Anthropic Targets $200B 2028 AI Revenue, but Can Margins Follow?

For the AI community, Anthropic's 2028 revenue projection is not just a financial story; it is a hypothesis about how quickly foundation-model businesses can convert research leadership into production revenue. Reaching $190–200 billion would require sustained enterprise adoption of Claude, API expansion, and premium pricing across a compute-heavy delivery model. The projection raises the stakes for every AI lab: can frontier model developers outgrow the cost of training, inference, and hiring fast enough to deliver software-style operating leverage?

Anthropic, the San Francisco-based AI company behind Claude, is reportedly anchoring its IPO valuation to a 2028 revenue projection of $190 billion to $200 billion, a figure that has not previously been disclosed and that dramatically reshapes how public-market investors are being asked to underwrite the company. The projection, attributed to two people familiar with its financials, compares with a $47 billion revenue run rate the company publicized as recently as May. Because the information comes from unnamed sources and Anthropic did not immediately respond to a request for comment, the forecast should be treated as an unconfirmed datapoint rather than official guidance. Nonetheless, the reported numbers offer a rare window into the assumptions circulating ahead of what could be one of the largest IPOs on record.

“

The company's revenue run rate was about $9 billion at one earlier marker, and by May it had publicized a $47 billion run rate.

The central valuation challenge is not whether Anthropic is growing quickly—it clearly is—but how far into the future buyers are willing to price that growth. Normally, bankers value fast-growing software businesses on enterprise value-to-revenue multiples against near-term forecasts, often the next twelve months. Here, according to four sources, bankers and investors are relying on EV/revenue multiples based on much longer-dated forecasts. The comparables illustrate the arithmetic: Palantir trades at 53 times expected current-year revenue, while SpaceX and Cloudflare each trade at 41.6 times expected 2026 revenue, according to LSEG data. Applying comparable multiples to Anthropic's possible 2028 revenue of $200 billion would produce valuations far beyond most current public market capitalizations, even if growth execution risk warranted a discount.

This long-horizon pricing is not unprecedented in frontier-tech deals. Cerebras Systems reportedly cited 2028 revenue expectations before its IPO this year, and SpaceX projections extended to 2029 before its June public offering. Those precedents suggest investors may increasingly accept a venture-style, forward-story approach in IPOs for infrastructure-heavy AI companies, but they also reveal why the method is controversial. A revenue multiple on 2028 numbers requires buyers to assume that Anthropic can sustain extraordinary growth for years while absorbing compute, model training, and hiring costs that currently keep margins from reflecting mature software economics. The bet is that revenue grows faster than the cost of computing power and talent, allowing operating leverage to emerge as the company scales.

The financial trajectory itself underscores the scale of that bet. The company's revenue run rate was about $9 billion at one earlier marker, and by May it had publicized a $47 billion run rate. If the 2028 projection is accurate, Anthropic would need to more than quadruple that May figure in roughly two and a half years. That acceleration would require not just expanding enterprise and consumer adoption of Claude but also maintaining premium pricing, broadening distribution through cloud partners, and converting AI experimentation into durable workloads. The capital requirements are equally significant: the compute capacity needed to support that revenue could run to tens of billions in annual infrastructure spending, making gross margin trajectory as important as top-line growth.

For bankers and early-stage backers, the incentive to emphasize 2028 revenue is clear. A company spending heavily on AI can look expensive on trailing or 2026 metrics but reasonable on 2028 if growth continues. That framing, however, shifts risk to public-market investors, who would be underwriting a period long enough for new competitors, open-weight models, pricing pressure, or regulation to disrupt the trajectory. If cloud providers or major enterprises accelerate their own in-house AI platforms, Anthropic's forecast could prove too optimistic even if the AI market expands. The IPO's eventual pricing will therefore test whether public investors accept a valuation constructed less on current results and more on a still-unproven 2028 income statement.