How was the $900 billion Anthropic built?

Author: Su Yang, Tencent Technology

Bloomberg, citing sources familiar with the matter, reported that AI startup Anthropic is in early-stage talks with investors, aiming to raise at least $30 billion in new funding, valuing the company at over $900 billion.

Sources revealed that this round of financing is expected to be completed as early as the end of May 2026, however, the deal has not yet been finalized, and no term sheet has been signed.

The sources revealed that this round of financing is expected to be completed as early as the end of May 2026, however, the deal has not yet been finalized, and no term sheet has been signed.

If the funding round is successful, Anthropic will not only surpass OpenAI (valued at $852 billion in March) but also challenge the market capitalization of tech giants like Apple and Microsoft. It's worth noting that early investors have largely chosen to remain on the sidelines this round. $30 billion in annual revenue and a 40% gross margin: Why can a company's valuation surge 15 times in 14 months? The answer seems obvious: growth rate. According to publicly reported data, Anthropic's annualized revenue will soar from $1 billion in December 2024 to $30 billion by the end of March 2026. This means that it has maintained a growth rate of over 10 times for several consecutive years. This growth curve is likely unprecedented in the history of enterprise software. Eight of the top ten companies on the Fortune Global 500 list are Anthropic clients. Over 1,000 enterprise accounts spend more than $1 million annually on Claude. In particular, its developer-facing coding product, Claude Code, achieved annualized revenue of $2.5 billion by February 2026, following its launch in May 2025, with enterprise subscriptions quadrupling in the first six weeks of the year. Based on a $900 billion valuation and $30 billion in annualized revenue, the price-to-sales ratio is approximately 30. This multiple sounds extreme, but proponents are betting on the future. They argue that a company growing tenfold annually cannot be valued in the conventional way. Their pricing logic assumes it will maintain a similar compound growth rate until 2028, at which point the current valuation will appear reasonable. Regarding Anthropic's revenue, competitor OpenAI raised its own questions. They believe Anthropic's reported $30 billion annualized revenue uses total revenue accounting, meaning that when customers use its models through platforms like Amazon Web Services and Google Cloud, it records all end-user consumption as revenue and only lists the fees paid to the cloud platforms as expenses. OpenAI estimates that after deducting these intermediary fees, Anthropic's true annual revenue is closer to $22 billion. This $8 billion difference is purely a matter of methodological choice, but it will become a focus of scrutiny from the market and regulators during the IPO. More noteworthy than revenue figures are the costs. According to data, Anthropic plans to spend approximately $19 billion on training and inference computing in 2026—a figure almost equivalent to its annual revenue. More problematic is that due to inference costs exceeding expectations by 23%, its gross margin has been compressed to approximately 40%, a level far lower than most established enterprise software companies. Anthropic is currently not profitable and is not expected to turn a profit until 2028. For a company with a valuation approaching one trillion dollars, this combination of financial metrics is indeed unusual. Valuation-Driven Computing Arms Race

Why is Anthropic raising so much money?

Nominally for development and expansion, but in reality, the vast majority of this $30 billion in funding is to pay for the computing infrastructure it has promised but has not yet built. This seems to be a completely different model from traditional fintech.

In the past, startups raised funds to refine their products and expand their markets, and then gradually matched their valuations with growth. But in the AI ​​era, startups need to raise funds at extremely high valuations first, using this money to lock in massive future computing power, and then hoping that this computing power will drive leaps in model capabilities, thereby bringing revenue growth, and ultimately proving that the high valuation is justified.

It's like a chicken-and-egg dilemma. Now, valuation drives computing power commitments, which in turn require an even higher valuation to pay for, accelerating the cycle. Anthropic is the ultimate embodiment of this model. Once this cycle starts, it's hard to stop. It can propel a company to the cloud, or drag it into the abyss in an instant.

In early 2026, Anthropic's valuation soared to $380 billion

Anthropic's CEO, Dario Amodei, told Fortune magazine just days after the company completed its previous $30 billion funding round that Anthropic would go bankrupt if progress in artificial intelligence was delayed by 12 months.

Anthropic's CEO, Dario Amodei, told Fortune magazine that if progress in artificial intelligence were delayed by 12 months, Anthropic would go bankrupt.

