Anthropic co-founder Daniela Amodei used a Bloomberg Tech conference appearance Thursday to defend the economics of frontier AI, hours after the company confirmed it had filed confidentially for an IPO at a $965 billion valuation. Annualized revenue crossed $47 billion in May, up from roughly $9 billion at the end of 2025 — a more than fivefold jump in five months that Amodei used to brush off questions about whether enterprise AI spending will hold. The pitch to public markets is straightforward: training and serving frontier models costs more than private capital can comfortably absorb, and the listed market is the natural next pool.
The IPO filing follows a $65 billion private fundraise announced last week that multiple investors described as greatly oversubscribed. That demand is the backdrop for the public-markets move, not a reason to avoid it. Amodei framed the listing as a structural requirement of the business rather than a financing tactic.
The revenue trajectory is the headline number that underwrites the valuation. Going from $9 billion to $47 billion in annualized run-rate inside two quarters puts Anthropic on a pace that, if sustained, would make it one of the fastest-growing software businesses on record. Whether that pace holds is the open question — and the question Amodei was repeatedly asked.
Key facts
- 01Anthropic annualized revenue hit $47B in May 2026, up from roughly $9B at the end of 2025.
- 02The company's $65B private fundraise at a $965B valuation last week was greatly oversubscribed.
- 03Anthropic filed confidentially for an IPO, with co-founder Daniela Amodei citing capital needs for training and inference.
- 04A compute deal with xAI disclosed in SpaceX's S-1 will cost Anthropic $1.25B per month.
- 05Unlike OpenAI and xAI, Anthropic is not building its own data centers.
The skeptical case has been getting louder across the customer base. Uber has said publicly that while parts of its AI spending have produced returns, not all of it has, raising the prospect that corporates start to throttle AI budgets and cool the sector's growth. Amodei's response is that enterprise adoption is still early, and the productive use cases — coding, financial services, legal, healthcare — will broaden as buyers learn the tools.
That argument needs to land with public-market investors who will price the stock on forward revenue assumptions, not on the May run-rate. The bull case is that Anthropic's growth reflects genuine enterprise pull and that today's named verticals are the wedge into broader workflow integration. The bear case is that a chunk of the $47 billion is experimentation budgets that get cut in the next downturn.
On the cost side, Anthropic is taking a different path than its two largest rivals. OpenAI and Elon Musk's xAI are building their own data centers; Anthropic is not. Amodei said the company prefers slightly underbuilding capacity to overbuilding it.
The practical consequence of that posture showed up last month, when Anthropic partnered with xAI for compute capacity in a deal later disclosed in SpaceX's S-1 filing at $1.25 billion per month. That is a striking figure on its own and a strategically awkward one given xAI's positioning as a direct model competitor. It also illustrates how thin the supply of frontier-grade compute remains: Anthropic will pay a competitor $15 billion annualized rather than build the capacity itself.
Anthropic now reaches the public-markets gate ahead of OpenAI, which has signaled its own IPO ambitions but has not filed. The order matters. The first frontier-model IPO will set the comparable that every subsequent listing is measured against — on revenue multiples, on gross margin disclosures, on how the S-1 narrates the relationship between training capex and inference economics.
The risk to the story is concentration. A revenue base growing this fast typically has heavy customer concentration and heavy product concentration, and the S-1 will have to disclose both. Investors will want to see how much of the $47 billion run-rate comes from a small number of enterprise contracts, how much from API usage by AI-native startups, and how much from coding products specifically. The compute-cost line, with the $1.25 billion monthly xAI commitment included, will also draw scrutiny.
For the broader AI market, Anthropic going public on these numbers changes the reference frame for every private valuation in the sector. A $965 billion mark on $47 billion of annualized revenue implies a multiple that only works if growth stays close to current pace and gross margins on inference improve materially. If the IPO prices well and trades well, it pulls forward IPO calendars at OpenAI, xAI, and the next tier down. If it stumbles, the private market resets faster than anyone in the current funding cycle is pricing for — and the $1.25 billion monthly compute bill becomes a lot harder to underwrite.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
