AI Investment and the U.S. Economy
Artificial intelligence (AI) has progressed rapidly in recent years, driven by rapid advances in large language models (LLMs). AI has led to a surge in business borrowing and investment in data centers, semiconductors, and other physical capital spending and has fueled a surge in related stock prices. As a result, investment in AI has been a significant source of economic growth in recent quarters. While a recent boon to the economy, AI investment has likely also put upward pressure on inflation, interest rates, and imports, slowing other parts of the economy. Further, if AI investment proves to be a bubble that subsequently bursts, it could be destabilizing for the economy.
Contribution to Economic Growth
One of the main ways in which AI has impacted the economy is in its effects on output, as measured by gross domestic product (GDP). GDP accounts for all expenditures within the economy, including businesses' investment in the plant (e.g., data centers), equipment (e.g., hardware), and intellectual property (e.g., software and research) needed to build AI.
The Bureau of Economic Analysis (BEA) does not a produce a specific line item for "AI investment" or similar category within GDP data. Therefore, estimates use proxy categories to account for AI investment, such as expenditures on computer equipment, software, or data center construction. Methodologies differ and likely either over- or underestimate the impact of AI investment, depending on their treatment of potential proxy categories, data centers, and imports, among other possible differences. On the one hand, estimates will be too high when categories include similar spending that is not AI related: For example, not all semiconductors are used for AI. On the other hand, a focus on AI investment is too narrow to fully capture AI's effects on growth: For example, it would not include growth in value added for non-AI firms attributable to those firms' spending on AI services.
Many aspects of AI investment in the economy currently rely on imports, as much of the hardware that is used to build data centers is produced abroad. For example, Federal Reserve researchers estimated that global trade in AI-related products (servers, graphics cards, and related parts) increased 65% in the first half of 2025 compared to the first half of 2024. GDP is calculated by subtracting imports. Therefore, capital expenditures related to the AI build-out are larger than the effect of the build-out on overall GDP growth. Figure 1 shows that AI capital expenditures have increased noticeably since the first quarter of 2025 but so have AI imports, resulting in a more modest acceleration of AI's contribution to GDP growth, as estimated by the Federal Reserve.
In addition to its effects on GDP, AI also affects productivity and the labor market, which indirectly affects economic growth in ways that are not directly measured and are harder to estimate. AI has been shown, in some cases, to improve task and worker productivity and potentially spur further innovation that begets more productivity growth. Potential impacts on the labor market are more uncertain, with the potential for some redistribution of jobs, as well as a potential further shift of the share of income from labor to capital. For more information, see CRS In Focus IF12762, The Macroeconomic Effects of Artificial Intelligence, by Lida R. Weinstock and Paul Tierno.
|
Source: CRS using Federal Reserve estimates. |
Potential Risks
While AI has undoubtedly been an engine of economic growth in the United States in recent years, it carries risks to the economy, including that AI investment is a bubble that could burst, could raise inflation, or could raise interest rates.
That the AI build-out has been a large driver of growth begets questions for economists and policymakers that do not have straightforward answers—is economic growth currently unbalanced and overly dependent on AI? So far, AI's contribution to growth has not been large enough to significantly change the overall growth picture; however, AI's contribution to growth was more significant in 2025 and the first quarter of 2026 than in previous years (see Figure 2). Additionally, Figure 2 could understate AI's contribution to growth because it does not consider any multiplier effect that occurs when businesses and individuals spend income that they received from AI purchasers.
On the other hand, Figure 2 may also overstate AI investment's effects on growth since it may also be increasing inflation, interest rates, and imports, thereby crowding out other economic activity. In other words, other parts of the economy would grow more rapidly in the absence of the AI investment boom.