For a company valued at $900 billion, the distance between "extraordinary success" and "operational bankruptcy" may be just a few bad quarters. This tense balance may be the reason why sensitive early investors have largely refrained from participating in this round. Early investors are collectively adopting a wait-and-see approach. According to Forbes, some of Anthropic's early backers—those who entered at a valuation of $4.1 billion in 2023 or $61.5 billion in March 2025—have shown almost no interest in participating in this round. The reason is simple: bankers privately predict that if Anthropic goes public as early as October 2026, its public market valuation could fall between $400 billion and $500 billion. This means that if someone invested in the final round of private placement at a valuation of $900 billion, theoretically, the investment would already be at a loss before the shares were even released from lock-up and traded. This kind of inversion—a late-stage private placement valuation significantly higher than the expected IPO valuation—is extremely rare in the history of tech financing. It signals either that the company is severely overvalued in the private market or that the public market will give it a drastically different price. Either way, it's fraught with uncertainty. The decisive event that was about to unfold was the IPO itself. We previously mentioned the key figure behind Anthropic's IPO and financing—the company's CFO, Krishna Rao. The Information reported that at the time, Anthropic's computing power was essentially dependent on Google. Rao felt this wouldn't work; they couldn't put all their eggs in one basket. He pushed for a new strategy internally and among investors: computing power providers had to be diversified. According to The Information, citing sources familiar with the matter, Rao had in-depth discussions about this strategy with Byron Dieter of Bessemer Venture Partners, one of the investors. Dieter later commented that it was Rao who made the company realize that finding more partners would accelerate development. Looking back now, Anthropic acted even faster than OpenAI. They have already signed deep agreements with the three cloud computing giants: Amazon, Google, and Microsoft. At the chip level, they have also incorporated Nvidia's GPUs, Google's self-developed TPUs, and Amazon's chips, forming a diversified supply network. But simply signing agreements isn't enough; the core is ensuring that suppliers actually provide the computing resources. By the end of 2025, Rao spearheaded two major deals: one a whopping $30 billion to use Microsoft's cloud servers to run Nvidia chips; the other a contract for up to 1 million Google TPUs. By early April 2026, Anthropic had gone a step further, reaching new agreements with Broadcom and Google to secure power supply capacity for several gigawatts of data centers. These actions were no longer simply "purchasing" computing power, but rather a large-scale "reservation" of future infrastructure. Since Rao joined, he has helped the company complete multiple rounds of financing totaling $60 billion. By January of this year, the company's valuation had risen to $380 billion. It can be said that, under Rao's strong impetus, Anthropic's computing infrastructure and financial resources have reached unprecedented levels. Is there a bubble? The answer will be revealed in six months. Following the current pace, if this round of financing is successfully completed, Anthropic is expected to seek an IPO between October 2026 and the first half of 2027. Goldman Sachs, JPMorgan Chase, and Morgan Stanley are reportedly discussing the matter. At that time, the market's focus will no longer be on "whether Anthropic can continue to grow," but rather on a referendum on the valuation logic of the entire AI industry: Have the private market's pricing methods for AI over the past three years been correct? The capital expenditure commitments of hyperscale enterprises, computing power reservation contracts spanning several years, 40% gross margins, the debate between the total revenue method and the net income method, and the accelerating cycle of "valuation-computing power-revaluation"—all these complex issues that can be blurred in the private market will be placed under the microscope of the public market during the IPO. If the public market is willing to give Anthropic a valuation of $1 trillion or even higher, then a $900 billion valuation entry price seems like a generous early positioning. However, if the market only offers $500 billion, then the last batch of private equity investors will be in a very awkward position. The third, and perhaps more far-reaching, possibility is that Anthropic's IPO will serve as a key data point, verifying or disproving the structural assumptions of AI finance. Remember Michael Burry, the real-life inspiration for the protagonist of "The Big Short"? He recently warned of a bubble in "tech stocks" and "chip stocks" in his paid column. If the assumptions of AI finance are disproven when Anthropic goes public, it will be the time when the bubble bursts. Therefore, for both Anthropic itself and the entire AI industry, which has become accustomed to soaring valuations over the past three years, the stress test has only just begun, and will soon be determined by a stock price chart, providing the most realistic and ruthless valuation.