Editor’s Note
AI demand stayed strong across the period, but the market focused on what it costs to build capacity and how those costs will be paid back. Meta’s revenue grew, yet its spending plan rose and free cash flow turned negative. Oil also climbed above $114 a barrel, adding to inflation pressure as investors weighed interest rates.
AI Demand Is Strong, And Financing Is Under Scrutiny
The period opened with a sharp selloff in AI and chip stocks, followed by a brief rebound and another drop. One market recap put the SOX chip index down 16% over eight straight sessions. On October 23, the Nasdaq 100 fell 2.90%, the Philadelphia Semiconductor Index fell 6.7%, and Nvidia fell 4%. Strong supplier forecasts did not prevent the earlier selling: TSMC projected 51% quarterly revenue growth, and ASML raised its 2026 sales outlook for the third time.
The concern is increasingly about financing and returns. CoreWeave reported $640 million in quarterly net interest expense against $2.58 billion in revenue. Its 2032 bonds traded below 80 cents on the dollar, with yields above 15%. Elice, a GPU rental and operations company, raised $151 million in South Korea’s largest IPO of the year, then fell 30% on its first trading day. Reuters said the debut revived concerns about the AI trade.
There is also evidence of demand across the supply chain. Nvidia CEO Jensen Huang called demand for AI computing infrastructure “off the charts.” Nebius secured a $3.5 billion credit facility from Citi and pointed to more than $40 billion in long-term contracts with investment-grade customers. In the period’s reported results, Broadcom’s AI semiconductor revenue was up 221% to $16.7 billion, while AWS revenue grew 37% to $42.2 billion and Google Cloud revenue grew 82% to $24.8 billion. These figures show reported growth in several parts of the buildout, while the financing questions remain.
Meta’s Growth Comes With a Larger AI Bill
Meta reported $64.6 billion in Q3 revenue, up 26% year over year. Ad impressions rose 10%, and the average price per ad rose 14%. It guided to $73 billion to $76 billion in Q4 revenue.
Its 2026 capital spending range is now $135 billion to $145 billion, after the low end was raised. Q3 spending was $34.9 billion, up 80% from a year earlier. Meta reported negative $2.1 billion in Q3 free cash flow. Its EPS was $3.83, against $6.80 expected; one brief said it would have been $7.40 without a $9.89 billion settlement charge. The operating business grew, but these figures leave less room to assume that growth is already translating into cash after investment and other costs.
Meta is also trying to sell AI services directly. Its personal agent has a free tier and monthly plans at $20 or $100. Zuckerberg said Muse had more than 10 million users, was costly to run, and was targeting more than 10 times better cost efficiency next year. Meta is exploring a share of agent-related shopping transactions, but says it has no concrete plan. The user count and paid plans are early signs of reach and a possible revenue model; the digests provide no subscriber figures.
Oil Adds Another Pressure Point
Brent rose above $114 a barrel on October 23, after reaching $111.64 in an earlier market brief. Reuters linked the latest rise to attacks in Saudi Arabia and no diplomatic breakthrough in the Middle East or Ukraine. The period also included September headline inflation of 3.7%, up from 3.4%, and one-year inflation expectations of 4.7% in the University of Michigan survey.
The rate picture remained unsettled. The supplied digest said the Fed was widely expected to hold rates on October 28, while a December increase remained likely. Earlier, Bank of America still forecast an October hike, while markets priced a 20% probability. Higher oil and reported price pressure complicate that debate; weak housing points the other way. Pending home sales fell 7.3% in September, and traders priced roughly a 15% chance of an October increase in another digest.
The cost pressure is already visible for one group of businesses: Reuters reported that resilient passenger traffic and sharply higher fares had not fully offset soaring fuel costs for US airlines.
Counter-Thesis and Risk Watch
Several commentators warned that AI spending and market conditions may be more fragile than current demand figures imply. One cited Gartner’s forecast that more than $3.5 trillion in AI infrastructure spending could face impairment by 2030, and that more than 80% of neocloud providers could cease operating independently, including through consolidation. Those are forecasts, not reported losses.
Other commentators described the market as speculative or warned of a sharp reset, citing high leverage, oil, yields, and stretched valuations. One pointed to a Shiller CAPE ratio around 41, compared with around 32.6 at the 1929 peak, while also noting margin debt was a little under 2% of market value today versus almost 10% in 1929. These are risk arguments, not evidence that a crash has begun.
Looking Ahead
Can reported AI demand produce enough revenue and cash flow to support the cost of building capacity, especially for companies relying on debt? And does oil stay high enough to keep adding pressure to inflation and the Fed’s rate decision?