The AI Bubble May Deflate, But It Will Not Burst
By Dr. Brett M. Decker, Northwood University
It has been a wild year for artificial-intelligence stocks. Beijing’s flaunting of China’s AI progress, among other factors, sent American stocks sliding over the summer. August saw a selloff of AI stocks, with Nvidia, Micron and others hit hard.
The pessimists were quick to warn that the AI bubble was bursting. Doomsayers are always on hand, waiting for every hiccup on Wall Street or data center setback to proclaim the technology is about to pop.
Their concerns aren’t unfounded. About 45% of the S&P 500 market capitalization consists of AI-linked stocks. The seven biggest U.S. tech companies, overwhelmingly invested in AI, account for more than a third of the S&P 500’s total market cap.
These stocks have grown at a rate that seems too good to be true. The S&P 500 Shiller CAPE Ratio — a valuation measure that smooths out business cycles to help investors see if stocks are overvalued — hasn’t been this high since before the dot-com bubble burst in 2000.
Yet history doesn’t have to repeat itself. Rather than a bubble, AI may simply be a technology that is landing slower than expected. The market might not be imploding so much as readjusting as the tech industry succumbs to more realistic expectations. The result could be a stable AI landscape that is better for investors and the rest of us.
Tech entrepreneurs have been making grandiose claims for years about AI’s capabilities. Predictions that AI would kill half of all white-collar jobs or create a permanent class of unemployed welfare dependents were always far-fetched.
New technologies often bring doom-and-gloom predictions. The difference with AI is that Silicon Valley elites promoted some of these claims. They contributed to absurd narratives to get businesses on board.
The limitations of AI have become clearer since then. Companies that laid off workers expecting to replace them with AI are hiring them back. An MIT study found that 95% of all generative-AI pilot projects fail to deliver a measurable return on investment.
This dovetails with AI’s PR problem. Polls find growing public skepticism of the technology. Data centers have sparked opposition across the country while online “AI slop” and perceived threats to employment remain sore spots. The initial narratives about AI were too bullish and opened a gap between expectation and reality. The expectation repelled many Americans while the reality proved underwhelming.
That doesn’t mean the technology is doomed to disappoint. In less flashy fields — data extraction, research, workflow automation, data-pattern summarization, customer service — AI is making a difference. The U.S. government is integrating AI into everyday operations, streamlining tasks like fraud detection and permit approval. The Pentagon is adopting AI for targeting and other functions. Ukraine has demonstrated how AI implementation can make a difference in battle.
The U.S. tech race with China has accelerated AI innovation. Chinese AI models put competitive pressure on their American counterparts. The elephant in the room is Huawei, whose AI dominion is so extensive that the U.S. reportedly approved a merger between Hewlett Packard Enterprise and Juniper Network to challenge it.
Global AI competition backed by industrial policy seems like the future. The world is becoming both an arena and a marketplace for AI. Wherever the technology is going, it isn’t a passing trend.
AI critics make the mistake of assuming the industry is frozen in time and can’t surmount its present problems. Take data centers, which are controversial in part because people believe they use enormous amounts of water. Nvidia recently pioneered a coolant that the company’s head of sustainability said “largely solved” the issue of data-center water consumption. Microsoft has been moving in this direction too.
This is the future of AI: a gradual adoption process that works through its problems and integrates into society without triggering economic apocalypse. The markets are correcting toward this reality. The transformation of the economy is proceeding, but it simply is moving at a slower pace than we were led to expect.
About this piece
Dr. Brett M. Decker, the Endowed Chair of Leadership at Northwood University, is a New York Times bestselling author. He is a former editor and editorial page writer for The Wall Street Journal. A version of this article originally appeared in The Wall Street Journal. This essay is the cover piece for the 2026 November edition of When Free to Choose, Northwood’s signature publication dedicated to exploring the importance of free enterprise and other tenets of The Northwood Idea. Click here to receive When Free to Choose in your inbox!