Interest Rates and the Current Account Deficit
A great deal of the AI build-out is being funded by debt. This amount of borrowing could increase overall interest rates and the current account deficit (colloquially, the "trade deficit") in turn. In economic parlance, investment must be financed with saving, and therefore if domestic saving is insufficient, foreign saving must make up the difference. This net inflow of foreign capital results in a current account deficit. Although the deficit has increased somewhat during the AI investment boom, Federal Reserve Bank of New York research attributes this to the larger government budget deficit, finding business saving and investment rates relatively unchanged, with higher AI investment offset by a decline in other business investment.
|
Source: CRS using Bureau of Economic Analysis and Federal Reserve estimates. Note: This figure shows only one estimate of AI investment's contribution to growth. Other sources will yield different results. Bars indicate percentage points; the solid line indicates percentages. |
Other business investment may be falling because this large amount of debt issuance has increased demand for capital and is thus likely also contributing to the increase in long-term interest rates throughout the economy, raising financing costs for businesses, households (e.g., through higher mortgage rates), and the government. This increase in rates make U.S. investments relatively more attractive to foreigners, resulting in a larger current account deficit.
Inflation
AI investment may be contributing to near-term price pressures as demand for AI-related goods such as chips has increased, potentially causing bottlenecks that cause their prices to increase. Data centers may also contribute to increases in the demand for, and therefore price of, electricity. Goldman Sachs estimates these factors will cause inflation to rise by 0.5 percentage points at its peak. To avoid this causing an increase in overall inflation, other prices would need to fall, either because AI allows other goods to be produced more inexpensively or because fiscal or monetary policy is tightened to reduce overall demand. To date, tightening has been modest, and policymakers have not successfully reduced inflation to the Federal Reserve's 2% target. These factors, while affecting inflation in the short term, would not be expected to continue to do so in the medium term, at which point AI's inflationary effects would be more dependent on the extent of productivity gains seen from its dissemination and use, for example.
The 119th Congress has shown interest in these issues. For example, the Ratepayer Protection Act (H.R. 9340) would require state utility regulators to consider standards to hold "large-load" customers responsible for higher energy costs. The bill passed the House, but a cloture vote failed in the Senate.
Investment Bubble
If AI investment and growth is sustainable, its effects on the growth of the rest of the economy may be acceptable for the economy as a whole—throughout history, new technologies have crowded out less advanced industries. But if there is an AI investment bubble, that bubble could burst and cause significant harm to the economy—especially if other parts of the economy could not generate growth to pick up the slack. The AI industry has a few characteristics that could make it prone to a bubble, meaning that the actual and planned level of AI investment will not ultimately prove to be financially viable:
- AI requires large physical capital spending to make further improvements, and that capital depreciates rapidly.
- Revenues are currently small relative to investment spending but are projected to grow rapidly.
- AI is a general-purpose technology, which could have unforeseeable applications across many industries, so plausible projections of future AI demand vary greatly.
- The AI build-out is financed by a large amount of debt.
- Projected revenue growth has led to high equity valuations for leading AI firms.
- There may be a "winner takes all" aspect of the industry, in which a small number of firms may end up capturing most of the market share—and the rest may go out of business.
- AI investment sometimes involves "circular financing," (financing is between AI-centric companies that also do business with one another), which may be artificially increasing demand.
Together, these factors carry a risk that if future AI revenues do not grow at their rapid projected pace for some or all of the leading firms, current AI investment spending will prove to have been a bubble that could subsequently burst, leading to a devaluation of AI firms' equity, debt, and capital assets. In that case, equity and debt holders could be exposed to large and unexpected losses, including through bankruptcy or default. The magnitude of potential losses could decrease economic growth through reduced investment spending and reduced consumption via a wealth effect (as investors bearing losses reduced their consumption spending in response) and might cause financial instability, with spillovers to the global economy.
Financial bubbles are difficult to accurately identify ahead of time, and an AI bubble's existence is hotly contested between skeptics and optimists. For example, a Bank for International Settlements working paper questions the sustainability of the AI boom, estimating overinvestment of 1.5 times the efficient level. Losses are expected to be greater as debt grows.
Despite these potential warning signs, many believe that the AI investment boom is likely to continue for now, suggesting that there is investor confidence in AI's prospects. For example, Goldman Sachs predicts "robust" short-term growth, with AI investment increasing to 2.8% of GDP in 2028, which is consistent with past technology build-outs.