Investors brace for slower AI spending growth from hyperscalers
Some investors are cutting exposure to AI chipmakers and shifting toward hyperscalers, software and sectors expected to benefit from AI adoption. The move comes as forecasts suggest Big Tech spending growth on AI infrastructure may slow in coming years.

WASHINGTON: Some investors are beginning to reposition for a deceleration in spending growth by the biggest technology companies on artificial intelligence infrastructure, even as enthusiasm for the sector remains intact.
For much of the past two years, investors favoured chipmakers and other companies tied to AI infrastructure on expectations that Microsoft, Amazon, Alphabet and Meta would keep stepping up investment in data centres. But that growth is now expected to ease. UBS estimates hyperscaler capital expenditure will rise 76% this year to $673 billion, then grow by 25% next year and by 6% in 2028, with total spending projected to approach $900 billion by 2028.
That outlook has led some active fund managers to reduce their exposure to semiconductor shares and increase positions in hyperscalers, which have significantly underperformed chip stocks during the rally. They are also adding software names and sectors seen as likely beneficiaries of AI adoption, including financials and healthcare.
Alexis Bossard, global equity portfolio manager at Edmond de Rothschild Asset Management, said a pause in ever-faster spending would change the picture for both tech buyers and suppliers.
"Once they stop increasing their capex, it will definitely be a relief for hyperscalers and a negative signal for the semi industry,"
Bossard said he had already reduced semiconductor exposure because he believes valuations have become too high compared with expectations. He has added to Amazon and prefers themes such as liquid cooling, cybersecurity and selected software companies.
"We have a massive underexposure to semis right now."
The Philadelphia Semiconductor Index, whose leading holdings include Nvidia, Broadcom, Micron, ASML and TSMC, has more than doubled over the last year despite falling nearly 18% from its June peak. That compares with an 11% rise in the equal-weighted S&P 500 and an 8% gain in Europe’s STOXX 600, which has less AI exposure. Bank of America’s July fund manager survey found 82% of respondents viewed semiconductors as the most crowded trade, while none said they were short the sector.
LFG+ZEST chief investment officer Alberto Conca has also sharply trimmed positions in memory-chip and equipment makers, while building stakes in hyperscalers and healthcare stocks. He has also purchased put options on selected semiconductor names.
After initially using their own cash to finance AI expansion, hyperscalers are increasingly turning to outside funding, raising concerns that pressure from capital markets could eventually limit the pace of spending. The corporate debt market has taken in billions of dollars in Big Tech bond issuance this year, although demand has softened. Apollo chief economist Torsten Slok said cover ratios, which measure investor demand relative to supply, dropped to below two times in July from nearly five times in February.
In June, the Basel-based Bank for International Settlements warned that weaker-than-expected returns could cause financing to retreat abruptly and turn the capital spending boom into a longer slump. Conca said mounting capital expenditure is consuming cash flow and argued the largest tech firms would become more disciplined.
"Cash flow is starting to be almost completely drained by capex,"
Empirical Research said there is an increasing mismatch between moderating growth in hyperscaler spending and the high revenue expectations attached to chipmakers and other AI infrastructure suppliers.
"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,"
Not all investors are turning cautious. Madeleine Ronner, senior portfolio manager at DWS, said commentary from hyperscalers during earnings season is still expected to support additional investment.
"The surprise would be if it's not like that,"
She also said buy-side projections for 2027 spending remain well above analyst estimates. DWS has taken some profit in semiconductor names after their strong gains but is still overweight the sector, while some funds have added exposure to industrial and electrical equipment companies following the pullback.
Another potential headwind is resistance to data-centre construction in parts of the United States. Empirical estimates that about 70% of projects face some level of opposition. On Tuesday, New York became the first US state to pause construction of large new data centres, imposing a one-year moratorium amid concerns over higher power costs, pressure on water supplies and the burden on local communities.
Even so, fund flows suggest investor interest in AI infrastructure remains strong. Morningstar data showed chip-focused funds drew a record $10 billion in net inflows through May, before increased volatility unsettled the sector in July.
Fidelity Investments director of global macro Jurrien Timmer said demand for computing power remains strong and recent swings may prove to be another temporary correction. He compared the pullback with repeated 20-30% declines suffered by leading stocks during the late-1990s internet boom before they resumed rising.
"The AI story is well known, it's ongoing, the earnings are still supporting the trend,"
Timmer said investors should still spread their bets more broadly, including toward sectors that benefit from using AI rather than only those building it.
"I want to participate in the boom, but I also want to protect myself in case that boom is overdone,"
Comments
No comments yet. Be the first to join the discussion!







