Wire flash
US AI infrastructure investment could hit $10.3 trillion, exceeding railroad and electrification eras
Editorial responsibility
- No named human review is recorded for this page.
- Source reporting is collected, normalized, translated or condensed automatically when needed.
- Automatically published source-backed update
A new analysis by economist Stijn van Nieuwerburgh, based on Brookings Institution estimates, projects that US investment in data centers and AI infrastructure could total $10.3 trillion from 2025 to 2032, averaging 3.63% of GDP annually. This would surpass the investment intensity of past booms like railroads (2.24% of GDP). However, the forecast is uncertain; Goldman Sachs estimates 2026 AI investment at 1.9% of GDP. The investment is already reshaping the US economy: private data center construction spending hit $370 billion by July, while other building sectors declined. Data centers are competing for power, labor, and land, potentially crowding out manufacturing. The boom has created over 750,000 AI-related jobs since 2023, with median salaries of $180,000. Rising demand for chips and energy is contributing to inflation, with computer and semiconductor import prices up 20% year-over-year. Chicago Fed President Austan Goolsbee warned of wage and electricity price pressures. The massive capital expenditure, estimated at $4.2 trillion for five major cloud firms over four years, is increasingly debt-financed through opaque off-balance-sheet vehicles, raising financial stability concerns if AI revenues fall short. Fed Chair Kevin Warsh linked this borrowing to rising long-term interest rates.
Source report
The United States is undertaking a massive buildout of AI infrastructure at a scale never seen before. According to the latest estimates from the Brookings Institution, cited by economist Stijn van Nieuwerburgh, cumulative investment in U.S. data centers and related AI infrastructure could reach $10.3 trillion between 2025 and 2032. This would represent an average of 3.63% of U.S. GDP annually.
This level of investment already surpasses previous waves of infrastructure development, including railroads, highways, electrification, and the internet. Van Nieuwerburgh's estimates show that railroad construction in the late 19th century accounted for about 2.24% of GDP annually, while AI infrastructure investment could be roughly 1.6 times that figure. However, these long-term projections carry significant uncertainty. Goldman Sachs, for example, estimates that U.S. AI investment will account for 1.9% of GDP in 2026.
AI Construction Is Reshaping the U.S. Economy
The most immediate impact of this investment wave is visible in the construction sector. According to U.S. Commerce Department data, seasonally adjusted spending on private data center construction reached $37 billion as of July this year—an increase of approximately $9 billion compared to the first seven months of 2025. Over the same period, spending on other private construction projects—including housing, apartments, and commercial real estate—fell by roughly $46 billion year-over-year.
The expansion of data centers is also competing with other industries for electricity, labor, and land. The Richmond Federal Reserve recently noted that data center construction is intensifying local labor supply pressures. In one case, an aluminum smelter planned for Mississippi—expected to create about 1,000 long-term jobs—ultimately relocated to Oklahoma after nearby data centers consumed the electricity it needed.
Didi Caldwell, a site selection consultant, said that in some regions, data centers are driving up land costs, and described the trend as "crowding out manufacturing."
At the same time, AI construction is generating new employment opportunities. LinkedIn estimates that from 2023 through early 2026, the U.S. added more than 750,000 AI-related jobs. The median salary for these positions is approximately $180,000, significantly higher than the overall median of about $80,000 across all job postings. Since the start of 2024, data centers themselves have added roughly 117,000 jobs, not including the large number of temporary construction roles.
Investment Boom Begins to Affect Inflation and Financing Risks
The resource demands of AI infrastructure are already influencing U.S. price levels. Data centers are purchasing large quantities of storage chips and other equipment, tightening supply for certain products. In August, prices paid by U.S. importers for computers, peripherals, and semiconductors rose 20% year-over-year. These cost pressures are being passed through to some consumer electronics, contributing to broader inflationary pressures.
Labor and energy costs are also being affected. Chicago Federal Reserve President Austan Goolsbee recently warned that data center investment is pushing up wages in related industries. In some regions with a high concentration of data centers, electricity prices have also risen noticeably. AI construction is no longer just a capital expenditure issue for the tech sector—it is beginning to affect broader prices through equipment, wages, and energy costs.
Meanwhile, the massive capital outlays are increasingly reliant on debt financing. According to FactSet data, five major cloud computing companies—Alphabet (GOOGL.O), Amazon (AMZN.O), Meta (META.O), Microsoft (MSFT.O), and Oracle (ORCL.O)—are expected to spend a combined $4.2 trillion on capital expenditures over the next four years, with a growing share funded through debt.
Van Nieuwerburgh noted that tech companies often borrow from banks and private credit institutions through off-balance-sheet entities, and public information on such transactions is limited. This makes it difficult to assess the full extent of financial risk embedded in the AI investment chain. If future AI revenues fail to cover the debt incurred for data center construction, the resulting pressure could ripple through the financial system.
Federal Reserve Chair Kevin Warsh has also cited borrowing by large tech companies as one of the factors behind rising long-term interest rates. AI investment is thus entering the broader macroeconomic environment through three channels: resource demand, inflation, and financing conditions.
Source
金十数据Neutral / independent