According to Woofun AI, CoreWeave (CRWV.US) successfully reversed the market's previous sentiment that continued to be pressured by uncertain returns on AI investment with the third quarter earnings report and full-year guidance that exceeded expectations, and established its position as a weather vane for AI computing power. The key customer structure data disclosed by CEO Michael Intrator not only explains the resilience of revenue growth, but also directly boosted the overall valuation expectations of the Neocloud (Neocloud) sector, setting a positive market tone for the subsequent Anthropic listing and NVDA.US (NVDA.US) earnings report.
Earnings details released on Tuesday revealed strong fundamental support from CoreWeave. Third-quarter revenue reached $3.66 billion, and losses per share were controlled at $1.27, all better than analysts' expectations. More critical is its guidance for the full year 2026: revenue is expected to be between $13 billion and $13.3 billion, which is about 3.4% higher than the market's agreed upper limit. Intrator revealed during the conference call that more than half of the new customers this year came from outside the AI field. This diversification trend has effectively mitigated the market's concerns that demand for computing power is too concentrated and may decline like a cliff as the AI boom recedes. As early as April, Intrator described core partner Anthropic's contract as “multi-billion dollar”, a judgment now backed up by financial data.
The response from the capital market was quick and intense. CoreWeave's stock price soared by about 25% in a single day, driving the Xinyun sector to rise by more than 12% during the intraday period. The NASDAQ also used this to recover lost ground after the midterm elections. Deutsche Bank (DB.US) analyst Brad Zelnick commented that the company provided “reassuring details,” reflecting a significant recovery in market sentiment. According to Goldman Sachs (GS.US) data, the generalized AI stock basket rose by about 5%, significantly outperforming non-AI stocks.
However, capital flows are fragmented: hyperscale cloud vendors such as Microsoft (MSFT.US), Google (GOOGL.US), and Amazon (AMZN.US), which bear huge capital expenses, are lagging behind, while “sellers” such as computing power leasers and semiconductors have clearly benefited. This shows that investors prefer to bet on the direct payee of the AI construction cycle rather than the payer that continues to pay for high expenses.
According to data compiled by Woofun AI, Anthropic plans to complete the listing next week, and its valuation logic is deeply tied to CoreWeave. As the first cutting-edge AI lab in the US to seek public listing, Anthropic lacks a traditional business moat, and its valuation depends entirely on the market's judgment on the future of AI. The issue price range disclosed in the prospectus corresponds to a fully diluted valuation of up to about 1.4 trillion US dollars, which is lower than the 1.5 trillion to 2 trillion yuan range previously discussed in the market; other market sources say that underwriters are considering raising the valuation, but this is not yet reflected in the disclosed range. As Anthropic's key computing power supplier, CoreWeave's strong performance helped Anthropic gain higher recognition at the roadshow.
Meanwhile, OpenAI's revenue reached $10 billion for the first time in a single quarter, further supporting the application-level demand narrative. Note, however, that Anthropic also relies on other uncertain new cloud service providers, including Nscale, which is reportedly considering delaying its IPO plans, which poses a potential supply chain risk.
Next week's Nvidia earnings report will be another litmus test. The market not only focuses on GPU demand, but also wants to confirm the intensity of procurement by cloud vendors, the degree of matching supply of new Blackwell products, and the sustainability of the transformation of capital expenditure into profit.
Although the Philadelphia Semiconductor Index recorded its biggest one-day increase in months, and Nvidia's stock price also rebounded, traders observed that some of the funds were still being sold before earnings were reported. Bloomberg columnist Dave Lee notes that as early as March 2025, when CoreWeave went public, he warned that it would become “the first real weather vane stock for the AI revolution,” and now its stock price is a major pump for market fluctuations. If Nvidia fails to provide strong enough guidance, the volatility of AI trading may increase again.
The hard constraints behind the rebound still exist, and financing gaps and supply expansion are the core variables that determine whether AI infrastructure deals can go far. According to multiple logic, CoreWeave can set up financing entities around customer contracts, cover construction costs with a high percentage of guarantee debt, and reserve GPU assets for re-leasing after repaying the debt during the contract period. However, bears focus on three risks: first, there is a gap between the revenue level per megawatt mentioned by management and actual monetization; second, the market believes that the company may face a capital gap of about 33 billion US dollars next year and still need to be covered by additional debt or customer advance payments; third, by 2028, the market is expected to invest about 55 gigawatts of additional computing power capacity. If supply is released faster than demand, the rent and residual value of older GPUs will be under pressure.
Uncertainties in the macro-regulatory and interest rate environment further exacerbate the complexity. After the US midterm elections, the Democratic Party took control of Congress, and Texas governor-designate Gina Hinojosa advocated a moratorium on data center construction during the election campaign. Intrator wrote to Hinojosa last week seeking consensus on the “responsible completion” of approved projects and trying to stabilize investor sentiment, saying it is not expected to be substantially affected by recent major regulatory changes. In addition to regulation, interest rates are still a valuation constraint. During this round of rebound, the implied yield of 10-year US Treasury futures once again approached 5%, and rising financing costs directly affected Xinyun's project return.
Furthermore, although VIX once fell back to around 17, cross-asset volatility was still at a relatively high level, and safe-haven positions were only partially closed.
Overall, CoreWeave's earnings report temporarily repaired the market's confidence in AI computing power demand and financing capabilities, creating a favorable window for Anthropic's IPO and Nvidia earnings reports. However, from short-term short recovery to long-term fundamental revaluation, we still need to answer the four problems of financing, supply, GPU residual value, and interest rates. Next week's Anthropic and Nvidia will be two key tests to test whether this round of AI rebound can continue. Failure by either party to meet expectations could re-trigger market concerns about the infrastructure bubble.