The explosive rally in artificial intelligence chipmakers is hitting a rough patch, as worries about sky-high valuations and the long-term sustainability of massive spending have prompted some investors to quietly change course — moving away from semiconductor stocks and toward the very tech giants writing the checks.
For much of the last two years, the prevailing strategy was straightforward: pour money into chip and infrastructure companies, betting that Microsoft, Amazon, Alphabet, and Meta would keep ramping up spending on data center construction at a breakneck pace.
But that era of rapid acceleration may be nearing its end. UBS estimates that the combined capital expenditures of these major tech companies will climb 76% this year to $673 billion — but then grow by only 25% the following year, and just 6% by 2028.
Some active fund managers have already trimmed their holdings in chip stocks and are instead purchasing shares of the hyperscalers themselves, which have notably underperformed the chip sector’s rally. They are also moving into software companies and industries expected to benefit as AI technology gets adopted more broadly, including financial services and healthcare.
“Once they stop increasing their capex, it will definitely be a relief for hyperscalers and a negative signal for the semi industry,” said Alexis Bossard, global equity portfolio manager at Edmond de Rothschild Asset Management. Bossard has already reduced his exposure to semiconductor stocks, which he believes have become overpriced relative to realistic expectations.
The Philadelphia Semiconductor Index — whose top holdings include Nvidia, Broadcom, Micron, ASML, and TSMC — has more than doubled over the past year, even after falling nearly 18% from its June peak. That compares with an 11% gain in the equal-weighted S&P 500 and an 8% rise in Europe’s AI-light STOXX 600.
A July fund manager survey from Bank of America found that 82% of respondents viewed semiconductors as the most crowded trade in the market, and not a single manager reported holding a short position in the sector.
The central question for investors now is how to position if AI spending remains strong but no longer grows fast enough to justify the lofty expectations already built into AI infrastructure stocks.
Bossard said he has increased exposure to Amazon and favors areas including liquid cooling, cybersecurity, and select software companies. “We have a massive underexposure to semis right now,” he said.
LFG+ZEST CIO Alberto Conca has made sharp cuts to positions in memory-chip and equipment makers while building up stakes in hyperscalers and healthcare stocks. He has also backed that view by purchasing put options on certain semiconductor names.
After initially funding AI infrastructure buildout with their own cash reserves, the major tech companies are increasingly turning to outside financing — raising questions about whether capital market pressures could eventually put a ceiling on spending growth.
The corporate bond market has absorbed billions of dollars in Big Tech debt issuance this year, and investors had been eager to buy — until recently. Apollo Chief Economist Torsten Slok noted that cover ratios, a measure of investor demand relative to bond supply, have dropped to below 2 times in July, down from nearly 5 times back in February.
In June, the Basel-based Bank for International Settlements warned that if returns disappoint, it could trigger a sudden withdrawal of financing and transform the current spending boom into a prolonged bust.
“Cash flow is starting to be almost completely drained by capex,” said Conca, who argues that hyperscalers will be forced to become more disciplined about spending growth.
Research firm Empirical Research has flagged a growing disconnect between moderating capital expenditure growth and the still-ambitious revenue expectations built into chipmaker and AI infrastructure supplier stocks — suggesting something will eventually have to give.
“Either the capex trajectory of the hyperscalers will be upgraded again, or the revenue growth pencilled in for their suppliers will have to come from elsewhere,” the firm stated.
Not everyone is turning bearish. Madeleine Ronner, senior portfolio manager at DWS, expects upcoming earnings season commentary from hyperscalers to remain supportive of continued investment. “The surprise would be if it’s not like that,” she said, adding that buy-side forecasts for 2027 spending remain well above analyst estimates.
DWS has taken some profits in semiconductor stocks following their strong run but remains overweight in the sector, and some of its funds have added exposure to industrial and electrical equipment companies after recent price declines.
Growing community opposition to data center construction in the United States could also put a brake on spending. Empirical Research estimates that roughly 70% of data center projects face some level of local pushback.
New York this week became the first U.S. state to halt construction of large new data centers, imposing a one-year moratorium amid concerns that the facilities powering the AI boom are driving up electricity costs, straining water supplies, and placing burdens on local communities.
Despite these headwinds, investor appetite for AI infrastructure remains strong overall. Data from Morningstar shows that chip-focused investment funds attracted a record $10 billion in net inflows through May.
Jurrien Timmer, Director of Global Macro at Fidelity Investments, argues that demand for computing capacity remains robust and that the recent volatility may simply be another temporary shakeout. He drew a comparison to the late-1990s internet boom, when leading technology stocks repeatedly fell 20% to 30% before eventually pushing higher.
“The AI story is well known, it’s ongoing, the earnings are still supporting the trend,” Timmer said.
Even so, Timmer believes investors should diversify their exposure, noting that companies benefiting from AI adoption — such as those in the financial sector — may increasingly matter alongside the companies building the AI infrastructure itself.
“I want to participate in the boom, but I also want to protect myself in case that boom is overdone,” he said.
