Editor’s Note
AI demand showed up in cloud, chip, and infrastructure results across the period. But spending and cash flow became just as hard to ignore, and investors punished some companies even as their AI businesses grew quickly. The clearest question is whether the new capacity can produce enough cash to justify its cost.
AI Growth Is Real, But Cash Is the Test
Microsoft reported $92.3 billion in revenue, up 19%, with Azure growing 48%. Its $39.4 billion in capital spending came in below the $46.6 billion expected. Alphabet’s revenue rose 25% to $127.58 billion, and Google Cloud revenue more than doubled to about $30.7 billion. Yet Alphabet’s free cash flow was negative $9.07 billion after $51.36 billion in capital spending, and its shares fell more than 9%.
Amazon made the same tension harder to miss. AWS revenue reached $46 billion, up 39%, and management projected growth above 40% in both 2027 and 2028. But quarterly capital spending was $77.3 billion, versus $57.5 billion expected, and trailing twelve-month free cash flow was negative $42.4 billion. Amazon kept its full-year cash CapEx plan at $220 billion.
Meta reported $64.6 billion in revenue, up 26%, while costs and expenses rose 68%. Operating income fell 37% after a $9.89 billion legal charge. Its CapEx outlook is $135 billion to $145 billion, and the company said it expects to keep investing aggressively in 2027.
Across the period, the Financial Times reported that Amazon, Meta, Microsoft, and Alphabet lost nearly $900 billion in market value after their third-quarter results. Their capital expenditures almost doubled, and shares in all four fell. The comparison matters: strong demand is not enough to settle the spending question. Investors are looking for cash returns as well as growth.
Suppliers See Demand, While Commitments Add Up
The suppliers’ results provide evidence that customers are buying computing capacity. SK Hynix reported $78.9 billion in Q3 revenue and $62.8 billion in operating profit, both above estimates. It guided to 25% growth in DRAM shipments next quarter and expects memory demand to outpace supply through 2030. Separate reports said Alibaba, Tencent, and ByteDance are seeking memory contracts lasting up to 10 years, through 2035.
Vertiv reported $4.3 billion in revenue against a $3.8 billion estimate, and $652 million in free cash flow against $373 million expected. These results fit the demand story, but they do not show whether every large customer’s spending will pay off.
Anthropic’s filings and reported plans show the scale of the commitments. One account of its filing put first-half revenue at $16.3 billion, operating loss at $693 million, and compute costs at $8.9 billion. The same account cited roughly $850 billion in compute commitments stretching to 2045, with around half locked in. An earlier report described $11.56 billion in second-quarter revenue, more than $80 billion in first-half losses, and almost $132 billion in cash, equivalents, and short-term investments at June’s end. Those are different reported figures, so the exact financial picture remains unclear in the supplied material.
The commitment risks are clearer than the final cost. One report said Anthropic’s filing listed $321.4 billion in estimated future lease obligations and $265.6 billion in expected future fixed or minimum-guaranteed commitments. It later entered agreements with about $245 billion in minimum future payments, without an updated aggregate commitment figure. Growth is accelerating, but the scale and timing of the spending obligations make profitability hard to judge.
Inflation Keeps the Rate Question Open
The Fed held its rate at 3.75% to 4.00% in a 10-2 vote, with two officials preferring a hike. September PCE inflation was 3.6% year over year, while core PCE was 3.1%. Third-quarter U.S. GDP grew at a 3.4% annualized rate, above the 3.2% forecast.
Oil remained near $114 to $117 for Brent in the period’s reports, with WTI near $101 to $102 in one account. That sits alongside persistent inflation and a Fed decision that left the next move unsettled. The combination matters for companies committing large sums to infrastructure: the evidence supports strong demand, but the eventual cash returns remain open, while rates and energy costs add pressure.
Counter-Thesis and Risk Watch
Several commentators argue that AI spending may outrun the returns. One questioned whether AI’s return on invested capital is a “huge question mark”; another focused on Amazon’s $220 billion spending plan and asked whether it will produce enough cash soon enough. These are concerns about the payoff, not evidence that demand has disappeared.
Alphabet faces a specific monetization risk. One commentator said AI Mode searches earn less than traditional searches, while Google Search revenue grew 12% to $63.35 billion. Google said query growth was its strongest in five years. The unresolved question is whether AI answers can earn as much as traditional search ads.
Looking Ahead
Can cloud and chip demand keep growing fast enough to support the spending commitments, and can that growth turn into free cash flow? The results show demand today; they do not yet answer how much cash the new capacity will produce.