Is $800 Billion in Capital Expenditure the Peak? Goldman Sachs: Actual AI Investment This Year Could Exceed $1 Trillion

Global AI investment is far larger than commonly perceived. Goldman Sachs' latest research report points out that the widely cited ~$800 billion capital expenditure forecast for hyperscale cloud service providers is systematically underestimated. When including investments from private companies and non-U.S. firms, and excluding non-AI-related spending, total global AI investment is projected to reach $1.019 trillion by 2026.

Goldman Sachs economists Joseph Briggs and Sarah Dong stated in their 'Global Economic Analysis' report released on Sunday (the 2nd) that the commonly referenced ~$794 billion hyperscaler capex figure both underestimates total global AI capital expenditure by ~$200 billion and overestimates U.S.-based investment by ~$200 billion.

After adjustments, Goldman estimates U.S. AI investment at ~$581 billion and global total at ~$1.019 trillion by 2026. Two cross-validation methods — upward revisions in public companies’ gross profit forecasts and official national accounts and trade data — point to global AI investment of ~$1.06 trillion and ~$1.002 trillion respectively by 2026, closely aligning with the main estimate.

For macro markets, this recalculation has direct implications: it means the scale and sustainability of the AI capex cycle are stronger than previously expected. Goldman’s leading indicators show recent growth momentum remains robust, though trade data from Taiwan and South Korea suggest a mild slowdown in investment growth for June and July.

The Goldman Sachs report identifies four fundamental flaws in using hyperscaler capex as a proxy for AI investment:

First, the metric ignores U.S. private company investments and capital expenditures by other listed firms playing key roles in the AI ecosystem. Goldman’s credit team data shows hyperscalers account for only 40% of AI-related supply in 2026.

Second, it completely omits investments by non-U.S. companies, especially those in China and other parts of Asia.

Third, U.S. hyperscalers had already exceeded $150 billion in capex before the AI boom, meaning part of current spending is unrelated to AI.

Fourth, U.S. hyperscalers operate globally, so a significant portion of their capex actually occurs outside the United States.

Based on these assessments, Goldman made multidimensional adjustments to standard hyperscaler capex data: incorporating capital expenditure forecasts for other listed companies in its AI investment basket, supplementing media-disclosed data from key private firms, adding capex from non-U.S. AI-related companies, and subtracting non-AI-related investment using 2022 capex as a baseline.