UBS raises AI spending forecast to $1.4 trillion for 2027 despite tech sell-off

UBS raised its AI capital-spending forecasts on November 2, projecting $1 trillion in 2026 and $1.4 trillion in 2027, and adding a $1.5 trillion estimate for 2028. Its assessment pairs confidence in infrastructure demand with a warning: investors increasingly want evidence that the spending will produce adequate returns, not just faster growth.

The revision follows a bruising Friday for technology shares. The Nasdaq Composite fell 2.9%, while the S&P 500 lost 1.8%. UBS interprets the reaction as a shift in emphasis: even with strong earnings and infrastructure demand, investors are putting more weight on overinvestment, financing and the risk that AI erodes existing businesses.

Bigger commitments, faster cloud growth

Stronger commitments from the largest cloud operators underpin the higher forecasts. Alphabet kept its 2026 capital-expenditure plan at $195 billion to $205 billion and signaled a significant increase next year. Meta raised the floor of its range to $135 billion, leaving a $135 billion to $145 billion spending plan, and pledged aggressive investment beyond 2026.

Both companies reported negative third-quarter free cash flow—the cash left after capital spending. Alphabet shares fell almost 13% the following day; Meta fell nearly 7%. UBS points to these reactions as evidence that the bar is rising, even as companies continue to commit money to additional capacity.

The demand evidence is substantial. UBS cites Google Cloud revenue growth of 103% year over year, up from 82% in the prior quarter. Azure grew 48% in the three months through September, versus 43% through June. Oracle’s growth accelerated to 62% from 47%. All three companies said additional capacity could help meet demand in their pipelines.

Debt comes due before computing pays

Financing is a more acute concern for neoclouds, specialized providers that mainly rent out computing capacity built around graphics processors. UBS says their debt has sold off because large, upfront investments depend on cash arriving later. Those future receipts remain sensitive to how fully equipment is used, customer renewals, project execution and the ability to refinance borrowing.

But UBS estimates these providers represent only about 8% of its AI spending forecast. Alphabet, Amazon, Microsoft, Meta and SpaceX account for 80%. That concentration among larger, generally better-funded companies is why the bank expects selected project delays, repricing and consolidation rather than a broad collapse in AI spending.

More usage does not settle the revenue question

Alphabet also illustrates a different risk: AI can attract activity while weakening the economics of an established business. UBS notes that Search query growth reached a five-year high, yet Search revenue growth slowed. Management indicated that AI Mode currently monetizes at a lower rate. More searches therefore did not translate into faster revenue growth.

UBS remains confident in the overall investment trend, but its report identifies specific developments that would challenge that view:

• Demand weakens: AI lab revenue growth slows meaningfully, cloud capacity is less fully used, or cheaper AI fails to attract enough usage to absorb new capacity.
• The computing equation changes: more efficient algorithms sharply reduce computing needs, or larger investments stop delivering matching improvements in AI capabilities.
• Funding tightens: a significant deterioration in financial conditions reduces major cloud operators’ willingness to finance new investment.