Midterms Could Deepen the K-Shaped AI Trade as Data Center Curbs Favor Big Tech Companies

The midterm elections could hand America's biggest technology companies an unexpected advantage from the restrictions on the data centers powering artificial intelligence, making their existing infrastructure more valuable while making expansion harder for smaller, heavily indebted rivals.

Democratic victories in gubernatorial races have sharpened concerns about permitting delays, electricity demands and the cost of building AI capacity. For investors, those pressures could deepen a K-shaped trade, with financially stronger companies advancing as more vulnerable operators struggle.

Wednesday's trading illustrated that divide. Microsoft and Amazon each gained more than 6%, while specialized AI cloud provider CoreWeave dropped about 10%. Micron fell 7%, and Nvidia declined 3.17%.

Texas has become a central test as Democratic Gov.-elect Gina Hinojosa has pledged to stop new data center approvals until lawmakers establish protections for residents. Projects already under construction would be exempt, potentially giving operators with facilities underway an advantage over developers still seeking permission.

Hinojosa's proposal follows restrictions imposed under Republican Gov. Greg Abbott, including a halt to new grid connections and an audit of projects. Her victory increases uncertainty about how long those constraints could last.

The investment implications extend beyond construction schedules. Companies with operating facilities can keep selling computing services while competitors wait for approvals. If demand remains strong and supply arrives more slowly, access to existing capacity could become more valuable.

James Sullivan, JPMorgan's managing director and co-head of global fundamental research, told CNBC's "Squawk Box Asia" on Thursday that financing costs help explain the split.

"The fact that you're seeing this bifurcation in equity performance is largely driven by the view that existing capacity is significantly more valuable now because of incremental supply constraints," Sullivan said.

He said borrowing spreads have widened substantially for neoclouds, specialized providers of AI computing infrastructure. That leaves them facing higher costs for additional capital than hyperscalers, the large cloud operators with broader businesses and more diverse revenue.

For debt-funded developers, delays can compound financial pressure. Equipment and construction require money before facilities generate revenue. A longer wait for permits or power can postpone customer deliveries while financing obligations remain.

The strain predates the election as CoreWeave, Nebius and Iren each lost at least 30% in October. CoreWeave's five-year credit default swap spread climbed above 1,100 basis points, signaling a higher price for protection against default.

In a research note, JPMorgan Asset Management said lower-rated neoclouds were particularly exposed to borrowing costs, energy prices and refinancing conditions. Hyperscalers have more internally generated cash and stronger access to credit, giving them greater room to absorb setbacks.

Much of the capacity scheduled to open in 2027 is already being built, limiting the immediate effect. Projects targeting 2028 are more exposed to permitting uncertainty, Nomura said in a Wednesday note cited by the outlet.