Zahlung innerhalb von 48 Stunden, Schätzung von 6 Billionen Yuan

There is a high demand for projects in popular industries, and the valuations of unicorn companies are skyrocketing. This phenomenon is not only occurring in the current Chinese primary market. When a startup asks investors to confirm their investment intention within 48 hours, it is a sign of a new phase in the financing history of the AI field in the spring of 2026.

According to several foreign media reports, Anthropic plans a new round of financing of up to $50 billion, with an estimated company value of $900 billion (about 6 trillion yuan). The sharp increase in the company's value within a month is remarkable. First, Bloomberg and Business Insider reported that Anthropic had received several pre - financing proposals with a company value of $800 billion. Later, TechCrunch, citing six informed sources, gave an estimate between $850 billion and $900 billion and reported that Anthropic asked investors to submit an allotment plan within 48 hours.

According to informants, due to the strong increase in investor demands, it is very likely that the final company value will exceed this amount. This means that the potential company value of Anthropic already exceeds that of OpenAI. Its main competitor completed a record financing round of $122 billion at the end of March and now has a company value of $852 billion.

Of course, Anthropic's deal is not yet finalized. A company spokesperson refused to comment. But that doesn't matter. In the primary market, such lightning investments almost only occur with the best assets. Since the release of ChatGPT at the end of 2022, Anthropic has been constantly catching up with OpenAI, and this trend was particularly strong last year. The growing revenues and company values at least show that it has finally caught up.

Annual ARR over $40 billion, competition for listing spots

A year ago, in May, Anthropic released Claude Opus 4 and Claude Sonnet 4 at the first developer conference, marking the beginning of an almost one - year technological upgrade. Opus 4, the flagship model, can work autonomously for seven consecutive hours and achieved a score of 72.5% in the SWE - bench programming benchmarks, surpassing GPT - 4.5 of the same period and consolidating its advantage in handling complex corporate tasks. Subsequently, Claude Opus 4.5 was introduced, which offers a context window of over 200,000 tokens and can fully process entire books and millions of lines of code. The technological implementation also accelerated, and the Claude model is the only large model that covers the three major cloud platforms AWS, Google Cloud, and Azure at the same time.

At this stage, Anthropic emphasized security and efficiency in its technological route and refused to relax the ethical criteria for AI for short - term interests. Anthropic even publicly rejected a $200 - million contract with the US Department of Defense and maintained the "red line" that the model should not be used for lethal weapons or mass surveillance. This enhanced the trust of highly compliant customers.

In contrast to OpenAI, the core driver of Anthropic's growth is the explosion of corporate demand and the improvement of the product portfolio. On the one hand, eight Fortune 500 companies have become in - depth customers of Claude, and the customized orders in industries such as finance, healthcare, and law have grown strongly. On the other hand, intelligent products such as Claude Code (programming assistant) and Claude Cowork (office assistant) were introduced, forming a complete matrix of "basic model + vertical application" and significantly increasing the value per customer.

Thanks to Claude Code, Anthropic announced last month that its annual revenues (ARR) had already reached $30 billion, surpassing OpenAI. By May 2026, this value rose to $44 billion. In addition, an analysis by SemiAnalysis showed that the gross margin of Anthropic's inference business increased from 38% in May 2025 to over 70% in May 2026, which is far above the industry average in terms of cost - efficiency.

For Anthropic, which is advancing rapidly in business, computing power is currently the most urgent problem. At the beginning of this month, Anthropic released its latest model, Claude Opus 4.7, and described it as the best product for the mass market. Earlier, Anthropic released Claude Mythos Preview, a model with advanced network security functions, which is only available to some selected companies. Mythos led to a series of high - level meetings between government officials, technology companies, and bank executives, which is also one of the reasons for Anthropic's new round of financing - Anthropic needs huge computing resources to run the Mythos model.

In the past few weeks, computing power has even been the most notable issue for Anthropic. Anthropic signed an agreement with Amazon, under which Amazon will immediately invest $5 billion in Anthropic and can invest an additional up to $20 billion if certain business milestones are reached, with a maximum total of $25 billion. In return, Anthropic will receive up to 5 gigawatts of computing power from Amazon Trainium chips for the training and deployment of the Claude model, about 1 gigawatt of which is expected to go online by the end of 2026.

