In Economics 101, you learn that markets work when everyone has information, everyone has choice, and everyone bears the consequences of their decisions. The Goldfinch proposal fails all three tests.
When those conditions break down, economists call it a market failure. The Goldfinch project is a textbook case — five distinct failures, all compounding at once.
1. The Winner’s Curse
When 38 US states compete for the same data centre investment, they bid against each other with tax breaks. The winner is often the one who overpays the most. Economists call this the Winner’s Curse — winning an auction by paying more than the prize is worth.
Alberta is competing against every one of those states. And the province already made a key concession: the data centre levy is net zero by design. Alberta collects it, then gives it right back.
Research confirms: economic development incentives are a prisoners dilemma for jurisdictions. Everyone would be better off if no one offered incentives, but no one can stop. The winner loses, and taxpayers pick up the tab.
2. Private Benefit, Socialized Cost
Diode captures the profit from rezoning and land sale — the private benefit. The community absorbs the costs: noise, light, traffic, water depletion, permanent farmland loss — socialized costs.
Economists call this an externality: when a decision’s costs fall on people who didn’t make the decision. Economics 101. And because the developer doesnt pay these costs, they have no incentive to reduce them.
3. Information Asymmetry
Diode knows everything about the project. Residents know only what Diode chooses to share. Diode hires consultants who produce reports favourable to their application. Residents have no independent experts to verify claims.
This is a textbook market failure. The standard solution is an independent environmental impact assessment, paid for by the developer but chosen by the community. Alberta is waiving these. After rezoning, the development permit stage has no public hearing. The community loses its voice entirely.
4. No Competition
If one hyperscaler is the only employer on 576 acres, they have all the bargaining power. Where would 100 specialized workers go if the operator cuts wages or demands concessions?
The facility has no alternative use. You cant convert a data centre to a factory or a school. Compare that to 576 acres of farmland: it can grow any crop, support any farm family, adapt to any market. Farmland is flexible. A data centre is not.
5. No Exit
Once 576 acres is rezoned industrial, it will almost never revert to farmland. Once infrastructure is built, the County has sunk costs. Once the data centre is built, the community is locked in for its 15-25 year lifespan.
If the operator leaves, the County gets a specialized building nobody else can use. There is no decommissioning bond. Cleanup costs fall entirely on the taxpayer.
The Cumulative Effect
These failures compound. The community has less information, less bargaining power, less flexibility, and more risk than the developer at every stage.
This is why 142 opposition groups have formed across 28 US states. This is why $64 billion in large data centre projects have been blocked or delayed. These communities arent anti-development. They recognize a market failure when they see one.
What Real Market Conditions Would Require
- Full disclosure of all incentive terms
- Independent studies chosen by the community, paid for by the developer
- A bond or security for decommissioning
- A community benefits agreement signed before zoning
- A named end user bound to commitments
- A public hearing at the development permit stage
None of these exist in the Goldfinch proposal. Not one.
This isnt about being against progress. Its about demanding that progress follows the rules every other industry follows — costs arent dumped on the community, information is shared fairly, and Wheatland County families arent left holding the bag.
Learn the facts. Visit the facts page to see the full picture.
Get involved. Join your neighbours standing up for fair treatment and real accountability.