Founded five years ago and valued at $96.5 billion, Anthropic has just “submitted” its filing.

On June 1, Anthropic confidentially filed a draft S-1 registration statement with the U.S. Securities and Exchange Commission (SEC), officially kicking off the IPO preparation process.

Just two days earlier, on May 28, Anthropic had announced the completion of a $65 billion Series H round, valuing the company at $965 billion (approximately 6.54 trillion RMB). At the valuation level, it was the first time in history that it surpassed OpenAI (latest valuation $730 billion). On the same day, Claude Opus 4.8 was released.

This star company, founded in 2021 by former core members of OpenAI, is knocking on Wall Street’s door at an unprecedented pace.

And after learning that Anthropic had privately filed for a listing, OpenAI CEO Sam Altman said in an interview with CNBC, “OpenAI will go public when we think the time is right. I think there’s a race right now, with everyone vying to offer the top-tier technology and build the best companies, but you also know that going public is a form of fundraising. I don’t think we’re currently focusing on deciding the specific timing of going public.”

1 Confidential “Filing”

What Anthropic submitted this time is a confidential S-1 draft, not an official prospectus.

Under the SEC’s confidentiality filing rules, Anthropic’s submission allows it to conduct internal review and communicate with regulators before officially releasing financial data. In an official statement, it said that after submitting the confidential S-1 document, once the SEC completes its review, the company can move forward with the listing when the time is right. The specific listing timing and the amount of funds to be raised have not been disclosed. The execution schedule depends on market conditions and various external factors, with the earliest possible listing targeted for this fall.

Right on the eve of submitting the filing, Anthropic had just announced the completion of a Series H funding round totaling as much as $65 billion.

This round not only brought together top-tier financial investors such as Altimeter Capital, Sequoia Capital, and Greenoaks, but also three of the world’s leading memory semiconductor giants: Micron Technology, Samsung Electronics, and SK Hynix. The post-investment valuation jumped directly to $965 billion—nearly touching the valuation ceiling of currently unlisted AI technology companies worldwide.

A company would not rush to submit its filing right after finishing a round of huge funding, with ample cash on its books, unless it had seen an even bigger game—namely, the liquidity premium in public markets.

Such a geometric leap in valuation also means the capital market’s pricing logic for foundational AI large-model companies is shifting: the market is no longer simply paying for “the grand vision of AGI,” but is seeing the real possibility of a perfect closed loop formed by computing power, models, and applications.

The idea of abandoning pure money-burning projects and instead prioritizing betting on top players that can realize revenue and support valuation has become one of the prevailing views.

2 From $965 billion to surpassing OpenAI, with $47.0 billion ARR

What does $965 billion mean?

If compared with the closing prices of domestic tech giants on June 1, it is roughly equivalent to 1.9 Tencent (closing: $508 billion) and 3.2 Alibaba (closing: $300.9 billion). If compared with the two domestic large-model leaders that are soon expected to return to the STAR Market—then it is about 11.5 Zhipu AI (closing: $83.6 billion) and 34 MiniMax (closing: $28.4 billion).

At the same time, Anthropic’s annualized revenue has already exceeded $47.0 billion, representing explosive growth from the $9 billion at the end of 2025. Behind this growth is the fact that some of the world’s largest companies are embedding Claude into their core business workflows, as well as Anthropic’s extremely strong product iteration cadence.

The recently released Claude Opus 4.8 version improved runtime speed by 2.5x, significantly cut costs by 3x, and also made major optimizations to the “hallucination” problems that trouble enterprise customers, reducing the probability of missing code defects by 4x.

Meanwhile, as more powerful Mythos-level models are set to be pushed to customers, Anthropic is building a workflow ecosystem centered on real commercial deployment. When computing costs are effectively controlled, and enterprise customers demonstrate a strong willingness to pay in a closed-loop manner, Anthropic enters a period of cashing in on commercial value.

It’s also worth noting Anthropic’s layout in computing infrastructure. Based on the H-round funding announcement and information previously jointly disclosed, the company has already locked in three core compute agreements:

• Reaching a cooperation agreement with Amazon AWS for up to 5 gigawatts of dedicated compute capacity, and in this H-round Amazon also completed a $5 billion strategic cash increase, tightly linking compute procurement and equity investment.

• Teaming up with Google and Broadcom to secure a combined total of 5 gigawatts of next-generation TPU compute resources, providing dedicated hardware support for continuous iteration of large models.

• Deep compute cooperation with SpaceX, connecting to the complete GPU compute cluster of Colossus 1.

In addition, the H-round financing also brought in Micron, Samsung, and SK Hynix—three of the world’s top storage chip giants—as strategic investors. The three firms monopolize the global mainstream high-bandwidth memory (HBM) and core storage capacity. This deep binding of equity and the supply chain also ensures Anthropic’s stable supply of core hardware needed for AI model training over the coming years, completing an end-to-end closed loop across compute, chips, and the supply chain.

3 With the IPO Gate Wide Open, the AI Business Landscape Enters a Race

Anthropic, OpenAI, and SpaceX are widely regarded as the three IPO targets most eagerly anticipated in this year’s market.

SpaceX submitted its listing application in April, and Anthropic’s top rival, OpenAI, is also expected to follow up and apply in the near term. The competition between the two has already upgraded from a contest of parameter counts and benchmark rankings to a final showdown for liquidity and capital pools in the secondary market.

After going public, ordinary retail investors will be able to invest in these top AI startups, and early investors may also reap exit returns amounting to billions of dollars.

Dan Ives, an analyst at Wedbush Securities, described it as “the opening of the IPO market gates.” The relatively quiet IPO market for tech stocks in recent years is being reignited due to AI’s maturation.

Why now? The answer is largely hidden in AI’s nearly bottomless demand for computing infrastructure. In the battle toward Artificial General Intelligence (AGI), private funding in the hundreds of millions or billions of dollars is no longer enough to support the consumption of clusters with thousands or tens of thousands of chips. Only Wall Street’s massive capital pool can sustain this “arms race” of computing power and energy.

Anthropic’s filing is what kickstarted this tech-stock IPO frenzy.