OpenAI has gone back to its investors with a smaller ask. According to reporting from the Financial Times relayed by Reuters on October 30, the company has cut the target for its latest round by $10 billion — from roughly $30 billion to about $20 billion — and reduced the valuation it is seeking from $1.4 trillion to $1.25 trillion. Investor subscriptions were reportedly due on Friday, and the first close may land below even the revised figure.
That is not a collapse. It is a repricing, and it arrived in the middle of the worst month the AI trade has had in years.
The month the market stopped taking the story on faith
The Nasdaq-100 fell 6.7% in October, its sharpest monthly decline since the tariff selloff, with more than $2 trillion in index market value evaporating. On October 30 alone the Nasdaq Composite dropped 2.86% to 25,437 and the S&P 500 lost 1.82% to 7,580.07, while the VIX jumped nearly 13% to 19.42 — a level that says investors are buying protection, not bargains.
The trigger was not a product failure. It was cash flow. Alphabet fell more than 12% in a single session, erasing roughly $500 billion in market value, after capital expenditure reached about $51 billion for the quarter and free cash flow came in around negative $9 billion. Search revenue growth slowed to 12%. Amazon reported negative free cash flow of $42.4 billion for the twelve months to September 30, a reversal from positive $14.8 billion a year earlier, even as AWS grew 39% and full-year capex headed toward $220 billion. Meta also posted negative free cash flow. Microsoft, still cash-flow positive, became the exception that sharpened the question: who can fund this out of operations, and who cannot?
The debt door is closing too
The equity side is one half of the story. Reports this week, citing the Wall Street Journal, said neoclouds CoreWeave, Nebius and Nscale have effectively been shut out of the bond market in recent weeks as lenders demand tougher terms — a bottleneck that lands directly on the GPU-leasing model that much of the build-out depends on. Credit markets are not repricing AI optimism; they are repricing AI leverage, and they are doing it faster than the equity market.
Put the three facts together and the shape is clear: a frontier lab takes a smaller cheque at a lower valuation, hyperscalers post negative free cash flow while guiding capex higher, and the specialist compute landlords can no longer borrow at will. Any one of those is a data point. Together they describe a shift in who pays for the next gigawatt — and at what price.
What actually changed, and what did not
Demand has not visibly cracked. AWS grew 39%. Alphabet's cloud revenue roughly doubled. Data-center revenue at Intel rose 78% this quarter, for whatever comfort that offers. Companies are still building. What has changed is the tolerance for spending ahead of returns, and the willingness of capital to accept a story instead of a cash-flow statement.
The original justification for the spending — that training and inference capacity would convert into revenue fast enough to self-fund — was always a timing bet. In October, the market shortened the clock on that bet.
What to watch next
- AMD and Nvidia earnings, the cleanest read on whether chip demand is still accelerating or merely still large. - Whether hyperscalers hold, raise or trim 2027 capex guidance. A hold is the bullish signal; a raise without an accompanying revenue acceleration is not. - Whether OpenAI's first close lands below $20 billion. A down-round that closes under its own reduced target is a different signal from one that fills. - Credit spreads on the neoclouds. If the bond window reopens, the financing squeeze was a pause. If it does not, the constraint moves from valuation to physics.
