Trump’s Energy Secretary Wants Data Centers to Cover the U.S.

Trump’s Energy Secretary Chris Wright says AI acceleration and the buildout of AI data-center infrastructure are the No. 1 scientific priority of the Trump administration, and the United States should lead the effort. Speaking in excerpts from TIME’s interview for a new AI-focused newsletter, Wright—who came to the post from the private sector—emphasized AI’s potential to accelerate scientific discovery, including cancer research and molecular design that could help turn some deadly cancers into manageable conditions in coming years.

On data centers spanning the country and the energy mix that would power them, Wright argues that un-retiring old coal plants matters for speed and capacity, even as the administration weighs broader climate and market considerations. “We won’t keep open all of them. But maybe the significant majority of all the plants that are slated to close are closing for political reasons,” he said. He added that adding net-generating capacity quickly requires avoiding new retirements and that AI demand has shifted the calculus. The Energy Information Administration noted in 2023 that many coal plants retired because they could not compete with highly efficient natural gas plants and low-cost renewables, but Wright suggested those dynamics could change with AI-driven demand. As data centers come online, he said, the largest near-term additions are likely to come from natural gas, with steps to bring conventional nuclear power online more rapidly and with solar continuing to develop even without subsidies. He suggested wind subsidies, long in place, may not be extended indefinitely, but stressed that solar and wind alone will not deliver large new AI capacity.

[IMAGE]

Regarding subsidies and public concerns about price spikes, Wright pushed back at critiques that funding cuts had kneecapped renewables. He said subsidies are winding down by design, and the question remains how to ensure enough electrons reach peak demand without driving up electricity prices. “There’s not a swimming pool in the back that stores electrons,” he quipped, underscoring the need for real-time capacity rather than storage miracles as data centers expand.

Climate worries also feature in the debate, and Wright pushed back on the idea that climate concerns should dominate policy. He argued that public perception of climate change is not the world’s biggest problem and cited broader risk comparisons he says are often ignored in public debate. While acknowledging the WEF’s Global Risks Report 2025 lists climate-related risks among the most severe in the longer term, he indicated a pragmatic approach to energy policy that prioritizes reliability and affordability as AI demand grows. He said the United States must expand electricity production to meet growing needs, even if that means accepting some tradeoffs on climate policy.

The core point, Wright said, is that AI will raise electricity demand and thus demand for new capacity. He asserted that the country’s energy strategy should be to grow generation—so that data centers can run at scale without forcing price rises for families and businesses. If hyperscalers need a gigawatt for a single data center, he said, they must bring a comparable amount of new power online and pay rates that reflect the cost of expanding the grid—rather than simply exporting demand onto a constrained system.

The episode sits within a broader dialogue about AI’s energy footprint. Separately, a UK assessment by the AI Security Institute said frontier AI capabilities are advancing rapidly across domains, including biology and cybersecurity, with models’ jailbreak detection times lengthening dramatically as safeguards improve. It also highlighted a rising prevalence of human–AI companionship and noted that Google recently rolled out Gemini 3 Flash within Search and other platforms, signaling continued acceleration in commercial AI capabilities. These developments underscore the tension between rapid AI progress and the practical realities of grid capacity, regulatory frameworks, and public acceptance as data centers proliferate.