Anthropic is pushing its initial public offering — potentially the largest in history — to November, and weighing whether to release a new model before listing, as it responds to the momentum OpenAI has built with GPT-6 Astra this month. Citing Reuters, The Wall Street Journal and The New York Times, Chinese outlets reported that Anthropic had originally planned to start its roadshow as early as mid-October; the target is now after the U.S. midterm elections.
The stakes are hard to overstate: investors have floate
[1][2]d a valuation near $2 trillion and a raise of up to $100 billion, both of which would surpass the records SpaceX set in its June listing (roughly $1.77 trillion valuation, $75 billion in regular financing). At $2 trillion, it would be the largest IPO on record, ahead of Saudi Aramco, Alibaba and Visa.
The stated reason for the delay is timing. Several people familiar with the matter said advisers argued that moving the offering to November would let the company publish third-quarter results first, showing investors its competitive position still holds after GPT-6 Astra shipped. Notably, those people said the decision was made before former Anthropic researcher Jacob Coxon's public resignation warning touched off a broader debate about whether AI is moving too fast — the schedule change and the safety argument are not directly linked.
What is genuinely awkward is the company's double narrative. Even as investors prepare to pay for a $2 trillion story, CEO Dario Amodei published a roughly 3,800-word essay on September 12 urging the industry to slow the pace of releases, arguing the risks of frontier tools are too large to continue at current speed. OpenAI's Sam Altman and SpaceX's Elon Musk echoed similar positions. Silicon Valley's response: several firms heavily invested in Anthropic told reporters they think Amodei's push for industry safety standards is well timed and will not hurt IPO interest; one investor argued that staking out a position on safety early could protect the company once public shareholders scrutinize it more closely.
The numbers remain strong. Bloomberg had reported annualized revenue run-rate above $65 billion by the end of July, up from roughly $9 billion at the end of 2025; people familiar with the matter told The New York Times the company expects annualized revenue above $100 billion this year, and existing investors project over $110 billion by the end of 2026. But the pressure is real: spend-management platform Ramp attributed about 13% of tracked enterprise AI spending to GPT-6 Astra versus roughly 8% for Claude Fable, and OpenRouter's data last week showed spending on OpenAI models exceeding Anthropic's for the first time in over two and a half years.
There is also the rival's shadow. OpenAI has said it will not go public before 2027 and is in early talks on a new round that could value it above $1.2 trillion; it raised over $120 billion in its last round. Some Anthropic investors warn that a similar-scale OpenAI raise before Anthropic's listing could drain market liquidity and divert demand. Meanwhile both companies are piling into compute: investors expect Anthropic to have about 5 gigawatts of capacity by year-end, roughly doubling by the end of next year, matching OpenAI. At a September 17 investor event at Anthropic's headquarters, Jared Kaplan, Benjamin Mann and Andrej Karpathy demonstrated products including Model Hardware Standard, which lets agents operate physical devices such as microscopes and robotic arms.
The people familiar stressed that every plan could still change with investor sentiment and market conditions. The company now faces two stories at once: a $2 trillion story for the market, and a "we should slow down" story for the industry. Which one breaks first is worth watching more closely than the IPO date itself.
[1][2]