AI Data Centers and the New Agora

Faculty Seminar with Olivier Sylvain, Professor of Law, Fordham University School of Law.

The project that I sketch here brings together ideas that have animated my scholarship over the past couple of decades. My recent book gestures toward the argument, with a more general non-law audience in mind. With this current project, I hope to set out a concededly ambitious normative claim about the limits of the prevailing regulatory approach to communications and information law.

The prevailing assumption among many, if not most, scholars of law and tech is that "code is law." This view asserts that policymakers should craft laws that promote innovation and user sovereignty because policymakers are likely to get information technology regulation wrong otherwise. 

This project means to show that the developers of communications infrastructure and consumer-facing networked devices and applications, as well as their corporate employers and financiers, should never be responsible for implementing public law norms. At least, there should be a strong presumption against delegating public lawmaking responsibility to them. The main reason: they cannot account for the distributional impacts of their products in ways that democratic legal process at least tries as a matter of course.  

The recent global movement against AI data centers, even as uncoordinated as it is, offers a productive challenge to the code-is-law framing. Developers of these facilities and their boosters assert that restrictions on data center development intrude on innovation and free speech. But data centers, this project will aim to show, are a general-purpose infrastructure  in the same way that power grids, roadway networks, water supply and drainpipes, and telecommunications are. If the proponents of regulatory forbearance prevail, all manner of general-purpose infrastructure would be vulnerable to "free speech" challenge by dint of their speculative connection to downstream AI activities. 

More importantly, municipalities have long protected local communities from environmental, public-health, and economic risks pursuant to land-use authority. The distinctive combination of earthly and quotidian harms associated with data centers – including extraordinary energy and water consumption, noise and air pollution, harmful emissions, and related infrastructure demands – falls squarely within that authority. This raw fact strongly suggests significant limitations to the code-is-law framing and, presumably, a path forward in the emergent age of AI.

Olivier Sylvain is a visiting professor at the Law School and the Open Institute for Digital Transformations, and is funded by the TIERED project, state support managed by the Agence nationale de la recherche (French National Research Agency) under France 2030 and by the European Union's NextGenerationEU programme, under reference "ANR-22-EXES-0014".