Anthropic's $2 Trillion Valuation Rests on Adjusted Math and 2028 Wishes

Anthropic has selected Nasdaq as the venue for an October IPO that could value the artificial intelligence startup at $2 trillion or more—a figure that would rank it among the most valuable companies on Earth, nestled somewhere between the entire GDP of the United Kingdom and the total market capitalisation of every automotive manufacturer combined. The company has also informed a select group of investors that it expects positive adjusted operating income for a second consecutive quarter. Both facts are true. Both facts are also deeply weird.

Start with the revenue picture, which is genuinely impressive by any measure. Anthropic's second-quarter revenue reached $11.5 billion, representing a 14-fold increase year over year. By the end of July, annualised revenue had climbed to $65 billion, a leap from $9 billion at the end of 2025. These are not the numbers of a company treading water. The top line is moving with the kind of velocity that tends to attract serious capital and serious scrutiny.

Then comes the valuation math, which is where things get interesting in the way that only financial projection games can be. A $2 trillion valuation at current revenue levels makes no sense—it implies a price-to-sales multiple of roughly 31 times the current run rate. So bankers have instead reached forward to 2028, applying a 10x revenue multiple to a projected $190 billion to $200 billion in annual revenue. That two-year projection is an unusually long reach. It says less about Anthropic's certainty than it does about the peculiar moment the AI market is inhabiting, where reaching further into the future has become the only way to justify reaching higher on the valuation.

But the real aperture onto the company's current situation opens when you examine the profitability claim. Anthropic is reporting positive adjusted operating income—emphasis on adjusted. The adjustment excludes stock-based compensation, which is to say it excludes the single largest expense item that most tech companies actually incur. It excludes the cost of the equity you need to hand out to keep engineers from leaving for OpenAI or Google or any of the dozen other places that will hand them eight-figure packages to build AI systems. Adjusted operating income is not, technically, a lie. It is simply a number that omits several categories of cost that happen to be quite large.

Gross margins above 80 per cent before accounting for revenue shared with distribution partners—including Amazon—and the cost of training models do suggest real operational leverage once you account for scale. But scale is the operative word. That $2 trillion valuation assumes not just that Anthropic will reach $190 billion to $200 billion in revenue by 2028, but that it will do so while maintaining the kind of capital efficiency that has historically eluded the entire technology sector.

Investors are asking a sharper question beneath all of this. Anthropic's business model exhibits a singular vulnerability: its revenue is concentrated almost entirely in AI applications, with no meaningful diversification. There is no enterprise software division. There is no cloud infrastructure play. There is no advertising network. There is, fundamentally, one bet. The company is closing in on a public valuation that would place it among the most valuable companies on Earth while remaining entirely dependent on a single revenue stream in a market that has not yet determined which vendor will actually win.

The market will price this IPO. It will do so by deciding whether Anthropic's path to $190 billion in revenue by 2028 represents either unprecedented operational excellence or unprecedented accounting creativity. The bankers will frame it as the former. The traders will price it as something in between. And the buyers—because there will be buyers, because this is 2026 and this is the AI moment—will discover which version proves correct when the adjusted numbers eventually stop adjusting and the real ones start mattering.