Equity Is Not the Same as a Lease

Commentary on "Why First Nations should be all-in on AI data centres" · The Hub

The Hub argues that First Nations are unusually well positioned to host AI data centres, on the reasoning that these facilities need land, reliable power, fibre, permitting certainty, and willing hosts rather than large local labour pools, and that characteristics long treated as economic disadvantages, remoteness and low population density, become advantages in this particular economy.

The economic logic is sound. The framing carries a problem worth naming, and the phrase "all-in" is doing more work than the evidence supports.

The argument contains its own difficulty

The piece treats it as favourable that data centres do not require large local labour pools. Set that beside what is happening elsewhere in the country. Manitoba refused a project because the jobs did not justify the resource demands. Vancouver residents objected that these buildings employ almost nobody. The same characteristic is a liability in one conversation and an asset in the other.

Both cannot be true, and the resolution is straightforward: if employment is not the benefit, then ownership and revenue have to be, and the structure of the deal becomes the entire question.

Two structures, very different outcomes

The projects announced so far illustrate the range precisely.

Woodland Cree First Nation holds 51 per cent of a proposed 650 megawatt facility in Alberta, using an idle power plant for supply, with a development partner holding the remainder. That is control, and it produces returns for as long as the asset operates.

Upper Nicola Band in British Columbia has been consulting its community on a proposed facility on its lands, with revenue arriving through a lease.

A lease is rent. Majority equity is ownership of the cash flow. Over a twenty-year asset life the difference between those two structures is not incremental, it is categorical, and it is determined entirely at the point of negotiation, usually when a community has the least information and the developer has the most.

The instrument that closes that gap is capital. The Canada Infrastructure Bank now targets at least $3 billion in revenue-generating Indigenous infrastructure across four priority sectors, one of which is explicitly digital infrastructure and AI, and it lends to communities specifically so they can take equity positions alongside its own investments. Quebec's procurement weights community control above 50 per cent as a distinct criterion rather than treating any participation as equivalent. These mechanisms exist because rent and ownership are not the same thing and policy has stopped pretending otherwise.

"All-in" flattens a real disagreement

There is also no single Indigenous position on this, and treating there as one does nobody a service. Sturgeon Lake Cree Nation has raised concerns about a major Alberta proposal. Communities differ on whether a project belongs on their land, and they differ for the same reasons every other community differs: water, land use, what is being given up, and whether they were asked before or after the decision was effectively made.

A recommendation that First Nations should be all-in has the structure of advice and the effect of pressure. The defensible version is narrower: where a community wants this and negotiates well, the returns are substantial and durable, and the financing mechanisms to secure real ownership now exist. Whether a community wants it is not for an industry publication to settle.

The federal criterion creates a risk

Ottawa's sovereign compute programme lists Indigenous participation among its evaluation criteria, asking proponents to describe proposed benefits including ownership and co-development, and the consultation steps taken.

That is the right instinct and it introduces a hazard the sector should name honestly. When participation becomes a scoring criterion, the incentive is to produce the minimum arrangement that satisfies a reviewer. A letter of support is cheaper than an equity stake. A memorandum is cheaper than a signed agreement. Programme design determines which one gets written, and a criterion that does not distinguish between them will reliably receive the cheaper option.

The fix is the same one this series has argued for elsewhere: specify the test. Distinguish consultation from co-development from ownership, weight them differently, and require the arrangement to be documented rather than described.

What this means going forward

The Hub is right that a genuine opportunity exists, and right that the characteristics of this asset class fit communities that other industries have overlooked. Indigenous equity in AI infrastructure has gone from nothing to a recognised project category inside two years.

What determines whether that becomes a generational asset or another round of land rent is not enthusiasm. It is whether communities enter these negotiations with independent technical advice, access to capital, and enough time to walk away. Those are unglamorous conditions and they are the whole thing.

VOLTEDGE

Reference: "Why First Nations should be all-in on AI data centres" · The Hub · read the article