Senator Bernie Sanders and Rep. Alexandria Ocasio-Cortez introduced legislation today that would ban construction of new data centers exceeding 20 megawatts until Congress passes comprehensive AI regulation. The bill cites concerns from tech leaders like Sam Altman and Geoffrey Hinton about AI risks, and polling showing most Americans are more worried than excited about AI development.
This is political theater that fundamentally misunderstands how AI infrastructure works. A 20MW threshold would block major cloud facilities while letting thousands of smaller GPU clusters operate unchecked. The real AI compute happening in research labs and startups doesn't need massive data centers — it needs dense, specialized hardware that can fit in converted warehouses. Meanwhile, the bill's demands for pre-release model certification and union-only construction read more like progressive wishlist items than serious AI policy.
As I noted when this story first broke, the legislation has zero chance of passing but signals growing infrastructure backlash. What's telling is how Sanders' office cherry-picked quotes from AI executives who've since changed their tune or built their own massive training facilities. The bill also calls for export controls on advanced chips to countries without similar rules — essentially every country except maybe the EU.
For developers, this matters less for the specific legislation than what it represents: policymakers are starting to target AI infrastructure without understanding it. If you're building AI applications, expect more regulatory uncertainty around compute access and data center availability, even if this particular bill dies in committee.