There are two seats at a data center deal and only one of them compounds. Landlord or owner. Everything else in the term sheet is detail.

The Pitch That Keeps Arriving at Tribal Council

Over 100 hyperscale data center projects have been proposed on or adjacent to tribal lands in the past two years (Honor the Earth). They don't show up with a hyperscaler's logo on the folder, and that's on purpose. They arrive through subsidiaries and intermediaries, and they usually open on solar before pivoting to a campus-scale server facility that drinks millions of gallons of water a year and pulls enough power to light a small city. Microsoft, Google, Meta, Amazon and their downstream partners are all in this game, and they're all shopping the same list: large tracts of sovereign land with available water, available power, favorable tax treatment, and regulatory frameworks that haven't caught up to the scale of what's being proposed.

This is the most consequential infrastructure decision most tribal governments'll face in a generation, and it's arriving fast, wrapped in NDAs, economic development language, and a jobs number that may or may not survive contact with reality. The question to settle before anybody signs is whether the tribe ends up a landlord collecting rent on somebody else's operation, or an owner building infrastructure that serves the nation for the next seven generations.

That's the difference between this era's version of the Dawes Act and an act of genuine self-determination. Same paperwork, opposite outcome.

What the Hyperscalers Want and Why They Want It from You

The appeal of tribal land here is structural, not sentimental, and I'd rather say that plainly than pretend anybody's being courted. Tribal lands often sit right where the renewable resource already is: wind and solar across the Great Plains, the Southwest, the Pacific Northwest. Water on or next to reservations supplies the cooling hyperscale facilities demand, between 300,000 and 5 million gallons per day depending on facility size (Futurism). Tribal sovereignty creates a regulatory environment distinct from state and county jurisdictions, which means permitting, environmental review, and tax obligations can look nothing like what a developer faces in a typical municipal negotiation. And plenty of tribal nations don't yet carry utility regulation codes written for a facility of this scale, so the developer meets fewer constraints on water withdrawal, power consumption, and environmental discharge than it would building the identical campus in, say, suburban Virginia.

None of that's a feature the developer's offering the tribe. It's a feature the developer's extracting from the tribe's sovereign status, and that distinction sets the entire power dynamic of the negotiation. A corporation asks for land, water, and a favorable regulatory posture, offers a lease payment and a jobs promise, and calls the result a partnership. It isn't. It's a tenancy, and the tribe's supplying the inputs that can't be replaced while receiving the outputs that can.

The companies walking into these meetings know exactly what they're getting. I'm a lot less sure the councils know exactly what they're giving.

The Playbook: NDAs, Intermediaries, and the Solar-to-Server Pivot

The pattern's been documented across multiple tribal nations now, and it runs the same way every time (ICT News). A subsidiary or a Native-owned energy company approaches tribal leadership about renewable energy development, which sounds benign and community-beneficial because at that stage it mostly is. Conversations start on solar installations or a wind farm. Then the scope shifts, and the renewable buildout gets reframed as the power source for a hyperscale data center campus sitting on the same ground.

Before substantive details get shared, the corporation asks the tribal council to sign a non-disclosure agreement. That's the moment the whole thing inverts. An NDA makes tribal leadership accountable to a corporation's confidentiality terms instead of to the tribal citizens whose land and water are under discussion. Community members can't evaluate what they can't see. Council members who've signed one can't tell the people who'll live with the consequences what's in it.

The Seminole Nation of Oklahoma saw this playbook firsthand when a tech startup approached their Tribal Council with an NDA and a letter of intention to develop a data center on Seminole lands. The community's answer wasn't ambiguous: an emergency town hall on March 3, 2026, followed by a 24-0 Tribal Council vote on March 7 to enact a permanent moratorium on hyperscale data center development within the Seminole Nation (Native News Online). As Mekusukey Band Representative Glen Chebon Kernell put it at the vote: "Once we've used up and contaminated all of our water sources, this is it."

The Muscogee Nation's National Council rejected a data center proposal by a 4-11 vote after citizens organized town halls exposing that the proposed facility would sit on a 5,570-acre plot used for the nation's food sovereignty initiative (Moms Clean Air Force). A hyperscale data center on the ground set aside to feed Muscogee citizens. I don't need to write the joke.

Landlord vs. Owner: The Only Framework That Matters

The Department of Energy held a webinar in February 2026 titled "Beyond Land Leases: Harnessing Data Centers for Tribal Economic Development," and the title gives the whole thing away (DOE Indian Energy). It concedes that a land lease is the default offer being made to tribes, then gestures toward something better without ever naming what that something is.

So let me name it. Every deal falls somewhere on a line running from pure extraction to full sovereignty, and you'll locate a term sheet on that line in about five minutes.

The Landlord Position is where most proposals land. The tribe's leasing land, providing water access, offering regulatory advantage. The corporation builds, owns, and operates the facility. The tribe receives lease payments and a handful of construction and maintenance jobs. The data processed in that building belongs to the corporation's clients. The AI models trained on it belong to the corporation. The economic value the compute generates belongs to the corporation's shareholders. The tribe's left holding the lease, the water bill, and the environmental consequences.

That's the Dawes Act in digital form. A tribal asset, land, gets converted into a vehicle for non-tribal wealth accumulation through a legal mechanism the tribe technically agreed to.

The Owner Position is the one almost nobody's proposing to tribal councils, because it isn't in the developer's interest to propose it. The tribe builds or co-owns the infrastructure. The tribe's the one deciding what data gets processed on its sovereign land. The tribe captures the value of the compute instead of the rent on the dirt underneath it. And the tribe can serve its own needs, running tribal health data, emergency management operations, governance documents, and grant administration on tribally owned hardware inside tribal jurisdiction, out of reach of a CLOUD Act warrant because no U.S.-incorporated technology company ever takes custody of the information.

