A forced slowdown might be exactly what some AI companies need right now.

With Democrats taking not just the U.S. House and Senate, but also the governorships of states such as Texas, Ohio and Georgia, investors are now contemplating more roadblocks emerging for the construction of data centers.

That could constrain AI companies from being able to keep growing at a breakneck pace. Anthropic, in its initial public offering prospectus, listed the inability to keep securing more computing power as one of its major risk factors.

But the market's major concern of late hasn't been AI's growth potential. It has been the fear that building the capacity to meet that growth would be so costly that even the companies that won the AI race would end up much less profitable than envisioned. Rising AI credit costs were a big factor leading AI stocks to sink in recent weeks.

So in that way, if Democrats' sweep into power is really going to spur a national slowdown in the AI computing power build-out, it could actually play into the hands of many players in AI, and maybe especially to the biggest "hyperscale" tech companies.

A slowdown even at the fringes of data center development can ratchet down cost pressures on supplies, labor and credit. And it would give anyone with an existing data center more confidence that the scarcity of their asset will make it more valuable down the road.

"If you're a data center that is already running, or have power in place, there's scarcity value," says Andrew Keches, who is tech, media and telecom credit analyst and head of high-yield research at Barclays. "And if construction slows, then maybe capital expenditures are lower, and that's also good for credit."

This dynamic helps explain what has played out in the markets since the election. Tech giants Amazon and Microsoft, which already control vast computing power and have the financial might to outbid for more, rallied more than 6% on Wednesday.

Meanwhile, shares of computing and data center companies with less sterling credit and fewer cash resources were down. The CoinShares Bitcoin Mining and Digital Power ETF, which tracks shares of many crypto companies that have pivoted to AI, was down about 9% on Wednesday.

So were shares of companies selling chips or other components in the AI race because they are vulnerable to a slowdown in orders. The PHLX Semiconductor index was down 3%.

Bonds, too, were showing a shakeout, rather than a wipeout, for AI computing. There were rallies for many bonds financing the construction of data centers not just for tenants such as Microsoft and Amazon, but also upstart computing provider CoreWeave, according to MarketAxess data. These bonds saw tighter yield spreads over benchmark Treasurys.

Some exceptions, however, included bonds funding projects in Texas, regardless of tenant. The political tide shift in that state might be making investors nervous about those projects crossing the finish line, or the cost of getting there.

Another interesting indicator: While shares of Nvidia and Broadcom were falling on Wednesday, a sign that there may be less demand for chips, their bonds were rallying, according to MarketAxess data.

This suggests that while investors might be pricing in some kind of slowdown in new AI projects, the market may also view the financing guarantees those companies have offered to others as becoming less risky.

Given how sharply many AI stocks and bonds have sold off lately, there could be room for more relief as Democrats' plans come into shape.

Even if a state were to implement or extend a moratorium on construction, for example, whether it would apply to already started or late-stage projects would be a crucial variable. Some bonds also have features that protect them at the expense of the tenant, such as rent payments that kick in at a certain date regardless of construction status.

And if new projects go to the firms most willing to make compromises, such as forgoing tax incentives, or making local investments, that again may just tilt the landscape toward the biggest or most-experienced developers.

The AI build-out will still be a fierce race, even if it is run at slower speeds.