Sanders's Data Center Moratorium Is Risky Strategy for AI Safety

Senator Bernie Sanders has announced forthcoming legislation calling for a moratorium on constructing new data centers, framing it as a response to the profound risks of artificial intelligence. In a video statement, Sanders cited concerns ranging from massive job displacement and threats to democratic institutions to the existential risk of superintelligent AI escaping human control. He described AI as the beginning of "the most profound technological revolution in world history," with potential impacts on warfare, education, and human identity.

However, a critical analysis published on LessWrong argues this strategy is both ineffective and politically risky. The author, Thomas Rodskog, contends that a unilateral U.S. construction freeze is unlikely to meaningfully slow AI development due to intense opposition from well-funded AI labs and venture capital firms, coupled with national security imperatives to compete with China. Even if enacted, a moratorium might only delay frontier AI capabilities by months, as capital would shift to optimizing existing compute and preparing energy infrastructure for a post-freeze boom.

The greater danger, according to the critique, is that the moratorium could backfire by associating the serious cause of AI safety with weaker environmental or populist political arguments, potentially generating a backlash that makes more substantive regulation harder to pass. The economic concerns about automation and the existential risk from superintelligence are not solved by a temporary slowdown. Instead, the author argues the real solution lies in developing an international treaty to ban superintelligence development, verified through monitoring of data center compute—a model akin to nuclear non-proliferation—rather than a domestic construction ban that fails to address the global nature of the threat.