Widely cited estimates of global AI investment are significantly lower than the actual scale, according to the latest research from Goldman Sachs. The bank argues that the frequently referenced figure of approximately $800 billion in hyperscaler capital expenditure is systematically underestimated.
Goldman Sachs economists Joseph Briggs and Sarah Dong, in their August 2 report, state that the commonly cited $794 billion in hyperscaler capital expenditure both understates the total global AI capital expenditure by about $200 billion and overstates investment within the United States by a similar amount. After adjustments, Goldman Sachs estimates that U.S. AI investment will total roughly $581 billion in 2026, with the global figure reaching approximately $1.019 trillion.
Four Major Flaws in the Conventional Metric
The Goldman Sachs report identifies four fundamental flaws in using hyperscaler capital expenditure as a proxy for AI investment. First, this metric ignores investments from U.S. private companies, which play a key role in the AI ecosystem, and capital spending from other listed companies, with Goldman's credit team noting hyperscalers directly account for only 40% of AI-related supply in 2026. Second, it entirely omits investments from non-U.S. companies, particularly those in China and other parts of Asia. Third, hyperscalers had capital expenditure exceeding $150 billion before the AI boom, meaning a portion of current spending is unrelated to AI. Fourth, U.S. hyperscalers operate globally, and a significant share of their capital expenditure actually occurs outside the United States.
Three Converging Methods Point to Over $1 Trillion
Goldman Sachs employs three independent methodologies to estimate global AI investment, yielding highly consistent results. The primary estimate (enhanced hyperscaler capital expenditure) suggests global AI investment of $1.019 trillion in 2026, with $581 billion occurring within the United States. A first cross-validation method, which tracks revisions to gross profit forecasts for AI-related listed companies relative to a 2022 baseline, shows global AI investment of approximately $1.06 trillion in 2026, with cumulative AI-related spending increasing by over $1 trillion since 2022. A second cross-validation method, based on official national accounts and global trade data, calculates U.S. AI-related hardware investment at an annualized rate of about $463 billion (above the 2022 baseline) as of May 2026. Adding roughly $100 billion in AI-related research and development and intellectual property investment brings the current annualized total U.S. AI investment to nearly $600 billion. For other countries with limited data, Goldman Sachs uses global trade data and the historical relationship between U.S. imports and total investment, resulting in a global AI investment estimate of about $1.002 trillion. The average of the three methods indicates cumulative global AI investment will reach $1.8 trillion from 2022 to the end of 2026.
AI Capital Expenditure as a Share of GDP Expected to Rise, Aligning with Historical Technology Cycles
Looking at the medium-to-long-term trend, Goldman Sachs extrapolates from market consensus expectations for public company capital expenditure, forecasting that AI capital expenditure's share of GDP will continue to increase. Specifically, U.S. AI capital expenditure as a share of GDP is projected to rise from 1.8% in 2026 to 2.5% in 2027, and further to 2.8% in 2028. Globally, the figures are 0.9%, 1.3%, and 1.4%, respectively. Goldman Sachs notes these levels are within the historical range of peak investment shocks, which were 2% to 5% of GDP during the build-out of general-purpose technologies (GPTs). The report also points out that the timing of a slowdown in AI capital expenditure growth is a key source of uncertainty for the macro market, recommending a "dashboard" approach to track multiple leading indicators, including semiconductor manufacturing equipment imports from Taiwan and South Korea, related PMI sub-indices, import prices, and memory procurement and GPU rental prices. All leading indicators currently remain in high ranges seen since 2022, suggesting near-term growth prospects are still solid.
Inflation Erodes Real Investment Growth, Limiting GDP Impact
Despite the continued expansion of nominal AI investment, Goldman Sachs cautions investors about the erosion of real investment growth by cost inflation. Official U.S. data shows that 8% of the increase in nominal AI-related hardware spending so far in 2026 is attributable to cost inflation rather than real investment expansion. If this trend continues through the second half of 2026, the increase in AI-related spending will provide a smaller boost to real investment in 2026 than in 2025. Goldman Sachs also emphasizes that the overall impact of AI investment on U.S. GDP levels remains limited due to two measurement biases: first, U.S. national accounts do not count semiconductor purchases as investment goods; second, the high import content of AI hardware is already netted out in GDP calculations. This means that even with continued rapid growth in AI capital expenditure, its direct contribution to overall macroeconomic output faces structural constraints.
Comments