For years, investors treated almost any major artificial intelligence investment as good news. Spending more on chips, data centers, and AI start-ups often sent technology stocks higher. Recent market activity suggests that confidence may be weakening.
The Dow Jones Industrial Average recently rose more than 500 points while semiconductor stocks sold off sharply. The PHLX semiconductor index fell nearly 6 percent in morning trading, briefly pushing the Nasdaq-100 into correction territory, meaning it was at least 10 percent below its recent high. The broader Nasdaq Composite also approached a correction before recovering some losses.
That split is important. Investors did not abandon the entire market. Instead, they moved money away from some of the AI-linked companies that had led the rally and into sectors that had previously been left behind. The market appears increasingly nervous about whether enormous AI investments will generate enough profit to justify current valuations.
Great Results Are No Longer Enough
Some of the clearest warning signs have come from the market’s reaction to supposedly excellent corporate results.
Samsung reported what was described as the largest quarterly profit in the history of any technology company. Its second-quarter operating profit reached 89.4 trillion won, or roughly $58.4 billion. Revenue more than doubled, and the company beat analyst expectations.
Samsung shares fell 6.9 percent.
Zavier Wong, a market analyst at eToro, said the result acted “more like confirmation, and confirmation is what people sell into.” The reaction suggested that investors had already priced extraordinary growth into the stock and were looking for reasons to take profits.
Alphabet experienced a similar response. The company reported its most profitable quarter ever, yet its shares dropped 7 percent in one day. Investors focused on negative free cash flow of $5.9 billion and a capital spending forecast that had risen to as much as $205 billion for 2026.
The concern was not that Alphabet had suddenly become unprofitable. It was that the company was spending so aggressively that even record income could not calm investors.
The Korean Warning
South Korea may provide a more dramatic example of what can happen when an AI-driven market becomes concentrated and heavily leveraged.
The KOSPI lost 33 percent of its value from June through late July. It fell another 7.3 percent between July 27 and July 28, triggering a circuit breaker that temporarily halted trading. It was reportedly the eighth such halt of the year.
Samsung and SK Hynix make up more than half of the index and had fallen 11 percent and 12 percent, respectively. When concerns about AI capital spending increased, leveraged investors faced margin calls, forcing liquidations that also affected markets in Tokyo and Taiwan.
Tim Tolka argued that the United States has several of the same vulnerabilities. Fourteen companies make up 45 percent of the S&P 500, while the top ten companies account for 67 percent of the Nasdaq-100. Most are closely connected to AI.
The United States also has record margin debt of $1.53 trillion. Margin debt equals about 4.71 percent of GDP, a level compared in the material to 2000 and 2007, both periods followed by major market corrections.
The People Warning About a Bubble
Several analysts and financial figures now see growing risks.
Annie Lowrey wrote that AI-linked companies had gained $27 trillion in value over three years. She warned that technology companies must quickly produce much larger revenues and profits to support those valuations.
Sam Altman has reportedly acknowledged that the industry is in an AI bubble. The International Monetary Fund has identified a potential AI collapse as a financial stability risk that could reduce investment, tighten credit, weaken consumption, and disrupt trade.
Michael Burry warned that “The End is Nigh” and described major chip spending commitments by Samsung and SK Hynix as “the beginning of the end.”
Chinese hedge funds have also expressed concern. Wealspring Asset reportedly called global AI stocks a “super bubble,” while Shanghai Banxia Investment Management Center said that a possible trigger for a correction may have already appeared.
Ryan Cummings of the Stanford Institute for Economic Policy Research summarized the changing debate: “Right now, the burden of proof is on the skeptics.” But after a continuing stream of disappointing information, he said, the burden could shift to the optimists.
Tether CEO’s Four Warning Signs
Paolo Ardoino is the CEO of Tether, the company behind the widely used USDT stablecoin. He has identified four structural mismatches that investors should watch.
First, AI companies may be charging customers too little compared with the actual cost of providing computing services. Low prices can attract users, but they may hide weak margins.
Second, AI profits may take much longer to arrive than the industry’s enormous spending. Companies must pay now for data centers, chips, land, and electricity while waiting for future revenue.
Third, AI chips can become outdated within three to five years, while financing arrangements often assume a much longer repayment period. Companies may have to replace equipment before it has paid for itself.
Fourth, open-source models are becoming more competitive. Cheap or free alternatives could weaken the pricing power of companies trying to recover massive infrastructure investments.
Circular Financing Adds to the Risk
Another concern is that major technology companies are investing in AI firms that then spend the money on those same companies’ cloud services and hardware.
Alphabet has invested in Anthropic, which is also a major customer of Alphabet’s cloud business. Nvidia has invested in OpenAI and has reportedly discussed providing enormous financial backing for OpenAI data center projects and chip purchases.
These arrangements may help expand the industry, but they can make revenue and profit harder to evaluate. If one major company weakens, the effects could spread through a network of investments, loans, cloud contracts, and hardware purchases.
Why the Bubble May Not Burst Yet
Not everyone expects an immediate collapse.
Supporters of the AI boom argue that today’s largest technology companies already have profitable businesses and strong cash flows. Unlike many dot-com companies, they are not merely speculative start-ups.
China may also change how the correction unfolds. Chinese companies are producing cheaper models that increasingly compete with leading American systems. That could damage the economics of expensive U.S. AI infrastructure, but it could also expand AI adoption and prevent the entire global industry from collapsing at once.
The market has not proven that the AI boom is over. However, sharp semiconductor declines, record short positions, rising leverage, negative reactions to record profits, and growing concern about capital spending suggest that investors are no longer accepting the AI story without question.
The bubble may not have burst, but the market is finally testing its surface.
