The United States should continue building the data-center capacity needed to compete with China in artificial intelligence. My view is that retreating from that investment would be the wrong response to the real difficulties of supplying power and earning local trust. The better response is to build the infrastructure properly and make its costs and obligations explicit.
Computing leadership needs an energy plan
The International Energy Agency’s 2025 Energy and AI report provides a useful historical baseline: it estimated that data centers used about 415 terawatt-hours of electricity worldwide in 2024, with the United States accounting for 45% and China 25%. Those are electricity-use shares, not a scorecard of model quality or a current measurement of who is winning the AI race. The report also identified reliable, affordable power and grid constraints as central to future development.
That distinction matters. Winning should mean that researchers and businesses can obtain dependable computing services, develop useful products and keep improving productivity. It should not mean celebrating a project announcement before the facility has power, customers and a workable operating plan.
Build capacity without shifting every cost to neighbors
I support a demanding bargain: let worthwhile projects move forward, but require a credible account of power supply, connection costs, water needs and reliability. Households and small businesses should be able to understand the proposed impact on their community. A developer’s promises should be specific enough to check after operations begin.
Faster decisions and stronger standards can coexist. A predictable review process is better than an indefinite queue. It is also better than approving every project on the assumption that all investment will automatically deliver the same public benefit. Local conditions deserve a real assessment.
Competition is broader than buildings
A data center is part of a larger system that includes chips, software, skilled workers, security and customer demand. More floor space alone does not prove leadership. Investment should be judged by usable capacity and the services it enables, including access for smaller companies that cannot build their own infrastructure.
The strongest argument for continued U.S. investment is practical: preserve the ability to develop and deploy useful technology here. That argument does not require hostility toward Chinese people or unsupported claims about any individual company. It requires a serious view of national capability and long-term competition.
My conclusion
Keep building, and make the supporting power system part of the plan from the start. America should compete to lead in AI with infrastructure that works for customers and host communities. Treating reliability and local accountability as design requirements strengthens that ambition.
Related AMS reporting: why infrastructure matters as much as innovation. Photo: Tobi &Chris / Pexels, illustrative solar panels, not a data-center project. This is opinion; it is not a forecast or investment recommendation.
Eric Kuvykin
Publisher and editorial director covering payments, fintech, merchant services, banking, AI, technology and operating strategy for small and medium-sized businesses.
