“USD 2 Trillion Valuation Is Excessive”: Anthropic’s Deteriorating Results Dampen Investor Sentiment—Could China’s Tech Stock Rout Be Repeated?

Anthropic’s revenue growth slows and earnings swing to a loss, while AI infrastructure investment burdens come into focus
Market doubts over credit rating and valuation deepen, cooling investor sentiment ahead of IPO
As China’s AI bubble bursts, limited profitability in AI services leaves hardware companies reaping most of the benefits

Anthropic’s rapid growth has hit the brakes. With revenue growth and profitability slowing sharply, the added burden of massive long-term infrastructure investments is rapidly dampening investor sentiment. Concerns are growing in credit markets that Anthropic could remain below investment grade even after its initial public offering (IPO), while major institutional investors are lowering their expectations and questioning previously discussed valuations. Some analysts even say the market conditions surrounding Anthropic recall the sharp correction in technology stocks previously seen in China.

Anthropic’s Results Take a Tumble

According to a Reuters report on the 2nd (local time), Anthropic reported third-quarter revenue of USD 16.96 billion in an amended IPO registration statement filed that day with the U.S. Securities and Exchange Commission (SEC), up 47% from the previous quarter. That represents a sharp slowdown from the second quarter’s growth rate of nearly 144%. Profitability also deteriorated. Anthropic posted an operating profit of USD 974 million in the second quarter but swung to an operating loss of USD 567 million in the third quarter. Higher costs for securing artificial intelligence (AI) computing capacity through short-term contracts were cited as a major reason for the deterioration. To meet surging demand for its services, the company secured USD 5 billion worth of computing capacity through short-term contracts, which are more expensive than long-term agreements. An additional USD 2 billion noncash charge related to its employee donation-matching program further weighed on results.

Spending pressure from aggressive infrastructure expansion is also substantial. Anthropic disclosed total spending commitments of USD 895.4 billion under purchase and lease agreements for AI computing infrastructure. Of that amount, USD 69.8 billion is scheduled to be paid before October 2027, while spending expected to be incurred by the end of 2030 totals USD 447.5 billion. Most of the remainder is expected to be spent between 2031 and 2045. Anthropic warned that the financial burden of its long-term contracts could increase if future revenue growth falls short of expectations or its partners face worsening financing conditions.

Table 1. Anthropic’s Third-Quarter Results and Infrastructure Spending Burden

Pessimistic Credit Rating Outlook

As Anthropic’s results deteriorate markedly, credit markets are becoming increasingly wary of its financial health. Blackstone, the world’s largest private equity firm, reportedly recently told investors that Anthropic is likely to fall short of an investment-grade rating even after its IPO and instead receive a rating around BB, in the upper tier of speculative grade. Anthropic previously received a BB+ issuer credit rating, the highest speculative-grade rating, from Kroll Bond Rating Agency (KBRA) in June. It has yet to receive an official rating from any of the three major credit rating agencies: Moody’s, Standard & Poor’s (S&P), and Fitch.

Remaining at speculative grade could place significant constraints on Anthropic’s future fundraising. Major institutional investors in bond markets, including global pension funds and insurers, often restrict investments in corporate bonds rated below BBB-, the investment-grade threshold, under their internal investment rules and prudential standards. Without that rating, Anthropic would have to rely on the high-risk bond market, which has a narrower investor base, and would be more likely to need to offer higher interest rates to attract funding. That would be a painful setback for a company that has already committed to massive AI infrastructure spending and will need additional borrowing over the long term.

USD 2 Trillion Valuation Also Under Question

The argument that Anthropic’s valuation is inflated is also gaining traction in financial markets. Anthropic plans to begin a full-scale investor roadshow next week as it prepares for a U.S. stock market listing. No specific listing timetable has been disclosed, but market participants widely view a debut in the fourth week of November, just before U.S. Thanksgiving, as the most likely scenario. The final offering size, proposed price range, and target valuation have yet to be determined, but the market has spent the past several months expecting Anthropic’s valuation to be set at around USD 2 trillion.

Recently, however, major institutional investors have begun questioning that valuation. According to The Wall Street Journal (WSJ) on the 2nd, institutional investors considering participating in Anthropic’s IPO recently asked the company and its listing advisers to adjust its valuation to around USD 1.5 trillion–1.6 trillion. The request follows an accumulation of headwinds: falling share prices across U.S. technology stocks, concerns about excessive AI infrastructure investment, and opposition to data center construction from local communities and political leaders. A key risk cited was the uncertainty over when major technology companies such as Amazon and Meta would recoup their investments and whether they could monetize them, despite pouring vast sums into AI capital expenditure.

China’s AI Industry on a Downward Path

Some analysts say weakening investor sentiment in the U.S. AI market recalls the earlier technology stock rout in China. The STAR50 index, which tracks 50 leading technology and innovation companies on the Shanghai Stock Exchange, plunged 40% in roughly four months after peaking in June. The CSI300 index, comprising major Chinese large-cap stocks, also fell about 20% over the same period. Bloomberg estimated that approximately USD 2.4 trillion in market capitalization had been wiped off China’s stock market since May. The losses followed a sharp correction in shares of AI and semiconductor companies, which had surged on the back of extensive government support and policies promoting technological self-reliance.

The severity of the situation is even clearer in individual AI companies’ share price movements. Chinese humanoid robot maker Unitree drew intense market attention when it listed in August, with its retail offering oversubscribed by more than 8,000 to 1, but its share price subsequently halved in roughly two months. Chinese memory chipmaker ChangXin Memory Technologies (CXMT) also saw its shares fall more than 25% over the past month. AI chip designers MetaX and Moore Threads posted gains of 700% and 400%, respectively, on their first trading days late last year, but both companies’ share prices have halved since July this year.

Profits Concentrated in Hardware Companies

Weak profitability among AI model developers is cited as a key reason for this situation. Chinese AI companies grew rapidly with full government backing and actively adopted open-source and open-weight strategies, releasing their models’ core architectures and weights to expand the technology ecosystem. The problem was that this led to a surge in startups offering inexpensive AI services, intensifying price competition. As consumers continued switching to relatively cheaper models, companies struggled to generate steady revenue. U.S. sanctions, including restrictions on imports of Chinese-made humanoid robots and rules limiting the use of Chinese-made components in AI data centers, have further raised barriers to entering overseas markets.

With AI model developers unable to establish a sufficient foundation for generating profits, what profits there were became concentrated in hardware companies making semiconductors, servers, and optical communications equipment. According to China’s Securities Daily, the combined net profit of 10 major Chinese semiconductor companies reached CNY 111.78 billion in the first half of this year, while three optical module companies posted net profit of CNY 22.384 billion. By contrast, only slightly more than 40% of 95 listed AI software-related companies were profitable, and some remained in the red despite spending more than 20% of their revenue on research and development. In its analysis of first-half results, the China Association for Public Companies likewise concluded that “profitability in the AI industry tended to decline further down the value chain, from semiconductors to computing infrastructure, large models, and industrial applications.”