The Cherokee Nation's taking the careful route here, and good on them. Principal Chief Chuck Hoskin Jr. signed an executive order in February 2026 establishing a nine-member task force, led by Secretary of Natural Resources Christina Justice, to study the environmental and economic impacts of data centers on the 7,000-square-mile Cherokee Nation Reservation (Cherokee Phoenix). The report's due June 30, 2026. That's a sovereign government running its own due diligence on its own clock, which is a different activity from signing an NDA in a conference room and hoping the jobs materialize.

Where Local AI Fits: Sovereignty You Can Hold

The conversation about data centers on tribal land and the conversation about local AI infrastructure are the same conversation viewed from opposite ends of the telescope. A data center's somebody else's data processed on your land. Local AI's your data processed on your own hardware.

So before a council answers the hosting question, I'd ask a smaller one: what are our own compute needs, and are we meeting them? If tribal health services are sending PHI to cloud AI platforms, if the emergency operations center is running incident analysis through commercial cloud, if governance documents and grant narratives are moving across servers the tribe doesn't own or control, then the tribe's already sitting in the landlord position with respect to its own information, just paying a different corporation for the privilege, and paying by the token.

An on-site deployment puts that inference capability inside the building: air-gapped where the posture calls for it, running on sovereign land, tribal data under tribal jurisdiction, no token meter, no CLOUD Act exposure, and no corporate intermediary standing between a government and its own records. That's the owner position applied to the tribe's own work.

What I won't do is walk into a council chamber holding a configuration. I come sit with the enrollment officer, with the housing authority staff who rebuild that quarterly report by hand every time, with the emergency manager reading incident reports faster than one person can read them, and I shut up until I understand the work the way they run it. Discovery names the use cases already in the building. Then I recommend a stack, we deploy and configure it together on site, and I onboard the team on workflows and agentic orchestration built for a tribal government. The order matters. Reverse it and you've bought a rack nobody wants to touch.

The OCAP Principles demand physical possession of tribal data. NCAI Resolution NC-24-008 defines digital sovereignty as authority over both the data and the physical infrastructure through which it moves. A tribally owned AI server on tribally owned broadband is the only configuration that satisfies both frameworks at once. The compliance briefs here map how OCAP, the CLOUD Act, and on-premises cost play out for tribal nations.

The Strategic Framework: Five Questions Before Signing Anything

Whatever's on the table, a hyperscale campus or a partnership offer, these five decide whether the deal serves the nation or drains it.

Who owns the infrastructure after construction? If the answer's the corporation, you're a landlord and everything after that is rent. Ownership that reverts to the tribe after a defined period changes the economics fundamentally, but only if those terms survive the corporation's legal team, and that's a bigger if than it looks on the page.

What happens to the water? A single hyperscale facility can consume 5 million gallons per day. In the arid West, in drought-stressed watersheds, in communities that're already fighting over allocation, that's an existential resource question, not an abstract one. Every gallon a data center drinks is a gallon that doesn't move through the watershed, doesn't recharge the aquifer, and doesn't reach tribal citizens.

What data is being processed on your land, and who controls it? If the tribe can't see what's running on servers sitting on its own sovereign territory, it's traded land sovereignty for digital occupation. At a minimum, demand data governance provisions that give the nation authority over which categories of data are allowed to run inside its jurisdiction.

Does the tribe get compute access for its own needs? A partnership with no tribal compute capacity in it is a lease with better adjectives, not a partnership. If servers are running on tribal land, some of them ought to be running tribal government work.

What's the exit clause? Twenty to thirty year lease terms with no meaningful exit provisions outlast most council terms, most economic cycles, and most corporate commitments. The facility'll likely change hands two or three times before the lease expires, and every transfer dilutes whatever got promised at the start.

The Real Choice

The Department of Energy calls data centers a "big economic opportunity" for tribes, and that framing is incomplete in a way history should make tribal leaders allergic to, rather than wrong.

Every era of American expansion has offered tribal nations an "economic opportunity" that required handing over a sovereign asset in exchange for a revenue stream somebody else controlled. Land for annuities. Water rights for irrigation contracts. Mineral rights for royalty payments. Same structure every time: the tribe provides the irreplaceable thing, the corporation captures the compounding thing, and the tribe gets the residual.

Data centers on tribal land follow that pattern unless the tribe breaks it, and the only way I've seen it break is ownership. Own the compute. Own the data governance. Own the infrastructure. And for the tribe's own AI, own the server holding the most sensitive records in the government, because that's the one piece of this where sovereignty isn't a negotiating position.

That's what I build. Not data centers. Not hyperscale campuses. Tribally owned AI inference that sits on sovereign land, processes sovereign data, and answers to the government that bought it and nobody else. What it takes to get there depends entirely on what your work turns out to need, which is a conversation on site rather than a number on a web page. What it costs to give sovereignty away compounds forever.

Summary: Over 100 hyperscale data center projects have been proposed on tribal lands, arriving through intermediaries and NDAs and offering lease payments in exchange for land, water, and regulatory advantage. There are two seats at that table. The landlord seat repeats the extraction pattern of every previous era; the owner seat captures the value of the compute. For a tribe's own AI work, on-premises inference on sovereign land is the only architecture that satisfies OCAP's Possession principle and NCAI Resolution NC-24-008's definition of digital sovereignty at the same time.