AI spending nears a turning point... and investors shift their bets

The near-vertical rise in shares of AI chip companies has faced disruptions, amid growing concerns about high valuations and the sustainability of the massive revenues these companies generate, as some investors have quietly begun repositioning themselves in anticipation of a slowdown in the spending boom, approaching a trillion dollars, which could benefit the giant cloud computing companies bearing the cost of this spending.

For most of the past two years, the prevailing trend was the opposite; investors rushed into shares of semiconductor and infrastructure companies on the assumption that Microsoft, Amazon, Alphabet, and Meta would continue to accelerate their spending on building data centers.

Massive spending approaches a slowdown phase

But this spending now appears set to slow; UBS expects capital expenditure by giant cloud computing companies to rise by 76 percent this year to $673 billion, before growth slows to 25 percent next year, then to just 6 percent in 2028.

Some active portfolio managers have already begun reducing their exposure to chip company shares while increasing their investments in the shares of the giant cloud computing companies themselves, whose performance has notably lagged behind the gains posted by chip companies. They have also turned to buying shares of software companies and sectors expected to benefit from the adoption of AI technologies, such as financial services and healthcare.

Alexis Bussar, global equity portfolio manager at Edmond de Rothschild Asset Management, who has already reduced his exposure to semiconductor shares because he believes they have become expensive relative to expectations, said: 'Once the cloud giants stop increasing their capital spending, that will be a relief for these companies and a negative signal for the semiconductor industry.'

Humanoid robots displayed at Ant Group's booth during the World AI Conference in Shanghai, China, July 17, 2026 (Reuters)

High valuations and risks of crowded trades

The Philadelphia Semiconductor Index, whose largest components include Nvidia, Broadcom, Micron, ASML, and TSMC, has more than doubled over the past year, even after falling about 18 percent from its peak in June, compared to an 11 percent rise in the S&P 500 Equal Weight Index, or gains of 8 percent in the European Stoxx 600, which has limited exposure to AI.

The Bank of America Fund Manager Survey for July showed that 82 percent of participants consider semiconductor stocks the most crowded trade in the markets, while none reported taking short positions on the sector.

This raises the question of how investors will position themselves if AI spending remains strong but no longer accelerates at a pace sufficient to support the expectations embedded in the valuations of AI infrastructure stocks.

Bussar has increased his investments in Amazon and favors areas such as liquid cooling systems, cybersecurity, and select software companies, adding: 'We currently have very low exposure to the semiconductor sector.'

Investors reposition

Alberto Conca, chief investment officer at LFG+Zest, has sharply reduced his investments in memory chip makers and chip manufacturing equipment, while building positions in cloud giants and healthcare stocks, and supported this view by buying put options on a number of semiconductor stocks.

After the cloud giants financed the first phase of AI infrastructure build-out from their own cash flows, they have increasingly relied on external financing, raising questions about whether capital market pressures might eventually limit spending growth.

Bond markets show signs of strain

The corporate bond market has absorbed billions of dollars in issues from big tech companies this year, and until recently, investors were eagerly buying them.

Torsten Slok, chief economist at Apollo, notes that coverage ratios, which measure demand for bonds relative to supply, fell to below two times in July, compared to about five times in February.

In June, the Bank for International Settlements, based in Basel, Switzerland, warned that any disappointment in returns could lead to a sudden pullback in financing and turn the capital spending boom into a prolonged downturn.

Conca said: 'Cash flows are being almost completely drained by capital spending,' adding that the cloud giants will become more disciplined in the pace of spending growth.

In this context, Empirical Research points to a growing gap between slowing capital spending growth on the one hand, and very high revenue expectations for chip companies and other AI infrastructure suppliers on the other, meaning one side must change.

The firm said: 'Either the capital spending path of the cloud giants is revised upward again, or the expected revenue growth for their suppliers will have to come from other sources.'

A person holding a baby-shaped robot at Fourier's booth during the World AI Conference in Shanghai (Reuters)

Meanwhile, Madeline Roener, senior portfolio manager at DWS, expects the cloud giants' comments during the earnings season to remain supportive of continued investment. She said: 'The surprise would be if that doesn't happen.' She added that institutional investors' expectations for 2027 spending remain much higher than analysts' estimates.

DWS has taken some profits in semiconductor stocks after their strong rally, but still maintains an above-average weight in the sector, and some of its funds have increased exposure to industrial and electrical equipment stocks after the recent pullback.

Regulatory hurdles threaten expansion pace

Rising local opposition to data centers in the United States could also slow spending growth. Empirical Research estimates that about 70 percent of data center projects face varying degrees of opposition.

On Tuesday, New York State became the first U.S. state to halt the construction of large new data centers, after imposing a one-year moratorium, amid growing concerns that the facilities driving the AI boom are raising electricity costs, depleting water resources, and increasing burdens on local communities.

Nevertheless, investor appetite for AI infrastructure remains strong; Morningstar data shows that chip-focused funds attracted a record net inflow of $10 billion through May.