The AI bubble fears return as chip stocks face their latest reality check
Investors are not abandoning artificial intelligence, but they are questioning whether hundreds of billions of dollars in infrastructure spending can generate sufficient returns.
If there is one thing investors in AI and semiconductor stocks have become accustomed to doing over the past two years, it is panicking, and sending shares of the companies at the center of the boom sharply lower. That is exactly what happened on Tuesday, as concerns over AI and chip stocks triggered declines across Asian markets, led by the Seoul Stock Exchange, which plunged 11%. On Wall Street, the selloff reshuffled the ranking of the world's largest companies by market capitalization, with Nvidia's decline allowing Apple to reclaim the top spot.
The latest wave of anxiety has several triggers, although no single event represents a fundamental shift in the AI industry. One of the main concerns involves Nvidia itself: the cost of insuring against a potential Nvidia default recorded its sharpest daily increase since such contracts began trading actively, amid reports that the company is involved in AI infrastructure deals worth more than $750 billion. Investors are concerned that Nvidia may be taking on excessive financial exposure. The company's shares fell 5%.
Another trigger was the IPO of Chinese memory chip manufacturer CXMT, which surged 466% on its first day of trading on the Shanghai Stock Exchange, reaching a market value of $484 billion. Investors viewed the debut as a sign that competition in the global memory chip industry could intensify.
The memory chip market is currently dominated by three major players: South Korea's Samsung and SK Hynix, and U.S.-based Micron. The massive demand for advanced memory chips required for AI data centers has sent their shares soaring over the past two years. Now, investors fear that CXMT could capture some of the enormous value created by the AI boom. Samsung shares fell 13.4% on Tuesday, while SK Hynix declined 14.7%.
However, the market reaction appears to be driven more by fear than fundamentals. Demand for AI memory chips remains far above production capacity, and although Samsung and SK Hynix are investing heavily in new manufacturing facilities, South Korea announced this week that the two companies would be involved in investments and transactions totaling $950 billion, those facilities are not expected to become operational before 2027.
Moreover, CXMT currently lacks the ability to manufacture the most advanced memory chips required for AI data centers, meaning it is unlikely to directly challenge the leading players in the near term. Instead, the company is focused on less advanced chips used in consumer devices such as PCs, tablets, and smartphones. Earlier this week, reports emerged that Apple is working with the White House to obtain approval to use CXMT chips in devices sold outside the U.S.
The expansion of CXMT's production could actually provide relief to the broader technology ecosystem. The surge in demand for AI-related memory chips has contributed to shortages of more basic chips used in consumer electronics, pushing prices higher. Increasing Chinese production capacity could ease supply constraints, limit further price increases, and potentially allow manufacturers to reverse some recent price hikes.
But the market's concerns extend beyond CXMT. They reflect deeper questions about the sustainability of the broader AI investment cycle.
On one side, spending on AI infrastructure continues to accelerate. Google, Meta, Amazon, and Microsoft are expected to invest nearly $700 billion in data centers this year. Meanwhile, AI companies such as OpenAI, Anthropic, and SpaceX AI are committing tens of billions of dollars, with future infrastructure obligations reaching hundreds of billions.
Those investments are beginning to affect even the strongest companies' financial statements. In the second quarter, Google's capital expenditures reached $44.9 billion, exceeding the company's $39.1 billion quarterly profit and pushing it into negative free cash flow territory. Google CFO Anat Ashkenazi said the pressure would continue, with annual AI-related infrastructure investments expected to reach $195 billion-$205 billion.
"Free cash flow will continue to be under pressure as a result of our investments in technology infrastructure, which allow us to leverage the opportunities that AI creates," Ashkenazi said.
The uncertainty is amplified by the fact that many AI companies have yet to demonstrate business models capable of justifying the scale of investment flowing into the sector. Investors are increasingly questioning whether the current pace of spending can generate sufficient returns.
Adding to those concerns is the rise of circular financing arrangements across the AI ecosystem. In these deals, major technology companies such as Nvidia, Microsoft, or Amazon invest in AI startups like OpenAI or Anthropic at valuations reaching hundreds of billions of dollars. Those AI companies then use the capital to purchase chips, cloud capacity, or other infrastructure from the same companies that invested in them.
As long as AI demand continues expanding, the model can reinforce itself. But if demand slows significantly, the ecosystem could face a negative feedback loop: AI companies reduce infrastructure spending, hurting chip and cloud providers, while the value of strategic investments made by those same companies declines.
For Nvidia, investors are particularly focused on the financial commitments it may assume as part of large-scale infrastructure partnerships. Despite the concerns, the company continues to deepen its ties across the AI ecosystem. This week, reports emerged that Nvidia would provide OpenAI with a quarter-trillion-dollar guarantee linked to the construction of a massive data center project using Nvidia chips. The company was also reported to be investing $5 billion in Ilya Sutskever's Safe Superintelligence, with the deal including access to Nvidia's Vera Rubin platform.
For now, markets remain caught between two competing forces: unprecedented AI investment and growing fears that expectations have moved ahead of reality. Every new announcement can send shares sharply higher or lower.
The next major test will come with upcoming earnings reports from the technology giants, particularly Nvidia's results on August 26. If the numbers meet investor expectations, optimism could quickly return. But until the market determines whether AI spending is creating sustainable value or an unsustainable bubble, volatility is likely to remain a defining feature of the sector.
