Why Is Meta Expanding Its Data Center Presence to Canada

Meta is expanding its data center presence in Canada, breaking ground on its first Alberta facility as part of a global build out that now spans 33 sites. The move answers a question investors keep asking: why is Meta expanding its data center presence so aggressively even as its stock slides and spending balloons into the hundreds of billions.

At a Glance

Meta's 33rd global data center, its first in Canada, will be built in Alberta
The facility is rated at 1 gigawatt and expected to cost about $9.18 billion
Construction will support 3,000 jobs, dropping to 300 once operational
Meta is spending $42.4 million on local infrastructure upgrades
The site will use closed loop liquid cooling with no operational water use

The Numbers Behind the Alberta Build

A gigawatt of capacity puts this facility in the same tier as Meta's largest existing sites, and the $9.18 billion price tag reflects just how capital intensive AI infrastructure has become. For comparison, that single project cost is a meaningful slice of the $125 billion to $145 billion Meta has guided for total data center and related spending in its current fiscal year. The closed loop, dry cooling design, which Meta says will avoid operational water use entirely aside from fire safety and domestic needs, signals the company is trying to preempt the water usage criticism that has dogged data center expansion elsewhere.

Why Is Meta Expanding Its Data Center Presence Now

The Alberta project lands at a moment when Meta's AI ambitions and its financial credibility are both under scrutiny. Llama 4 underwhelmed last year, prompting a reorganization into Meta Superintelligence Labs, which has since shipped Muse Spark in April and, this week, a follow up model called Muse Image. Building out raw compute capacity is the prerequisite for training and running whatever comes next from that lab, and Meta appears to be betting that more gigawatts now translates into faster iteration later.

Monetizing the Buildout: Meta's Hyperscaler Pivot

Perhaps the more consequential development is Bloomberg's report last week that Meta plans to sell excess computing capacity to outside customers. That would put Meta in direct competition with Amazon, Google, and Microsoft, along with neocloud specialists like CoreWeave. Zuckerberg had floated this idea on prior earnings calls, but a concrete plan changes the calculus for investors trying to model returns on Meta's capital expenditure.

Selling surplus compute would give Meta a new revenue line to offset a spending plan that has already pressured the stock, which is down 16% over the past 12 months and 8.6% year to date. It also gives Wall Street a clearer framework for valuing infrastructure that until now has been justified almost entirely by internal AI ambitions rather than external demand.

Can Compute Reselling Ease Investor Doubts

The open question is whether reselling capacity and shipping competitive models fast enough will convince markets that Meta's data center spending, now running well north of $100 billion annually, is generating returns rather than just liabilities. Alberta is one data point in a much larger bet, and the next few earnings cycles should show whether that bet is paying off.

Frequently Asked Questions

Why are data centers growing?

Data centers are growing because demand for AI training and inference requires far more computing power, electricity, and cooling infrastructure than traditional cloud workloads did.

What drives data center growth?

Growth is driven by companies racing to build AI models, the need for massive compute capacity to train and run those models, and rising demand for cloud services from businesses and consumers.

Why are so many data centers being built?

Tech companies including Meta, Amazon, Google, and Microsoft are each building multiple facilities to secure enough compute capacity for their own AI roadmaps and to have capacity available to sell or lease to other customers.