Source: Jin Ten Data

According to the Financial Times, some investors in Anthropic expect the company could go public as early as this October, with a valuation reaching $2 trillion or even higher. This figure doubles the previous preliminary target of about $1 trillion, but Anthropic has not yet finalized its IPO valuation.

If this expectation comes true, Anthropic's valuation at the time of its IPO would surpass SpaceX's approximately $1.77 trillion IPO valuation, making it the largest IPO globally. Previously, foreign media reported that Anthropic secretly submitted its IPO application to the U.S. in June, but the scale of the issuance and specific terms have not yet been disclosed.

Investors are daring to double their valuation expectations based primarily on revenue. Anthropic reportedly had revenue close to $11 billion in the second quarter this year, more than doubling from $4.8 billion in the first quarter.

When Anthropic completed $65 billion in financing at the end of May, the post-investment valuation had already reached $96.5 billion, exceeding OpenAI. At that time, the company revealed that its annual revenue had already surpassed $47 billion in May.

Currently, investor expectations are even more aggressive: by the end of 2026, Anthropic's annual revenue could reach between $100 billion and $120 billion, more than a tenfold increase from the beginning of this year. This means that from May to the end of the year, the company’s annual revenue needs to increase by more than double.

$2 trillion bet on continued rapid growth in revenue

Based on annual revenue of $100 billion to $120 billion, a $2 trillion valuation corresponds to a revenue multiple of approximately 17 to 20 times. For Anthropic to support this valuation, it cannot rely solely on an increase in Claude user numbers; it is more critical whether enterprise customers continue to expand their AI spending.

Anthropic has not disclosed the official user scale of Claude. Statista estimates that, as of June this year, Claude had approximately 245 million monthly users globally, while OpenAI's ChatGPT users had already surpassed 1 billion.

Despite the significant gap in user numbers, Anthropic's valuation has at times exceeded that of OpenAI this year. One of the key reasons behind this is that Anthropic is gaining more customers in the enterprise market, where enterprise clients typically can generate higher single client revenue.

This advantage also faces increasingly direct cost competition. Reuters reported that some models launched by Chinese companies are continuously approaching the capabilities of U.S. leading closed-source models in tasks such as coding, while the usage cost is significantly lower, leading companies to re-evaluate whether they must pay higher fees for cutting-edge models.

Marc Bhargava, General Catalyst's Managing Director, noted that many basic operational tasks do not require the most advanced models, and companies will ultimately return to the issue of return on investment. However, in high-difficulty tasks such as coding, he expects that Anthropic and OpenAI will still be able to maintain an advantage.

Anthropic has recently begun intense communications with potential investors. According to the Wall Street Journal, the company is responding to competition from low-cost models, friction with the U.S. government, and issues regarding data center construction costs.

Government relations have already affected Anthropic's business. Earlier this year, the U.S. government temporarily banned Anthropic from exporting two high-performance cybersecurity models due to security concerns, a restriction that was lifted after slightly more than two weeks.

The Financial Times quoted investors stating that this restriction once hindered Anthropic's June revenue, but after the ban was lifted, the company’s revenue bounced back at a "remarkably rapid" rate. Meanwhile, Anthropic still has ongoing legal disputes with the U.S. government regarding whether its products can be officially used by government departments.

Nearly $100 billion in funding this year, computing power becomes another battlefield

The rapid growth in revenue has another aspect: Anthropic needs to continuously increase its power investment. The company had previously limited some services during peak demand periods due to insufficient capacity, and infrastructure supply has started to directly affect how much new demand Claude can handle.

This year, funding from venture capital firms, tech companies, and institutional investors has approached $100 billion for Anthropic. In addition to expanding model training and inference infrastructure, the company also plans to develop its own AI chips to alleviate long-term supply pressure on computing power.

In July, AMD committed to investing $5 billion in Anthropic. This collaboration will enable Anthropic to access 2GW of computing power supported by AMD's latest generation of chips.

Amazon announced in April that it would further invest $25 billion in Anthropic. As part of the collaboration, Anthropic will continue to use Amazon's cloud computing infrastructure on a large scale, with related cloud services investments exceeding $100 billion.

Anthropic has also been reported to be negotiating to acquire the AI infrastructure company Decart AI, supported by NVIDIA, with a potential deal size of about $6 billion. Decart primarily develops AI infrastructure and inference optimization technology, and if the transaction is completed, its team is expected to join Anthropic's department responsible for inference and performance.

These actions mean that the competition faced by Anthropic before its IPO is no longer limited to who can train stronger models. Inference costs, chip supply, data center capacity, and the ability to continuously secure capital will all impact its ability to truly convert model demand into revenue.

For investors, the most important data over the next few months remains revenue. Anthropic's second-quarter revenue has increased from $4.8 billion in the first quarter to nearly $11 billion, but the goal of reaching $100 billion to $120 billion in annual revenue by the end of the year means that growth should not significantly slow down.

If Anthropic can proceed with its IPO in October as planned, the officially disclosed prospectus will allow public market investors to see for the first time the actual revenue growth rate, customer structure, infrastructure expenditure, and capital consumption of this approximately five-year-old AI company.

At that time, the market will also be able to assess more directly whether the $2 trillion valuation is based on realizable growth rates, or if investors are paying extremely high revenue multiples in advance for AI demand in the coming years.