Anthropic also announced that it will cooperate with Google and Broadcom to obtain about 3.5 gigawatts of computing power from the next - generation TPU, which is expected to go online in 2027. Google has subsequently promised additional investments and vowed to provide an additional 5 gigawatts of TPU computing power in the next five years. Google said it plans to invest up to $40 billion in Anthropic, with $10 billion immediately and up to $30 billion more if the performance targets are met.

Just like OpenAI, Anthropic is also in a hurry to go public, both to satisfy the capital market and to have sufficient funds to acquire computing power.

According to further reports from Techcrunch, some early investors, especially those who invested in 2024 or earlier, have decided to skip this round of financing despite the high market pressure. These investors are waiting for the expected IPO later this year to sell their investments. Earlier, the Financial Times also reported that Anthropic has already selected a law firm to start preparations for the IPO, and the IPO may take place as early as this year.

The FOMO game of capital

The current round of financing is also regarded as the last round of financing before the IPO.

In September last year, Anthropic completed an F - series financing round and raised $13 billion, with a post - investment company value of $183 billion. The financing round was led by ICONIQ, Fidelity, and Lightspeed, and top capital providers such as Blackstone, Goldman Sachs, and the Qatar Investment Authority followed. After this financing round, Anthropic officially joined the club of billion - dollar AI companies and, together with OpenAI, formed the seeds of a duopoly.

In February this year, Anthropic completed a G - series financing round and raised $30 billion, with a post - investment company value of $380 billion. The financing round was led by the Singaporean sovereign wealth fund GIC and the hedge fund Coatue, and Google and Amazon continued their strategic investments. This was the second - largest private - equity financing deal globally, only behind OpenAI's $40 - billion financing round.

In the latest round of financing, which aims for a company value of $900 billion, the most notable change is not the amount but the extreme shortening of the transaction rhythm. The traditional financing process in the primary market includes due diligence, negotiations, determination of conditions, approval by the investment committee, and transfer of funds and takes weeks or even months. However, Anthropic's financing round follows a completely different logic: The subscription window is opened, investors compete for spots, and the shares are determined within 48 hours.

Behind this is the fact that Anthropic's growth logic has evolved from a tool product to a corporate infrastructure: It is no longer just a "promising AI startup" but the fastest - growing B2B company in the history of the software industry. Eight of the ten largest companies in the world are already paying customers, and the number of corporate customers with an annual consumption of over $1 million has exploded from a few initial values to hundreds or even thousands. This means that Claude is no longer a pilot project for companies but is gradually becoming an indispensable tool that is deeply integrated into core business processes.

Once AI models become the "power, water, and gas supply" of the digital era, the financing logic changes fundamentally. Capital no longer invests in future growth but tries to gain core access. Missing out on Anthropic could mean missing the core sector of corporate AI services in the next ten years. This fear makes top capital providers willing to give up the right to negotiate and shorten the decision - making time.

Of course, both Anthropic and OpenAI face an inevitable structural burden: The costs of model training are increasing exponentially, and the cost of a single ultra - large - scale training can easily reach several billion dollars. The inference costs remain high, and the computing power consumption caused by the large - scale use of corporate customers requires continuous huge financial resources. Most importantly, the global demand for corporate AI is exploding, and the computing power capacity directly determines the supply capacity and market shares.

For leading AI companies, financing means more survival resources. If financing is done too late, one may fall behind in the race for computing power, lose customers to competitors, and widen the gap. This vital pressure forces Anthropic to complete the financing as soon as possible, and capital is willing to increase the speed because the later one invests, the fewer shares are left.

Previously, Anthropic was not a favorite of capital and was rejected by 21 top VC firms. Today, its company value has increased by several orders of magnitude, and it has become a "hot commodity" for global capital. This dramatic turn has completely changed the behavior of investors. One can imagine that many top investment institutions have waived the traditional due - diligence process, simplified the approval by the investment committee, and even transferred funds before the formalities were completed, just to secure the shares in time.

For them, the risk of missing out on Anthropic is far greater than the risk of a high company value. After all, so far, Anthropic is the only company that can compete with OpenAI, and its uniqueness is irreplaceable.

This article is from the WeChat account "Touzhongwang", written by Liu Yanqiu and published with the permission of 36Kr.