Anthropic vs. OpenAI: Who Chose the Right Strategy in the AI Race?

Following the sell-off in the tech sector, potential investors in Anthropic have begun to reassess the company. Before the crash, many considered a $2 trillion valuation justified, but now it has fallen to $1.2–1.3 trillion; there are also those in the market who believe a fair value of $1.1 trillion is appropriate. OpenAI, which had long lagged behind its competitor, was valued at $1.25 trillion following its latest funding round. Thus, the valuations of the two main rivals have practically evened out, although their strategies differ significantly. Mikael Gorsky, an AI researcher and lecturer at the Holon Institute of Technology, explains why it is impossible to definitively identify a potential leader in this race.

In May 1956, Mao Zedong declared: “Let a hundred flowers bloom; let a hundred schools of thought contend.” The leader of the Chinese Revolution wanted to encourage Chinese writers, scholars, and engineers to openly debate and propose different solutions, rather than wait for the one “correct” answer from above, believing that a diversity of opinions would help the country develop more quickly.

This formula can also be applied to contemporary scenarios for the development of technology companies. In the marketplace, just as in science, there is no single right path to success, and the largest companies of recent decades have achieved it in very different ways.

Anthropic’s S-1 prospectus, published in late October, and a subsequent supplement to it showed that the company’s business is doing very well. In the first half of 2026, its revenue totaled $16.3 billion, with nearly $17 billion generated in the third quarter alone.

According to analysts’ calculations based on the prospectus, the gross margin for Anthropic’s main product—tokens—is approximately 85%. Just a year ago, it was around 12%. Token prices are falling, but the cost of production is falling even faster. Customer demand is clearly not saturated.

There is no official valuation yet for the IPO that Anthropic expects to conduct. In its most recent private funding round in May, it was valued at $965 billion. For several months, the press discussed a valuation of $2 trillion, but after the market’s dramatic crash in the first week of November, investors became more cautious. The Financial Times reported that Anthropic’s target valuation range has dropped to $1.25–$1.75 trillion. According to Bloomberg , it could be $1.2–1.3 trillion, and some potential investors are even considering a valuation as low as $1.1 trillion.

How much did Anthropic spend to reach its current position? According to the prospectus, as of September 30, 2026, the company had raised a total of $141 billion and held $115 billion in cash, cash equivalents, and short-term investments.

Operating cash flow for the first nine months of the year was positive, totaling $6.6 billion. Of the $104 billion net loss for the nine-month period, $94.3 billion is attributable to the revaluation of convertible notes and other financial liabilities; since 2023, such non-cash losses have accumulated to more than $150 billion. However, these losses are growing as the value of the company itself increases.

In 2025, Anthropic's operating expenses totaled $12.6 billion, with revenue of $4.6 billion and an operating loss of $8 billion.

OpenAI is also preparing for an IPO (in 2027). As reported by The Information on Friday, November 6, the company plans to hold a meeting with investors to discuss its third-quarter financial results. In its latest funding round, which closed on the same day, it was valued at $1.25 trillion, Reuters reported.

Since 2015, OpenAI has raised approximately $180 billion. In 2025, its expenses totaled $34 billion, with revenue of $13 billion and an operating loss of $21 billion.

In other words, to keep up in the AI race, the company spent more than its closest competitor.

There is a widespread view in the market that OpenAI's original ambitions to dominate the AI sector have clearly gone unfulfilled, and a number of analysts consider the company less successful than Anthropic.

For example, journalists at The Information reported that OpenAI is trying to close the gap with its competitor in the enterprise segment by positioning itself as a cheaper and more flexible provider of AI models. And this year, OpenAI even released the cheapest models (based on the cost of performing tasks in tests) compared to products from Google, Anthropic, and even its own. “But the publication of Anthropic’s prospectus shows that OpenAI will have to do much more to significantly shift the balance of power,” The Information’s newsletter stated.

I disagree with that point of view, and here's why.

