AI Data Center Push Collides With Local and Tribal Opposition

Major artificial intelligence developers are turning to smaller data center projects as they race to expand computing capacity. Anthropic and OpenAI have pursued deals for facilities in the 20 to 30 megawatt range after securing larger multi-hundred megawatt and gigawatt sites earlier. Sources familiar with the talks said Anthropic has explored such agreements in the United Kingdom and Nordic countries, while OpenAI has looked at similar options in the Nordics and the United States.

These moves reflect practical needs in a competitive market. Different workloads require varied infrastructure, and smaller allocations let companies bring new capacity online faster than waiting for giant builds. OpenAI described its approach as building a diversified portfolio to match demand, weighing factors like performance, reliability, timing, and cost. The companies have already announced extensive infrastructure partnerships over the past year to train and serve their models.

In California, Governor Gavin Newsom faces decisions on legislation that would shape where some of these facilities can locate. Bills S.B. 886 and S.B. 887 advance data center development without granting tribes the same formal consultation role on water issues that another measure, A.B. 2218, would provide. The latter bill, backed by the Karuk Tribe, aims to address historical water inequities through required state consultations. Critics of the data center bills argue they leave tribes without equivalent standing despite potential impacts on local resources.

Some Native nations have responded by restricting hyperscale projects outright. The Cherokee Nation, Sault Ste. Marie Tribe of Chippewa Indians, and Seminole Nation have already banned such centers on their lands. Honor the Earth has tracked more than 100 proposed or operating data centers near tribal territories nationwide. Native voters represent a notable group in several swing states, and observers note that state-level handling of these issues could influence broader political calculations.

Market pressures continue to drive the search for sites regardless. Data centers support the computing backbone for emerging technologies that have already delivered productivity gains across industries. Regulatory layers that prioritize group consultations over uniform rules risk delaying projects that generate jobs, tax revenue, and innovation spillovers. Past patterns show that special carve-outs based on historical grievances often produce uneven outcomes rather than broad improvements, as economic activity responds more reliably to clear property rights and predictable permitting than to layered identity-based reviews.

Newsom’s expected actions on the paired bills will test how state policy balances these competing claims. The AI sector’s expansion depends on timely access to power and land, and repeated delays from fragmented approval processes tend to raise costs passed along to users. Evidence from prior infrastructure buildouts indicates that voluntary negotiations between developers and landowners produce more efficient results than mandates that embed one group’s preferences into every decision. Smaller deals now under discussion may offer one path around some of those bottlenecks while still meeting immediate demand.