Artificial intelligence is driving almost all of America’s growth. In the first half of 2025, 92 percent of U.S. GDP expansion came from investment in data centers and information processing equipment—essentially the backbone of AI infrastructure. Without this surge, the economy would have grown by just 0.1 percent, according to Harvard economist Jason Furman.
Such concentration is historically rare and deeply revealing. It means that without AI, the U.S. economy would be stagnant. While every major innovation cycle has periods of overexcitement, the scale of AI’s contribution suggests that the country may already be living inside a bubble. The question is not whether it will burst, but how—and what will follow.
Economists estimate that corporate spending on AI infrastructure could exceed $500 billion annually by 2027, rivaling the GDP of medium-sized nations. Venture capital mirrors the frenzy: roughly two-thirds of all U.S. deal value this year has gone to AI-related startups, up from about a quarter in 2023. As a result, valuations have surged to extremes reminiscent of the late 1990s, when dot-com optimism blinded investors to the limits of the early internet.
That era offers a cautionary parallel. When the dot-com bubble collapsed, trillions in value evaporated, but the infrastructure it left behind—fiber optics, data centers, and web standards—paved the way for the next decade of digital growth. AI may follow the same pattern: an initial phase of speculative excess followed by a long, slower period of genuine utility.
The signs of a bubble are visible across the board. Many AI firms remain unprofitable, their valuations based more on promise than performance. Some chipmakers are expanding so aggressively that global supply may soon outpace realistic demand. Even power grids are being strained by data-center construction, an external constraint that no algorithm can fix. Meanwhile, productivity gains—the supposed justification for the boom—have not yet appeared in national data.
Still, the United States has always emerged stronger from its bubbles. The housing crash reshaped finance. The dot-com collapse birthed the platforms that now dominate the internet. Bubbles, in a sense, are how America experiments at scale. They test the limits of capital and optimism, often wasting both, but in the process they accelerate innovation.
If AI spending cools or corrects sharply, the immediate pain will be real: slower GDP growth, layoffs in tech-adjacent industries, and tighter credit conditions. But as history shows, the capital already invested will not vanish. The data centers, chips, and cloud networks will remain, forming the skeleton of whatever comes next—perhaps quantum computing, biotechnology, or a new form of digital productivity that has not yet been imagined.
The United States has a habit of inflating the future faster than it can arrive. The AI bubble, if it bursts, will be another reminder of that impatience. But it may also be proof of something enduring: the country’s uncanny ability to turn its excesses into infrastructure.