• Kevin Hassett, director of the National Economic Council, expressed optimism that the AI-driven data-center investment boom has room to run, but recent developments suggest electricity access, permitting, and community opposition are becoming the binding constraints.
  • Voters rejected data-center measures in 40 of 41 local ballot initiatives across 10 states, and roughly 40% of pending U.S. data-center capacity is planned in states where newly elected governors have pledged to pause or stop construction, according to Bloomberg.
  • Developers are adjusting: OpenAI committed to avoiding NDAs, disclosing community impacts, and ensuring households don't bear added electricity costs, while analysts estimate 75%–80% of 2027 deliveries are already under construction.

Hassett’s Optimism Meets Grassroots Resistance

Kevin Hassett, director of the National Economic Council, told CNBC (NASDAQ:VSNT) in an interview aired November 2 that there is “plenty of room” for data-center growth to continue—a headline that underscores the White House’s confidence in the AI infrastructure boom. But the latest developments paint a more nuanced picture: the main constraint is increasingly whether projects can secure electricity, permits, and community support, not whether demand for computing exists.

The political landscape shifted markedly after this week’s U.S. elections. Voters sided against data-center development in 40 of 41 local ballot measures across 10 states on November 3, according to CNBC. While the measures involved varying restrictions and shouldn’t all be read as outright construction bans, the results signal intensifying local opposition. Bloomberg reported on November 6 that approximately 40% of pending U.S. data-center capacity is planned in states where newly elected governors promised to stop or pause construction—exposure to potential restrictions, not a forecast of cancellation.

“What institutional investors like us are really focused on is regulatory stability,” said one industry executive, speaking on condition of anonymity. “The election results introduce a new layer of uncertainty.”

Developers Adapt to a New Reality

OpenAI moved quickly to address community concerns. On November 5, the company announced commitments to avoid nondisclosure agreements in data-center negotiations, disclose community impacts, and ensure households do not bear the electricity costs of its projects. It also pledged peak-demand reductions and local hiring goals. These are corporate commitments and proposed policy principles, not binding nationwide regulations, but they illustrate how community acceptance is becoming part of project economics.

OpenAI’s chief global affairs officer, Chris Lehane, acknowledged that some projects could be delayed. The company did not respond to a request for comment on whether the commitments would apply retroactively to existing facilities.

The tension between demand and deliverability is central. Continued appetite for AI computing does not guarantee that every announced facility can be financed, permitted, and connected to the grid on schedule. Recent reporting shows that concerns about borrowing costs and potential construction restrictions have weighed on AI-related shares, because slower building would reduce purchases of chips, power, and associated equipment. Those market moves reflect changing expectations—not proof that underlying AI usage has declined.

The effects are global. MarketWatch (NASDAQ:NWSA) reported November 5 that South Korea’s Kospi fell more than 5%, with Samsung (KRX:005930) and SK Hynix (KRX:000660) each down approximately 7%, as investors reassessed the AI infrastructure outlook. Neither company could be reached for comment.

The Power Crunch

An important industry shift is toward “speed to power”: obtaining dependable electricity quickly has become a competitive advantage. Grid-connection delays, equipment and labor shortages, and local approval are obstacles to expansion. Recent industry reporting describes developers in major European hubs facing grid-connection waits of up to a decade—a parallel example of computing demand running ahead of available power infrastructure.

The public debate centers on electricity affordability, water consumption, noise, transparency, tax incentives, and whether permanent local jobs justify the infrastructure burden. Opposition is not exclusively partisan: Reuters (NASDAQ:TRI) reported that some Republicans also turned against projects over utility bills, lighting, and perceived inadequacy of jobs or tax benefits.

Construction Already Underway Provides a Cushion

Near term, existing construction supports Hassett’s optimism. JPMorgan Asset Management (NYSE:JPM)’s Michael Cembalest estimated that approximately 75%–80% of AI data centers expected for delivery in 2027 are already under construction. He therefore sees the larger political risk in projects requiring approval for 2028, rather than an immediate halt to the entire buildout. That remains an analyst assessment, not a guarantee of completion.

Longer term, expansion may continue under more demanding conditions: developers paying their own grid costs, tighter water limits, labor requirements, and reduced tax incentives. Cembalest estimated that 37 gigawatts planned for 2028 could face outright pauses and another 12 gigawatts could face significant conditions. These figures describe potential exposure under his analysis, not realized losses.

This is not solely a U.S. problem. Resistance was already disrupting projects before this week’s election: Bloomberg cited Data Center Watch’s finding that at least 45 projects, valued at $68 billion, experienced disruption from local opposition in the second quarter of 2026. “Disrupted” includes delays and other obstacles, rather than only permanent cancellations.

The most useful reading of Hassett’s headline is therefore that substantial room for computing growth can coexist with a slower, costlier, and more geographically selective construction cycle. The decisive evidence will be actual permits, power connections, financing, and completed facilities—not announced spending alone.

Correction: An earlier version of this article misstated the date of the CNBC interview with Kevin Hassett. It aired November 2, not November 3.