
Brendan O'Connor, Bluesky · 19:15 UTC
“I don't think ROIC on a company that creates superintelligence is infinite. Nothing can defy physical law, nothing can teleport, nothing can reduce the universe's entropy, nothing can go faster than C, so there are structural limits keeping everything finite.”
curuinor, Hacker News, skeptic · 21:42 UTC
“I'd skip the $518B and go straight to the purchase obligations footnote once the S-1 is public. How much of it can Anthropic actually walk away from?”
olgava, Hacker News, skeptic · 09:03 UTC