Over the past several months, Atlas Analytics has warned that technology valuations may have become increasingly disconnected from underlying economic fundamentals.
In our Big Trouble for Big Tech series, we questioned whether the extraordinary expectations surrounding artificial intelligence could justify the valuations of America’s largest technology companies.
But the implications of the AI boom now extend far beyond Wall Street.
The construction of data centers, purchases of advanced semiconductors, and development of computing infrastructure have become important drivers of investment in the U.S. economy.
Our latest analysis suggests that AI-related investment categories have accounted for a median 35.5% of quarterly U.S. GDP growth since Q2 2025.
That raises an increasingly important question: What happens to America’s economy when the AI investment boom begins to slow?
The Scale of the AI Investment Boom
The numbers are extraordinary.
According to Goldman Sachs research published in August 2026, investment in artificial intelligence infrastructure is expected to approach $600 billion this year, equivalent to approximately 2% of U.S. GDP.
Technology giants, cloud computing providers, and specialized AI developers are committing enormous sums to constructing data centers, purchasing semiconductors, and expanding the computing infrastructure required to power artificial intelligence.
Yet there is an important distinction between total AI investment spending and its contribution to domestic economic output.
Much of the advanced computing equipment purchased by American companies is manufactured overseas. Consequently, while these purchases appear in U.S. investment statistics, the corresponding imports offset their direct contribution to domestic GDP.
The investment boom is enormous. But not every dollar spent translates into an additional dollar of American economic output.
More importantly, the macroeconomic significance of this expansion depends not simply on how much companies invest, but on how quickly that investment grows.
How Much of U.S. GDP Growth Depends on AI?
In January, economists at the Federal Reserve Bank of St. Louis published research examining the contribution of AI-related investment to U.S. economic growth.
Their analysis considered four investment categories: information-processing equipment, software, research and development, and data-center construction.
The researchers estimated that these categories contributed approximately 0.97 percentage points to real GDP growth during the first three quarters of 2025, accounting for roughly 39% of total growth.
Atlas Analytics has extended this research through Q3 2026 using data from the Bureau of Economic Analysis (BEA).
Our analysis adapts the St. Louis Fed’s methodology, measuring the contribution of information-processing equipment, software, and research and development to quarterly real GDP growth. We exclude data-center construction, which requires separate Census Bureau data and represented a relatively small component of the original estimates.
The results are striking.
Across the six quarters from Q2 2025 through Q3 2026, these investment categories accounted for a median 35.5% of quarterly real GDP growth.
In Q3 2026 alone, they contributed approximately 1.23 percentage points to real GDP growth of 3.36%, equivalent to 36.6% of total growth (see the headline graphic above).
The quarterly figures also reveal considerable volatility.
In Q1 2026, the three categories contributed 1.42 percentage points to GDP growth. Their contribution fell to just 0.42 percentage points in Q2 before rebounding to 1.23 percentage points in Q3.
The share of GDP growth attributable to these categories has fluctuated considerably, even exceeding 100% in Q4 2025, when weakness in other components offset much of their positive contribution.
The median provides a more representative measure of their typical quarterly importance than an average distorted by that unusually weak GDP quarter.
These findings require an important qualification: our analysis measures broad categories of AI-related investment, not spending exclusively attributable to artificial intelligence. The figures also do not separately identify the offsetting contribution of imported equipment.
Nevertheless, the results suggest that a substantial share of recent U.S. economic growth has been associated with the technology investment boom.
What Happens When the Boom Slows?
This brings us to the central macroeconomic concern.
An investment boom does not need to collapse to stop supporting GDP growth. It only needs to stop expanding.
Consider the implications of our latest findings.
In Q3 2026, the three investment categories contributed 1.23 percentage points to annualized real GDP growth of 3.36%.
If their contribution had instead been zero, with all other components of GDP growth unchanged, the economy would have expanded by approximately 2.13%.
This is a mechanical illustration, not a forecast. It does not account for changes in imports, consumption, or broader economic feedback effects.
But it illustrates the importance of the investment cycle.
If AI-related investment growth moderates, one of the economy’s major recent sources of momentum could diminish. An outright contraction in investment could create a more substantial headwind.
The sustainability of this expansion increasingly depends on three questions.
First, can companies generate sufficient returns to justify the enormous capital expenditures required to build and operate AI infrastructure?
Second, will investors and lenders continue providing the financing necessary to sustain the expansion?
And third, what happens if interest rates remain elevated while expected returns on AI investment disappoint?
As we argued in Higher for Longer, monetary policy could remain more restrictive than financial markets anticipate.
Higher financing costs, combined with uncertainty over future AI revenues, could eventually force companies to reassess their investment plans.
If that happens, the consequences could extend beyond technology stocks to construction, equipment suppliers, utilities, and the broader economy.
A Storm Is Brewing
Artificial intelligence may ultimately prove to be one of the most transformative technologies in modern economic history.
But the long-term productivity benefits of AI adoption should not be confused with the near-term GDP growth generated by constructing its infrastructure.
Our updated analysis suggests that AI-related investment categories have accounted for a median 35.5% of quarterly U.S. GDP growth over the past six quarters, with a contribution of approximately 37% in Q3 2026.
This is a remarkable concentration of economic growth within a relatively narrow group of investment categories.
The question is no longer simply how much companies are investing in artificial intelligence.
It is what happens to the U.S. economy when that investment stops accelerating.
In Part II of A Storm Is Brewing, we’ll examine the debt and financing structures supporting America’s data-center boom, including the growing importance of private credit and project-level financing.
From there, we’ll investigate what a slowdown in AI infrastructure investment could mean for U.S. economic growth in 2027.
Because while artificial intelligence may transform the economy over the coming decades, the investment cycle supporting that transformation could prove considerably less durable.
Methodology: Atlas Analytics calculations using Bureau of Economic Analysis National Income and Product Accounts data, adapting the methodology of Rubinton and Patro (2026), Federal Reserve Bank of St. Louis. Contributions are calculated for information-processing equipment, software, and research and development. Data-center construction is excluded. The 35.5% figure represents the median of six quarterly ratios from Q2 2025 through Q3 2026, not the share of GDP’s level or a cumulative contribution to growth. Investment categories include non-AI activity, and gross contributions do not separately account for offsetting imports.
Atlas Analytics develops proprietary macroeconomic indicators and forecasts using satellite imagery, computer vision, and advanced statistical modeling.
Disclaimer: The analysis and opinions expressed here are subject to uncertainty and revision and do not constitute investment advice.

