Bloomberg and CNBC reported on July 8, 2026 that Meta has selected Sturgeon County, Alberta as the site of its first-ever Canadian data center — a campus projected to cost roughly $10 billion (approximately C$13 billion, as corroborated by BNN Bloomberg). The facility will carry approximately 1 gigawatt of power capacity, enough electricity to supply around 750,000 homes. It will be Meta’s 33rd data center globally.
Alberta won the bid on cold economics: natural gas priced at a steep discount to the U.S. benchmark, and a climate cold enough to slash the mechanical cooling costs that quietly consume a meaningful share of every hyperscaler’s operating budget. The province offered Meta something the U.S. grid increasingly cannot — dispatchable, affordable, large-scale power, available now, without a decade-long interconnection queue.
The timing lands inside Meta’s broader infrastructure acceleration. The company is simultaneously standing up a third-party cloudbusiness to sell excess compute capacity — meaning this Alberta campus is both an internal cost center and the early physical inventory of a future revenue line. The capex arms race, in other words, is doubling as a land-grab for infrastructure landlord status.
THE AI POWER CRUNCH — CONTEXT TIMELINE
2022–2023
Chip scarcity dominates AI buildout narrative. GPU allocation becomes the primary constraint discussed in earnings calls.
2024–2025
Power emerges as the binding constraint. U.S. grid interconnection queues stretch 5–10 years. Hyperscalers begin scouting internationally for available gigawatt-scale sites.
July 8, 2026
Meta announces ~$10B / 1 GW campus in Sturgeon County, Alberta — gas-fired, cold-climate cooled. First hyperscaler to cross the Canadian border for AI-scale power at this magnitude.
Next
Compute maps redraw around energy access. Alberta’s model — cheap gas, cold air, regulatory speed — becomes a template other jurisdictions will race to replicate or outbid.
The key insight: This is not a data-center story. It is an energy-arbitrage story. Meta is doing exactly what aluminum smelters did a century ago — following stranded, cheap, dispatchable power to its source — except the product being manufactured is not metal. It is intelligence at scale.
The Structural Read
The binding constraint on AI has migrated up the stack. In 2022 it was chips. In 2024 it became power. By 2026, the firms that secured gigawatt-scale sites with dispatchable energy are building a structural moat that cannot be replicated in two or three years — because the permitting, grid negotiation, and construction timelines are themselves measured in years. The Meta Alberta campus is, in that sense, a supply-chain hedge masquerading as a capex announcement.
What makes Alberta distinctive is the combination: natural gas available at a meaningful discount to Henry Hub, a climate that delivers free cooling for much of the year, and a provincial regulatory posture that accelerated the approvals process. These are the same variables that once made Iceland irresistible to Bitcoin miners — except Meta’s timeline is decades, not cycles, and the capex commitment is an order of magnitude larger.
The second-order move is the cloudbusiness angle. Meta is not simply building infrastructure for its own models. It is building sellable capacity. The logic maps cleanly onto what Amazon did in the early 2000s: build infrastructure for your own operational need, discover the infrastructure is a product in itself, and monetize the overhang. AWS was not a strategic pivot — it was a consequence of operational scale. Meta’s cloud ambitions follow the same structural grammar. When you are spending $10 billion on a single campus, the marginal cost of selling idle compute to enterprise customers approaches zero.
The Map of AI — Infrastructure Layer
“The infrastructure layer is where value is pooling in the AI supercycle. Compute maps are becoming energy maps. The winners will be those who can co-locate models with cheap, dispatchable power — and the losers will be those who assumed the grid would simply expand to meet demand.”
The risk embedded in this bet is real and worth naming plainly. A gigawatt of gas-fired compute is a fossil-fuel commitment at a moment when regulators in the EU, and increasingly in North America, are scrutinizing data-center demand curves with fresh urgency. Alberta’s grid, while currently accommodating, is not infinitely elastic. And the political economy of a foreign hyperscaler drawing down a province’s natural-gas reserves for AI inference — while Canadian energy prices remain a live domestic issue — will not stay beneath the public radar indefinitely.
Meta is making a calculated bet: that the value of owning AI-scale compute now outweighs the regulatory and reputational risk of the energy source that powers it. That may be the right call. But it is a bet, not a given.
Three Implications
IMPLICATION 1 — THE GEOGRAPHY OF COMPUTE IS BEING REDRAWN
Alberta just proved that energy economics can override proximity-to-talent or proximity-to-customers as the primary site-selection variable for AI infrastructure. Jurisdictions with cheap, dispatchable power and fast permitting — regardless of where they sit on the map — are now in the bidding for hyperscaler capex. The next decade’s data-center map will look more like a 20th-century industrial energy map than a 21st-century tech-hub map.
IMPLICATION 2 — THE CAPEX ARMS RACE IS ALSO AN INFRASTRUCTURE LANDLORD RACE
Meta’s cloud ambitions mean every dollar of AI infrastructure capex is simultaneously a bet on becoming an AWS-style enabler for the next generation of enterprise AI workloads. The companies that build the most physical capacity now — even at negative near-term returns — are positioning to collect rent from every company that did not. The precedent is unambiguous: Amazon built AWS because it had to solve its own infrastructure problem at scale. Meta is following the same playbook, one gigawatt at a time.
IMPLICATION 3 — GAS-FIRED AI IS A REGULATORY AND REPUTATIONAL RISK THAT WILL COMPOUND
One gigawatt of natural-gas-powered compute is not a transitional measure — it is a decade-long infrastructure commitment. As the EU’s AI Act, Canada’s own federal climate obligations, and U.S. EPA data-center scrutiny converge, hyperscalers betting on fossil-backed AI campuses are acquiring a liability alongside the asset. The question is not whether regulators will act, but whether Meta’s competitive advantage from cheap gas outlasts the window before they do.
Business Engineer Framework
The Map of AI — Infrastructure Layer
The AI stack has nine layers, and the infrastructure layer — physical compute, power, cooling, land — is where the deepest moats are forming in 2026. The Map of AI framework shows exactly where Meta’s Alberta play sits in that stack, why the infrastructure layer is capturing disproportionate value in the current supercycle, and which companies are positioned to benefit as compute maps become energy maps. If you want to understand where the AI economy is actually being built — not where it is being announced — this is the read.
Meta’s $10 billion Alberta campus is the clearest signal yet that the AI supercycle has moved from a softwarecompetition to a physical infrastructure competition — one where the scarce input is not code or even chips, but electrons, and where the winners are those willing to cross borders, sign gas contracts, and tie their AI roadmap to the grid. The excess capacity becomes a cloudproduct, the cold air becomes a cost advantage, and the natural gas becomes a regulatory liability. All three of those things are true simultaneously. That tension is the businessstory of AI infrastructure in 2026.
Gennaro is the creator of FourWeekMBA, which reached about four million business people, comprising C-level executives, investors, analysts, product managers, and aspiring digital entrepreneurs in 2022 alone | He is also Director of Sales for a high-tech scaleup in the AI Industry | In 2012, Gennaro earned an International MBA with emphasis on Corporate Finance and Business Strategy.
Scroll to Top
Discover more from FourWeekMBA
Subscribe now to keep reading and get access to the full archive.
