AI spending has become one of the biggest investment stories in the U.S. economy. Companies are pouring money into chips, data centers, cloud infrastructure and software as they race to build and deploy the technology.

But the broader economy has not accelerated in the way many may have expected it to. Real U.S. GDP grew at a 2.2% annualized rate in the second quarter of 2026, close to its 2.3% average since the start of 2025, even as AI-related investment has surged.¹ That leads to an obvious question: If companies are spending so much on AI, why isn’t the economy growing faster?

One reason is that the headline spending figures don’t tell investors the full story.

“The AI spending numbers are enormous, but the boom adds much less to U.S. growth than they suggest,” Senior Markets Economist at J.P. Morgan Private Bank Joe Seydl said. “Imports absorb much of the hardware spending, and higher rates squeeze investment elsewhere.”

In other words, the AI boom is real – but much of the hardware is imported, and AI spending appears to be crowding out investment in other parts of the economy. So far, it appears to be reshaping U.S. economic growth more than accelerating it.

The AI spending boom is real – but GDP math matters

The AI boom is not imaginary. Companies are spending heavily on chips, graphics processing units (GPUs), data centers, servers, cloud infrastructure and software – all of which can support business investment, a critical part of GDP.

But GDP measures what the U.S. economy produces, not simply what U.S. companies buy. If a large share of AI hardware is imported, or if AI spending pulls capital away from other parts of the economy, the direct boost to U.S. growth can be much smaller than the headline numbers suggest. We cover these reasons more deeply below.

Reason 1: A lot of AI spending leaves the US economy

The first reason for the apparent mismatch between AI spending and real U.S. economic growth is that AI is hardware-intensive. Building AI infrastructure requires semiconductors, GPUs, servers, computers, peripheral equipment and networking gear. Much of that equipment is produced outside the United States.

That matters because of how GDP is calculated. Business investment adds to GDP, but imports subtract from it. So, if U.S. companies spend heavily on AI hardware that is made overseas, the investment boost is partly offset by the import drag.

Since mid-2023, annualized private investment in computers and peripheral equipment has risen from about $150 billion to roughly $365 billion.² But net imports of those products have risen sharply as well, nearly matching the level of investment.

“In other words, almost every extra dollar spent on computer hardware (including the chips inside it) is going abroad," Seydl said.

That doesn’t mean AI spending is unimportant; rather, the direct growth benefit may show up more clearly in economies that produce and export critical AI hardware. In the U.S., imported equipment absorbs much of the near-term GDP boost.

Reason 2: AI spending can crowd out investment elsewhere

The second reason for the apparent mismatch may be that the AI boom is shifting investment around the economy rather than creating a large net increase in investment.

Economists may use the term “crowding out” to describe what happens when heavy spending or borrowing in one part of the economy reduces investment elsewhere – usually because it competes for capital or keeps financing costs elevated.³ In this case, the pressure is coming from AI: Heavy demand for chips, data centers and related infrastructure is making it harder for other, more rate-sensitive investments to compete.

That dynamic matters more when the economy is already operating near full employment, because capital, labor and construction capacity are not unlimited. AI spending appears less sensitive to interest rates than many other kinds of investment. Companies may feel they need to invest quickly to stay competitive, and GPUs can lose value quickly as newer technology becomes available.

By contrast, housing, commercial real estate and other long-lived capital projects are typically more sensitive to higher rates. If AI spending rises while housing or other business investment falls, the overall growth effect is muted.

That phenomenon helps explain why the net investment boost has been modest. Since early 2023, business investment in information processing equipment and software has risen by about 1 percentage point of GDP. But over the same period, investment in other parts of the economy has fallen by about 0.8 percentage points, leaving total investment as a share of GDP only marginally higher.⁴

In short, the AI boom is changing where investment goes, not necessarily increasing the total amount by much.

Why this isn’t like the 1990s tech boom

The 1990s tech boom may seem like a natural comparison, but today’s AI cycle looks different. Back then, technology investment rose alongside other types of investment, helped by strong capital inflows from abroad.

Today, there’s less evidence of a similar surge in foreign capital coming into the United States. Without a larger pool of funding, AI investment is more likely to compete with other uses of capital – making it harder for the boom to lift total investment and U.S. growth.

The bottom line: What this means for the US economy

The AI spending boom is real, and it may reshape the economy over time. But for now, the direct boost to U.S. growth appears smaller than the headline numbers imply.

Much of the hardware spending tied to AI goes toward imported products, which offsets the contribution to U.S. GDP. At the same time, AI spending is crowding out investment in other parts of the economy, especially in more rate-sensitive areas.

That doesn’t mean AI spending doesn’t matter, though.

“The boom is changing what U.S. growth is made of more than it is speeding it up," Seydl said.

Despite the more muted effect on GDP today, as AI technology gets broadly adopted over the next 10 to 15 years, J.P. Morgan Wealth Management strategists expect a 50% increase in productivity. That view helped drive our largest upgrade to G7 (a group of major advanced economies) growth in 15 years, raising the nominal growth (including inflation) forecast to 4.2%.