What is a data center?
A data center is a physical building housing servers, cooling systems, and power infrastructure. Three types exist:
Regional or mid-size centers (50K–250K square feet) serve local needs: a university, a government system, a company’s internal operations. Power draw: 10–20 MW. Water: 100K gallons per day.
Enterprise centers (250K–500K square feet) serve larger organizations or multiple customers. Power: 20–50 MW. Water: 300K–500K gallons daily.
Hyperscale centers (1M–4M+ square feet) are what Google, Microsoft, Meta, and Amazon build. One facility uses as much power as 100,000 homes. Power: 50–100+ MW. Water: 1–5 million gallons per day, equivalent to a town of 30K–50K people. Meta’s Louisiana facility spans 3,650 acres (twice New Orleans’ airport, four times Central Park).
What tribes actually need falls in the first category. What’s being proposed sits in the third.
Current need vs. speculative infrastructure
The infrastructure tribes use now is real: enrollment systems, permitting platforms, health records, school systems. Navajo Nation, one of the largest sovereign nations, operates comprehensive schools, healthcare, and government services on maybe 15 MW peak demand. Even with casinos running real-time gaming systems, actual tribal computing needs are orders of magnitude below hyperscale capacity.
The data centers being proposed aren’t for tribes. They’re being built for:
AI training and inference at scale. Running ChatGPT, Claude, or other models requires enormous compute. Query processing alone justifies building massive infrastructure. Hyperscale centers exist to process billions of queries globally, not to serve one tribal nation.
Mass surveillance data storage. Tech corporations collect and store vast amounts of data extracted from all of us: location history, browsing patterns, app usage, search queries, communications metadata. That data lives in data centers. Hyperscale infrastructure is expanding partly to house this surveillance dataset as it grows. Tribes are being asked to host the physical space for infrastructure extracting data from Indigenous people and everyone else.
Corporate growth projections. Revenue models for tech companies assume growth. New data centers aren’t justified by current use. They’re built based on where analysts project demand will be in 3–5 years. Corporations bet that more people will adopt AI, that AI workloads will expand, that compute demand will surge. Tribes get asked to provide land and water to back corporate growth bets.
How siting works
- Developer identifies a potential site (usually where land is cheap, water is plentiful, and renewable energy is nearby or accessible)
- Land acquisition or lease negotiation with the tribe or property owner
- Utility agreements for power connection and water supply
- Environmental review (NEPA, state environmental laws)
- Permitting from tribal and/or county/state authorities
- Construction and operation
Tribal consultation should happen early. It often doesn’t. Many proposals move through environmental review with minimal tribal input until suddenly they’re seeking permits. By then, deals are half-done.
The NDA problem
This is where sovereignty gets circumvented.
When a tribe negotiates with a corporation about siting a data center, the corporation insists on confidentiality: non-disclosure agreements (NDAs) prevent the tribe from revealing the terms, the water commitment, the promised revenue, the environmental impact studies, or even the fact that negotiations are happening.
The stated reason: competitive sensitivity. The real effect: silence.
Elected tribal leaders can’t share terms with their members. The tribe’s community can’t vote or deliberate on whether to host the infrastructure. Members who ask “what was decided?” get told “it’s under NDA.” Public process becomes impossible because the information is contractually hidden.
This is how sovereignty gets overridden without the appearance of override. The tribe’s government says yes, but the tribe’s people never got to say anything. The NDA does the work of preventing consent.
In November 2024, Honor the Earth and the Indigenous Environmental Network showed up at the National Congress of American Indians (NCAI) annual convention. NCAI had been promoting generative AI and hyperscale data centers across Indian Country without disclosing the impacts. They refused to meet with organizers raising concerns. They’d planned to pass a resolution supporting AI expansion. The only response was organizing outside the convention: a mural, a rally, a press conference. NCAI tabled the resolution. But their silence beforehand shows the pattern: promote expansion, withhold information, prevent public deliberation.
Why this matters to tribal governance
Water sovereignty. Water is relation. It connects the land, the people who depend on it, the generations before and after. A hyperscale data center in a water-scarce region commits millions of gallons daily to cooling corporate servers instead of supporting the community. In the Southwest, where many tribes sit, water is jurisdictional sovereignty. Deciding how water gets used is deciding what the land is for.
Energy and renewable resources. Hyperscale centers locate near renewable energy (wind, solar, hydroelectric) because those sites are cost-effective and marketable to climate-conscious customers. When a data center absorbs tribal renewable energy, it means that energy isn’t available for tribal energy independence. The tribe generates the power; the corporation uses it.
Employment narrative vs. reality. Corporations promise jobs. Reality: hyperscale centers employ hundreds, not thousands. The jobs are operational and technical, concentrated in a few roles. Most employment is construction, which ends when the building is done. Long-term operational staffing is small, and much of it requires expertise corporations bring in from outside.
Precedent for future decisions. Once a tribe says yes to hosting speculative infrastructure, it sets precedent. The next proposal becomes easier to approve. The next corporation knows the approval process, knows the leverage points, knows the NDA strategy works. Refusal now is harder than refusal later, but it sets the boundary.
Seven-generation impact, forward and backward. Data is relation. The land holds knowledge from ancestors. The decisions made now about what the land is used for echo backward to that history and forward to generations who’ll inherit what gets left. If a tribe’s water goes to cooling servers for corporate profit, future generations inherit depleted aquifers and forfeited sovereignty, not renewable resources or independent energy.