Anthropic’s IPO Math Leans on a $190 Billion 2028 Revenue Forecast
Bankers and investors are reportedly looking two years ahead to value Anthropic, exposing how much of a potential blockbuster IPO rests on exceptional growth assumptions.
Anthropic’s prospective initial public offering is forcing Wall Street to value a fast-growing artificial-intelligence company using forecasts much further into the future than is customary. Reuters reported that the company projects roughly $190 billion to $200 billion of revenue in 2028, citing two people familiar with its finances. Anthropic did not comment on the figures.
The projection is striking even by the standards of the current AI investment cycle. Anthropic said its revenue run rate exceeded $47 billion in May, up from about $9 billion at the end of 2025. Reuters reported that the company had projected at least $10.9 billion of revenue for the second quarter of 2026 and its first quarterly operating profit, at $559 million. Moving from that base to nearly $200 billion of annual revenue in 2028 would require rapid adoption to continue while enterprise spending, compute supply and product pricing all remain supportive.
According to four people cited by Reuters, bankers and investors are applying enterprise-value-to-revenue multiples to forward estimates as they prepare for Anthropic’s analyst day. Revenue multiples are common for software businesses whose profits have not matured. What is less conventional is looking two years ahead to find a base large enough to support the valuation under discussion.
The approach reflects both Anthropic’s speed and the limitations of conventional earnings measures. Training frontier models, serving inference workloads and hiring specialized researchers consume enormous amounts of capital. Current operating profit can therefore understate the margins investors expect if usage scales faster than compute and personnel costs. It can also understate risk if competition drives model prices down or if the cost of supplying additional intelligence remains stubbornly high.
Potential reference points reportedly include Palantir, Cloudflare and SpaceX. LSEG data cited by Reuters placed Palantir at 53 times expected 2026 revenue, while Cloudflare and SpaceX traded at about 41.6 times expected revenue. None is a clean comparison. Palantir is established enterprise software, Cloudflare operates a distributed internet network, and SpaceX combines launch services, satellite communications and large physical assets. Anthropic’s economics depend on model performance, enterprise contracts, developer usage and access to computing infrastructure.
That mismatch is precisely why the 2028 forecast matters. A valuation built on distant revenue combines two judgments: that Anthropic can reach the forecast and that investors will still award a high multiple when it does. Small changes in either assumption create very large changes in enterprise value. A lower growth rate, heavier inference costs, more aggressive competition or multiple compression could materially reduce what public investors are willing to pay.
The IPO process will also test how much financial disclosure the market requires from frontier AI companies. Investors will want to separate contracted revenue from usage-based run rates, identify customer concentration, understand the duration and price of compute commitments, and see how gross margins change as models become more capable. Capital expenditure borne directly by Anthropic must be assessed alongside capacity financed or supplied by strategic cloud partners.
For the wider startup market, an Anthropic listing could set a valuation template. Private AI companies have raised capital against expectations of unusually fast revenue growth and very large addressable markets. A successful offering would provide a public benchmark and potentially reopen exit routes for venture funds. A difficult pricing process would force private investors to revisit forecasts and the premiums attached to AI exposure.
The report is best treated as a view into valuation work, not a company-issued forecast or a confirmed IPO price. Reuters based the revenue figure on confidential sources, Anthropic did not respond, and no final valuation was reported. Forecasts prepared for bankers can change before a filing, while market conditions and comparable-company multiples can move sharply.
Why it matters
Anthropic’s IPO could become one of the largest tests yet of whether public markets will underwrite frontier AI on distant revenue rather than current cash generation. The outcome will influence venture valuations, compute financing and the public-market appetite for other AI issuers. The essential question is not whether Anthropic is growing quickly; it is whether growth can remain exceptional long enough for revenue to outrun the cost of building and operating its models.