Meta will spend C$13 billion to build its first Canadian data center in Alberta, a marker of how far the AI infrastructure race has pushed hyperscaler capex beyond US borders. The commitment lands Meta in a province that has spent the past two years pitching itself to American cloud operators as a low-cost, cold-climate alternative to Texas, Virginia, and Arizona. It is the company's largest single infrastructure disclosure in Canada to date.
The Alberta project reflects the calculus every AI hyperscaler is running right now: where can you find enough power, enough land, and a fast enough permitting path to stand up training and inference capacity within the next 24 to 36 months. Alberta offers a deregulated power market, gas-heavy generation with growing renewables, and average winter temperatures that reduce cooling load for most of the year. That combination has already drawn interest from other US operators evaluating Canadian sites.
For Meta specifically, the C$13 billion figure sits alongside a US capex plan that has ballooned as the company builds out compute for its Llama model family and its ranking and recommendation systems. Meta guided to $70–72 billion in capex for 2025, most of it tied to AI infrastructure, and Mark Zuckerberg has repeatedly said the risk of underbuilding outweighs the risk of overbuilding. Alberta becomes one node in that global buildout rather than a standalone bet.
Key facts
- 01Meta will invest C$13 billion in a new data center in Alberta, Canada.
- 02It is Meta's first data center in Canada.
- 03The project extends the hyperscaler AI capex wave into a new jurisdiction with cheap power and cool climate advantages.
Canada has been aggressively courting AI infrastructure investment on the argument that it can host US-adjacent compute with lower power costs and a stable regulatory environment. Alberta in particular has pitched surplus generation capacity to hyperscalers, and the provincial government has streamlined approvals for large industrial loads. Meta's decision validates that pitch in a way earlier, smaller deals did not.
The site selection also fits a broader pattern of hyperscalers spreading new builds across more jurisdictions to reduce grid-interconnect delays. In several US markets, queue times for new large-load interconnects have stretched past four years, and utilities in Virginia and Texas have started rationing capacity for new data center customers. Building in Alberta sidesteps that queue entirely.
The cooling and power profile of a modern AI data center is meaningfully different from the cloud data centers hyperscalers built a decade ago. Racks are denser, per-rack power draw is higher, and liquid cooling is standard for training clusters. Alberta's climate reduces the energy penalty on heat rejection for a large portion of the year, which shows up directly in effective power usage effectiveness numbers.
Meta has not disclosed the site's target IT load in megawatts, its expected online date, or which workloads will run there. Those details typically come out in staged announcements as construction milestones are hit and utility filings become public. The C$13 billion topline includes construction, land, and the multi-phase compute fit-out over the life of the campus.
Local reception will matter. Large data center builds in North America have drawn scrutiny over water use, grid strain, and property-tax abatements. Alberta's grid mix and the specific cooling design Meta selects will determine how much of that scrutiny lands here; closed-loop and air-side economization designs have become common in cold-climate builds and dramatically reduce water draw.
The competitive frame is straightforward. Microsoft, Google, Amazon, and Oracle have all announced multi-billion-dollar data center projects across North America and Europe in the past 12 months, and each is racing to bring capacity online before the next generation of frontier model training runs. Meta's Alberta commitment plants a flag in a jurisdiction the others have circled but not yet closed on at this scale.
The story to watch is whether Alberta becomes a hub or a one-off. If Meta's build proceeds on schedule and the province delivers on power and permitting, expect Google and Microsoft to follow with their own Canadian projects within 18 months. If interconnection or local politics slow the buildout, the C$13 billion headline will read as the ceiling rather than the floor for hyperscaler investment in the region — and the AI compute buildout will keep flowing to the US jurisdictions that can still say yes.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
