Goldman Sachs has increased its 2026 U.S. capital expenditure growth forecast to approximately 7.8%, driven by significant investments in AI infrastructure and supportive government policies. This revised outlook, released on May 26, 2026, predicts continued strong business investment despite higher oil prices, with GDP growth projected at 2.1% for the same period. The bank emphasizes that the current investment wave is unique, fueled by secular trends in AI rather than cyclical factors.
While strong corporate profits, with the S&P 500's return on equity (ROE) reaching a record 22% in Q1 2026, have supported high U.S. equity valuations, Goldman Sachs warns that the extensive AI infrastructure buildout will increase asset intensity for major technology firms. This surge in capital expenditure (capex), depreciation, and financing needs is expected to compress ROE for the largest tech companies. Specifically, the seven biggest technology names—Nvidia, Apple, Microsoft, Amazon, Alphabet, Meta Platforms, and Broadcom—whose combined ROE reached 44%, are projected to see their ROE fall by an average of seven percentage points next year.
Hyperscale technology companies are becoming more asset-intensive to support AI workloads. Depreciation and amortization expense is projected to climb from about 7% of revenue in 2022 to roughly 12% by 2027. Major hyperscalers are expected to commit approximately $770 billion to capital expenditures in 2026, an amount equivalent to roughly 100% of their operating cash flow. Goldman Sachs forecasts that hyperscalers will direct about 98% of operating cash flow into CapEx by 2026, with big tech capital expenditures potentially reaching $920 billion by 2027, and possibly up to $1.4 trillion under a more aggressive scenario. This level of spending is nearing those seen during the dot-com era, raising questions about whether AI spending can generate sufficient returns to justify the massive investments.
