Markets and Economy What a slower AI buildout may mean for markets

One of the bigger takeaways from the 2026 midterm elections was that people don’t like data centers. And it’s not hard to understand why. They consume enormous amounts of electricity and water. They can create persistent noise for nearby communities. Their growing demands have raised concerns about electricity costs. And increasingly, they’ve become a physical symbol of another concern voters have about artificial intelligence (AI) — technology being built inside data centers could eventually displace jobs.

There’s another side to the story.

Data centers can bring significant investment and construction jobs to communities. Perhaps more importantly, they can generate substantial property tax revenue that lessens the burden on homeowners. But that argument didn’t carry the day.

There were 41 data center-related ballot measures across 10 states in last Tuesday's election. Voters backed restrictions or rejected development-friendly measures in virtually all of them.¹ Opposition to data centers was an important campaign issue in some of the country's largest markets for their development. That matters for investors because the AI story has become about more than software and technological innovation. Chips, memory, compute, and power are all part of what has become an enormous physical infrastructure story.

Stock market midterm response

The market initially responded positively to the midterm elections last Wednesday.² One likely explanation was relief that contested outcomes and prolonged legal battles were largely off the table. But there was already an interesting divergence underneath the surface. The AI infrastructure trade struggled even as the S&P 500 Equal Weight Index reached a record high.³

Then came Thursday.

The S&P 500 fell 4.4%, its 43rd worst day in nearly 60 years, while the Nasdaq-100 suffered its 32nd worst day since the index began in 1985.⁴ The selloff was particularly severe among companies tied to the AI buildout. Nvidia fell more than 9%, AMD more than 17%, and the Philadelphia Semiconductor Index roughly 12%.⁵

Investors were suddenly confronting a new question.

Longer AI buildout?

What happens if the AI buildout takes considerably longer than expected? That’s not an unreasonable concern. Texas, Georgia, and Ohio are among the important markets for data center development,⁶ and the midterm election results raised questions about how welcoming some states and local communities may be to additional construction.

Concerns about a widespread moratorium on data center construction are likely overstated, in my view. But markets don’t need an outright ban to reconsider expectations.

If AI-related capital spending has become an important contributor to economic growth,⁷ (compounded by the fiscal stimulus from the One Big Beautiful Bill Act fading), and AI-related earnings growth has become an important contributor to corporate profits,⁸ then a slower buildout will matter. The market attempted to price that very quickly.

Putting AI reset into perspective

There are several reasons I’d be careful about extrapolating the past few days too far.

First, the market hasn’t been correcting from the kinds of extreme valuations typically associated with the end of a structural growth story.⁹ Valuations across many of the largest technology companies had already compressed this year because earnings growth outpaced stock price gains.¹⁰ Some of that disconnect reflected doubts about the durability of the AI trade. Some reflected concerns about how much money companies had been spending. And some reflected the pressure from higher interest rates.

Expectations had been high, but skepticism had hardly been absent. That matters. The market may need to adjust to a slower infrastructure buildout, but it doesn’t necessarily have to unwind the type of extreme valuations we would associate with a speculative bubble, in my view.

Second, a slower buildout may alleviate one of the existing market concerns. Many investors have become increasingly worried about the extraordinary amount of capital being committed to AI infrastructure and the potential impairment of free cash flow at some of the largest technology companies.

The slowdown in data center construction may ultimately impose greater discipline on capital expenditure. It may also potentially favor the companies that have already controlled enormous amounts of compute capacity. The hyperscalers¹¹ have scale, existing infrastructure, and access to capital that newer and more speculative participants don’t. Ironically, therefore, restrictions on new data center development may reinforce some of the competitive advantages of the largest incumbents.

Third, slower investment may take further pressure off interest rates. The moderation in AI infrastructure investment may slow broader economic growth, and put further downward pressure on US interest rates.¹² That may ease some of the pressure on valuations and reduce the need for the Federal Reserve to raise rates further, in my view.

And then there’s my most important point — the fundamental case for AI hasn’t disappeared.

Demand for AI has continued to rise.¹³ Businesses have continued to adopt AI.¹⁴ Technology has continued to advance. None of that has changed because of the elections last Tuesday or a market selloff last Thursday.

What has changed is the timeline for building the infrastructure required to meet that demand. Projects may migrate to states and communities more receptive to development. Some capacity may ultimately be developed on federal lands. States interested in investment, employment, and tax revenue associated with data centers may compete for projects that other communities reject. The incentives to build remain enormous, in my view. That changes the timeline, not necessarily the outcome.

Market implications

For the near term, I believe a slower AI infrastructure buildout could favor more defensive areas of the stock market and higher quality bonds. But it’s important to distinguish between an elongation of the AI investment cycle and the end of it.

Finally, the stock market has been positive in the 12 months following every midterm election since 1938.¹⁵ Greater fiscal discipline from the largest companies, reasonable valuations, and the prospect of a more accommodative Fed gives me reason to believe the next 12 months may not end that streak.