Retail investors’ fear of an “AI bubble” appears to have fallen off after spiking this summer. It could mean the stocks have further to balloon before they ultimately top out. The number of U.S. and worldwide web searches for the term “AI bubble” peaked on Aug. 20 and Aug. 21, respectively, according to Google Trends data spanning the past three years. Searches for the term dramatically outpaced searches for “stock market bubble,” “AI boom” and “crypto bubble” at the time. Peak interest in AI bubble searches came shortly after a Massachusetts Institute of Technology report found that 95% of organizations are getting zero return despite funneling between $30 billion to $40 billion in enterprise investment into generative AI. Around the same time, Meta confirmed it paused hiring for its new AI division after going on a hiring spree, and OpenAI CEO Sam Altman also said investors seem to be “overexcited” about AI, and OpenAI’s long-awaited ChatGPT-5 model release failed to impress consumers . The danger of an actual bubble in artificial intelligence stocks remains, but history shows it likely has further to go. “Bubbles are not [a] neat linear process,” Deutsche Bank strategist Adrian Cox wrote in a note first pointing out the decline in the Google searches. What has happened before Cox noted that the Nasdaq Composite waded through gains and dips before seeing explosive growth in 1999 that led to its peak in March 2000, the top of the ‘dot-com bubble’ — but then the tech-heavy index wiped out lost nearly 78% of its value between 2000 and its trough in October 2002. The tech-heavy benchmark “carried on shooting into the stratosphere well after talk of a bubble became commonplace, doubling in the year to October 1999, then almost doubling again over the following five months until it turned,” Cox said. He added that other manias have followed similar trends, including the British railway boom-and-bust of the 1840s. During the peak year of spending in 1847, investors put the modern equivalent of over $1 trillion dollars into constructing public infrastructure, before railway shares hit their lowest level of that decade in 1849, according to a paper written by University of Minnesota emeritus professor Andrew Odlyzko. “Earlier booms and busts followed similar patterns … bubbles have a variable lifespan, with the South Sea bubble blowing itself out in seven months while the dot-com bubble took five years to pop,” Cox wrote, referring to the speculative financial crisis caused by major losses in British joint-stock company South Sea Company. What it means for the stocks Given previous bubble patterns, Wall Street strategists think AI stocks may continue ballooning higher before bursting. “We continue to think a bigger bubble will emerge from AI before it’s over and expect valuation-sensitive investors and those who thought US exceptionalism had peaked to get ‘stopped in’ over time,” Bank of America’s Nitin Saksena wrote Tuesday. GQG Partners also pointed out that AI stocks could go even higher as investors are stuck in a “TINA” — there is no alternative — trade, where tech giants remain the winners despite a major shift the firm sees within the sector. “For the first time in our firm’s history, we believe many large technology companies today — particularly those with meaningful roles in the AI infrastructure build-out — represent backward-looking quality. For much of the last 15 years, investors who compared the exuberant periods in the technology sector to the dotcom era have been repeatedly proven wrong. Is it different this time? We believe so,” the firm, which was historically overweight in the tech sector just a few years ago, wrote in a Sept. 11 post on its website. “In our view, the consequences of the current AI boom could be worse than those of the dotcom era, as its scale — relative to the economy and the market — is far greater,” the firm said. Going forward, MRB Partners strategist Salvatore Ruscitti believes investors should have higher-than-usual portfolio diversification to the rest of the U.S. equity market as well as in international names, where he said the risk-reward appears more favorable. Ruscitti pointed out in a Tuesday note that investors continue to face twin risks of ongoing AI exuberance and a highly concentrated broader market, with the top 10 stocks in the S & P 500 comprising over 40% of the index’s market cap. Investors should broaden exposure to the rest of the U.S. equity market, as well as internationally where the risk-reward trade-off is more favorable, he said. — CNBC’s Ryan Sammy contributed reporting. ( Learn the best 2026 strategies from inside the NYSE with Josh Brown and others at CNBC PRO Live. Tickets and info here . )
