Meta (META.US) cash flow turns negative for the first time; AI funding anxiety triggers market revaluation

According to Woofun AI, on October 22, the US stock market experienced a rare turbulence: the Nasdaq Index closed down by about 2.6%, the Philadelphia Semiconductor Index plummeted by about 6.7%, Meta (META.US) shares fell about 7%, while the Dow was nearly flat. On the surface, this appears to be a retreat from AI crowded trades, but the pricing logic has fundamentally shifted: in the past two years, the market focused on "how many more chips to buy," while now the focus is on "who will pay for AI."

This change was triggered by Meta's financial report released the previous day, causing the market's attention to quickly shift from the demand side to the financing side.

Meta’s third-quarter financial report shows adjusted earnings per share reached $7.40 (excluding $9.89 billion in legal expenses), while GAAP earnings were only $3.83. However, free cash flow switched from a positive $10.63 billion in the same period last year to a negative $2.12 billion this quarter, marking the first quarterly net outflow in more than a decade. The 2026 capital expenditure guidance was adjusted to $135 billion to $145 billion, with the lower end raised from the previous $130 billion. The board approved the issuance of non-voting Class C stock on a 1:1 basis in January 2027. Capital expenditure for the quarter was about $35 billion, and ad revenue growth failed to cover server and data center expenses. The Class C stock is designed to retain Zuckerberg’s control while reserving space for future fundraising. The market sell-off was not due to the stock split but rather concerns over potential equity supply. Valuation is based on the perpetual growth of cash flow—how financing is done directly impacts the discount rate, not just single-quarter profits.

According to data compiled by Woofun AI, there are significant differences in the upstream and downstream industry chains regarding the causes of the decline. Nvidia (NVDA.US) CEO Jensen Huang stated in a CNBC interview that neocloud demand is "off the charts," and that the current bottleneck is operating capital, with capital demand far exceeding supply, but "this problem will be solved." His remarks pointed at liquidity constraints rather than disappearing orders, yet the market interpreted them as indirect evidence of financing bottlenecks. If upstream players admit clients are short on funds, the pace of chip orders will depend on financing progress rather than demand strength.

On the other hand, Bernstein (AB.US) analyst Madison Rezaei cut CoreWeave’s (CRWV.US) target price from $74 to $51, maintained an underperform rating, and estimated that fulfilling its committed orders would require about $85 billion in financing. The market's response was more direct: CoreWeave’s bond yields rose to about 15.8%, and its 5-year credit default swap spread exceeded 1,100 basis points. When the marginal cost of borrowing exceeds 15%, the expansion speed of GPU rental businesses is determined by creditors’ limits, rather than market demand.

Jensen Huang sees demand as robust and funding issues as solvable, while Bernstein views rising capital costs as a signal of weakened effective demand. The October 22 session marked the first large-scale market preference for the latter. Chip stocks fell more sharply than cloud providers, with Nvidia down about 4% and CoreWeave down about 10%. Lam Research (LRCX.US) exacerbated market volatility: its December-quarter revenue guidance midpoint slightly outpaced analyst consensus but fell short of buy-side private expectations, while also signaling that the proportion of revenue from Chinese customers declined and supply chain bottlenecks could delay shipments.

Although Lam emphasized strong AI-related demand and forecasted that global wafer fab equipment spending could exceed $200 billion by 2027, with record gross and operating profit margins for the quarter, its share price still fell by about 10%. Lam’s guidance can only explain its own retreat and not the sector’s overall 6.7% drop, but as a specific case, it clearly demonstrates how financing concerns can be transmitted from cloud provider balance sheets to the visibility of equipment orders. The market’s sensitive reaction to “margins falling short of expectations” confirms the high dependence of valuations on capital expenditure slopes.

From a macro perspective, Meta has become the first visible sample of capital structure change. Morgan Stanley (MS.US) estimates that next year, hyperscale cloud providers plus neocloud could see related capital expenditures reach $1.58 trillion, with a free cash flow gap (including leased equipment) of $540 billion. Goldman Sachs (GS.US) has a lower estimate—about $800 billion in capital expenditures for top cloud providers by 2026. Despite the near-doubling difference in the two banks’ data, both point to the fact that the scale of AI investments now exceeds the coverage ability of individual companies’ internal cash flows. The capital structure is shifting from endogenous cash flow to external capital, and the market has begun pricing in dilution and spread risk.

Current evidence does not support “AI demand peaking” or “the financing window closing”—CoreWeave’s bonds are expensive but still have buyers; Meta Class C merely reserves a channel and hasn’t issued new shares yet. The key observation point is when “expected dilution” turns into “actual dilution.” If leading companies actually issue large blocks of equity or new types of debt, valuation pressure will shift from discount assumptions to real supply. If credit spreads in the tech sector widen further and non-neocloud firms are also required to offer higher risk premiums, financing constraints will become industry-wide rather than isolated cases. Conversely, if next quarter’s capital expenditures are revised up again and order income follows, this adjustment will more likely be a phase of repricing. What October 22 left was not the conclusion of “the end of AI,” but a new equation: demand remains, but behind every GPU, the source of capital must be answered first.