A $30 Trillion Market and the Risk of an Apocalypse: How Much Could Anthropic Be Worth? — An FT Analysis

Debates over whether artificial intelligence could destroy humanity have, for obvious reasons, diverted investors’ attention from the question of how much AI companies are worth, according to the Financial Times. The publication attempted to analyze whether Anthropic’s projected valuation of $2 trillion has a financial basis or is merely a “financial hallucination.”

Anthropic may announce details of its stock offering in the coming weeks and begin trading in November, according to sources cited by The New York Times. This IPO could be the largest in history. The FT’s analysis centers on three approaches: multiples of projected revenue, the size of the potential market, and risks—ranging from the devaluation of AI models to doomsday scenarios.

It is relatively simple to value companies with a predictable business model: forecast earnings for the next year or two—or revenue, if the company isn’t profitable—and multiply that by a multiple comparable to those of similar public companies, the FT writes. However, in the case of Anthropic, this approach is of little use—even by Silicon Valley standards, the company is growing rapidly.

Anthropic was founded in 2021 and generated no revenue until 2023. By August, the company had reached an annual revenue run rate of approximately $65 billion (according to sources at CNBC and Bloomberg). This growth appears to have come as a surprise even to the AI lab itself—just 18 months ago, it had projected revenue of $12 billion for 2027.

Some investors expect Anthropic’s annual revenue to reach $320 billion by the end of next year, according to the FT. Investors aren’t impartial, but if they turn out to be right, a $2 trillion valuation would imply a price-to-revenue multiple of just seven for 2028. By comparison, SpaceX shares are trading at 16 times the company’s revenue for that year. Using SpaceX as a benchmark, Anthropic could eventually be worth $5 trillion, the publication claims.

Another approach to valuing Anthropic is based on the size of the total addressable market (TAM). This metric has become a staple of virtually every venture capitalist presentation or IPO prospectus, although its usefulness is questionable. For example, in 2019, when Uber went public, it estimated its potential market at $12.3 trillion. Currently, the company’s annual revenue is less than $60 billion. WeWork’s addressable market for office real estate was estimated at $3 trillion, but that did not save the company from bankruptcy.

In its IPO prospectus four months ago, SpaceX estimated the potential market for “enterprise applications”—which includes AI products—at $22.7 trillion. Hypothetically, if Anthropic were to capture just 3% of this market, its revenue would total nearly $700 billion, which—at a revenue multiple of 10 and discounted over three years—would result in a valuation of $4.5 trillion, according to the FT.

Anthropic may estimate the potential market at $30 trillion in its investor documents, The Wall Street Journal reported, citing sources. Meanwhile, analysts at Morgan Stanley—which is likely to underwrite the upcoming IPO—estimate the generative AI market to be twice as large. Let’s assume that AI ultimately captures 10% of the “knowledge work” and digital consumer spending market, which Morgan Stanley estimates at $60 trillion. If Anthropic can capture a third of that share, its revenue would reach $2 trillion. Using the same 10x multiple and a five-year discount period, this corresponds to a valuation of $10 trillion, according to the FT.

In a recent study, Anthropic estimated that, in an “extreme” scenario, AI could add more than $10 trillion to the U.S. economy by 2030. A rough calculation puts the company’s market capitalization at $100 trillion, according to the FT.

Anthropic and OpenAI are competing not only with each other but also with Alphabet, Meta, Nvidia, and Chinese developers for a share of AI revenue. The main risk is that, at the end of the race, AI architects will end up with only pennies. This could happen if their models become interchangeable—that is, if one model is virtually indistinguishable from another to the user, who will consequently choose the cheapest one, the publication notes.

There are already signs of a price war: Chinese companies Moonshot AI and DeepSeek are offering cheaper alternatives, Google and OpenAI are lowering prices on some products, and some customers are using “router” programs that direct tasks to the cheapest model.

Not everyone agrees that a price war poses a risk. Founders of AI companies generally argue that there will always be a premium for the best models, according to the FT. In addition, leading labs are striving to achieve “recursive self-improvement,” in which AI models will update themselves with virtually no human intervention. If this happens, the frontrunners in the race could gain a huge competitive advantage, the article states.

“In a sense, this is more of a risk than an opportunity,” writes the FT. “If a self-improving superintelligence were to emerge, it could indeed pave the way for economic growth of more than $10 trillion, as Anthropic believes. But it could also turn the world upside down in such a way that asset prices, as we know them today, would become meaningless.”

Another source of potential revenue growth is the ability of AI labs to displace traditional software-as-a-service (SaaS). The launch of Anthropic’s plugin for lawyers in January triggered what has been dubbed the “SaaSpocalypse”—a sell-off of software company stocks. Salesforce and ServiceNow lost a combined $300 billion in market capitalization in just a couple of days—a figure that illustrates the scale of the potential payoff for Anthropic, according to the FT.

The publication notes, however, that any valuation model for Anthropic or OpenAI would be incomplete if it did not take into account one of the most unusual risk factors—a potential apocalypse.

“It is clear that if AI were to actually lead to the extinction of humanity, all investments would lose their value,” writes the FT. “But less extensive damage—say, ‘hundreds of billions of dollars in losses,’ as Anthropic CEO Dario Amodei writes in a recent essay — could seriously undermine Anthropic’s valuation if the company is found to be at fault for such an incident, not to mention its reputation.”

On the other hand, the newspaper notes that a company that develops a sufficiently reliable and powerful AI-based defense against malicious attacks stands to make enormous profits. Following Amodei’s call to slow the pace of AI development, the stock prices of cybersecurity companies rose sharply.

“As paradoxical as it may sound, the fear of mass destruction could prove useful, even if it slows down the pace of work at all the labs. Since superintelligent AI will occupy less space in their minds, Anthropic and OpenAI will have far greater incentives to generate revenue and profits from existing technologies—and to spend funds more cautiously or direct them toward more targeted projects. This could make the $100 trillion valuation unattainable. “However, a valuation of $2 trillion seems more reasonable under this scenario,” the Financial Times notes.

This article was AI-translated and verified by a human editor