Over the past 30 years, we have seen a flourishing variety of success stories in the technology industry.

Both the hyper-focused Amazon—a super-champion in logistics and cloud technology—and the “opportunistic” Alphabet holding company (Google’s parent company) can be successful. In addition to internet search and advertising, Alphabet is involved in autonomous transportation (Waymo), a mobile operating system (Android), and, of course, artificial intelligence (Google DeepMind). Both companies are worth trillions of dollars, and both have reached this point through completely different paths.

The AI developer is clearly focused on producing the best language models tailored for practical applications, and programming has become one such application. The business focuses on its core product, Claude Code, and on token sales. Corporate clients account for about 80% of the company’s revenue.

Its flagship product, ChatGPT, is used by 1.2 billion people every week, but most of them use the free plan. The company is building a whole range of initiatives around ChatGPT: it is developing its own agent-based programming system, Codex; a medical service, ChatGPT Health; and advertising within ChatGPT. By August 31, 200 days after its launch, advertising had brought OpenAI its first billion dollars in revenue.

In May 2026, OpenAI established a separate company, DeployCo, which implements AI for corporate clients using its own engineers.

Together with Broadcom, it has developed its own Jalapeño chip; it is working with former Apple chief designer Jony Ive to develop its first consumer device; and it is building data centers as part of the Stargate project.

In addition, OpenAI continues to acquire startups and made several acquisitions in 2026 alone. Like Alphabet, OpenAI is making many bets at once, and not all of them need to pay off for the company to remain successful.

The two companies have different strategies, different business metrics, and different political and social stances. Anthropic, for example, is in a legal dispute with the U.S. Department of Defense, which has identified the company as a risk to the supply chain. OpenAI, on the other hand, is collaborating with that agency.

Nevertheless, the companies share the same core objectives. Both must achieve sustainable profitability and convince investors of their long-term potential for rapid growth. Anthropic is already generating a positive cash flow, while OpenAI expects to do so by 2030.

So far, in my view, the situation looks promising. David Friedberg, a guest on the All-In podcast, demonstrated in early November how money is distributed throughout the AI value chain. His calculation starts with a $1 production cost for Nvidia hardware. That dollar generates $4 in revenue for Nvidia, with a gross margin of 75%.

The equipment is installed at the SpaceX AI data center, and SpaceX AI uses these chips in its data centers and leases computing power to customers. For this, it earns $8.60 per year (for every dollar of the equipment’s cost from Nvidia) with a margin of about 83% after depreciation and electricity costs.

At these capacities, Anthropic generates $17.8 in annual revenue for every initial dollar of cost of Nvidia chips. Its gross margin on customer services is 85%, and 52% after accounting for all computing costs.

The last figure is the author's estimate, but the other links in the chain are based on the companies' financial statements. Every participant in the chain enjoys excellent returns.

The AI labs themselves are taking it a step further: according to Epoch AI’s estimates, Anthropic generates about $9 million in revenue per employee per year, which is significantly more than that of the largest publicly traded tech companies. OpenAI ranks second, with $5.6 million per employee.

Revenue at the largest tech companies is growing rapidly, while their workforce is shrinking. Between 2021 and 2025, Amazon’s revenue increased by $247 billion, while its workforce decreased by 32,000. In fiscal year 2026, Microsoft had 5,000 fewer employees than the previous year, while revenue grew by 18%. Meta’s revenue per employee rose by 89% over the three-year period starting in 2022.

It's hard to say that OpenAI has lost this race. The history of the tech industry shows that both approaches can be successful.

Both companies are heading toward trillion-dollar valuations by different paths. Anthropic is banking on specialization and efficiency, while OpenAI is banking on scale, audience, and diversification. As long as demand for computing power and AI services continues to grow, both strategies may prove viable. The key question is whether AI can deliver real economic value to the companies that adopt it, and how large the market that Anthropic and OpenAI are targeting will ultimately be.

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