Anthropic's IPO filing slipped to late September, with bankers discussing up to $2T
The company filed confidentially on 1 June. The public S-1 has moved to late September and the roadshow to mid-October, on a valuation case that rests on 2028 revenue of $190bn to $200bn.
Anthropic's public IPO filing has moved to late September and its investor roadshow to mid-October at the earliest, pushing back a listing that bankers and investors have discussed valuing at up to roughly $2 trillion.
The company submitted a draft registration statement on Form S-1 to the Securities and Exchange Commission confidentially on 1 June. As of late August no public S-1 or amendment had appeared on EDGAR, which means no prospectus, date, ticker, exchange, share count or price range is confirmed.
The last private mark was a $65 billion Series H at a $965 billion post-money valuation. A $2 trillion listing would be slightly more than double that, and it rests on projected 2028 revenue of $190 billion to $200 billion.
The number the valuation depends on
The revenue trajectory underneath is genuinely steep. Anthropic's annualised run rate crossed $47 billion earlier in the year and reportedly exceeded $65 billion by the end of July, driven by enterprise adoption of Claude for coding and agentic workflows. Ramp data cited by TechCrunch put Anthropic at about 43.5 percent of US enterprise AI spending in August, ahead of OpenAI at 39.7 percent.
Getting from $65 billion to $195 billion by 2028 requires roughly tripling in two and a half years. That is slower than the company's recent growth, so the projection is not obviously heroic on its own terms. What it assumes is that enterprise AI spending keeps compounding, that Anthropic holds its share of it against OpenAI and a Google-Accenture push into deployment, and that pricing does not compress.
Pricing is the weak assumption. Anthropic cut cached-context costs by 75 percent with Fable 5.1 on 1 September and put typical workloads about 25 percent cheaper than the prior generation. Per-token prices are falling across the industry, so revenue growth has to come from volume outrunning deflation — and volume growth is precisely what a slowdown in enterprise experimentation would remove.
The costs on the other side
Anthropic has committed to compute at a scale that will be visible in the prospectus. It signed a $35 billion cloud agreement with Nvidia-backed Lambda in early September, following a reported $45 billion contract with Nscale and a $10 billion agreement with Volta.
Those are multi-year obligations that do not flex if revenue growth disappoints, and they will sit in the risk factors next to a revenue projection that has to nearly triple. Public investors will read them together in a way private investors did not have to.
The political footing
The listing arrives with the company's Washington position improved. Commerce Secretary Howard Lutnick said on 2 September that "we trust Anthropic," a reset after disputes over export controls, and the company has been positioning its safety architecture — Enterprise Frontier Safeguards, the restricted Mythos tier, its published alignment research — as commercial differentiation rather than cost.
Against that, Congress is considering a Sanders-Casar bill that would pause frontier research, authors are contesting the allocation of an existing Anthropic copyright settlement, and the company's own disclosures describe models escaping sandboxes. Every one of those becomes a risk factor the moment the S-1 goes public.
Anthropic has not commented on the timetable.
Runs the newsroom. Rename this profile in the studio to your own byline.
Related
Every weekday, the AI stories that moved money or shipped code.
No cross-posting, unsubscribe anytime. See all newsletters