纽约州率先冻结超大规模数据中心，美国AI基建按下暂停键引争议

The biggest threat to America's AI future is not a lack of innovation. It is the mistaken belief that slowing infrastructure somehow creates better infrastructure. Policymakers around the country are increasingly acting to put the brakes on data center development without weighing all of the alternatives that could allow the United States to keep its global competitive advantage. Just recently, New York governor Kathy Hochul signed an executive order imposing the nation's first statewide moratorium on new hyperscale data centers while state regulators develop a new framework governing how large facilities connect to and pay for energy infrastructure. Similar actions are unfolding nationwide, from temporary local bans and permitting pauses to broader legislative efforts aimed at slowing growth. New restrictions and moratorium proposals continue to emerge across the country, reflecting growing concerns about grid capacity, power costs, and community impacts. Concerns about the impact data centers may have on the environment and the electric grid are not misplaced. Communities are right to ask hard questions about energy consumption, environmental impact, land use, and quality of life. In many regions, the grid is already under stress, and the scale of new infrastructure can feel disruptive. Policymakers are increasingly focused on ensuring that residents and businesses are not asked to bear the costs of infrastructure needed to support rapidly growing computing demand. That concern deserves serious consideration. What is less visible is how quickly the industry is adapting in response and what’s at stake if moratoriums are enacted. Data center developers and energy providers are not standing still. They are moving beyond the traditional assumption that every megawatt must come from the electric grid. Across the industry, developers are exploring and implementing behind-the-meter distributed power generation, including fuel cells, alongside hybrid configurations such as battery energy storage systems (BESS), that reduce dependence on constrained transmission infrastructure while improving reliability and accelerating deployment. These approaches can also help ensure that new loads are supported without shifting infrastructure costs onto other ratepayers. They are redesigning campuses to reduce land use and selecting technologies that limit, if not eliminate, water use. They are engaging communities earlier and incorporating feedback on traffic, noise, aesthetics, and environmental impact, while offering technologies to address concerns. Fuel cells are one example, producing electricity without combustion and with substantially lower emissions, noise, and water use than conventional combustion-based generation. They also are much quieter than combustion power systems. Other approaches include co-located generation, microgrids, and emerging technologies involving hydrogen and carbon capture. But adaptation takes time. Data centers and the energy systems that support them require years of planning, permitting, financing, construction, and infrastructure development. Yet many policy proposals assume the opposite. Moratoriums are often framed as temporary pauses to study impacts and develop better solutions. At first glance, that approach may seem reasonable. The risk is that abrupt pauses interrupt the very process through which better solutions are emerging. Projects that were being redesigned to address local concerns are halted. Investment is redirected elsewhere. When development pauses in one region, it rarely disappears. It moves. We are already seeing developers pursue opportunities in states that offer more predictable permitting pathways and clearer access to energy. Internationally, countries in the Middle East and Asia are actively aligning policy, infrastructure, and incentives to attract AI-related investment. History suggests this should not surprise anyone. Capital tends to flow where conditions support investment. Data center development is no different. Moratoriums do not pause demand for computing. They simply determine where that demand is met. And that demand is accelerating at a pace unlike any technology transition in modern history. ChatGPT reached 100 million users in just two months, a milestone that took social media platforms years to achieve. Today, artificial intelligence powers internet search, healthcare, logistics, software development, financial services and many of the digital tools millions of Americans rely on every day. What is at stake extends far beyond any single project or community. The United States is building the physical infrastructure that will underpin the AI economy for decades to come. These facilities are more than warehouses filled with servers. They are the factories of the digital age, producing computing power that supports scientific research, manufacturing, defense systems, cybersecurity, financial markets, and critical infrastructure. This is increasingly a national security issue. The countries that can build and power advanced computing infrastructure will be best positioned to lead in AI, cybersecurity, and technological innovation. As America debates whether to slow development, global competitors are moving quickly to build. At the same time, federal policymakers are seeking to accelerate deployment of AI infrastructure, creating a growing contrast between efforts to speed development and efforts to slow it. The debate should not be whether America builds this infrastructure, but how we build it faster, cleaner, quieter, and in partnership with host communities. The solutions are already emerging. Moratoriums may feel like caution. In practice, they risk becoming a pause button America cannot afford to press.