On July 8 local time, Meta dropped some huge news: they’re pouring a staggering $10 billion into Sturgeon County, Alberta, Canada, to build their very first data center in the country. This beast of a facility will boast a 1-gigawatt installed capacity—think about the power needed for roughly 750,000 homes—and is slated to take about two to three years to finish up.
Once it’s up and running, this place will officially become Meta’s biggest data center outside the US, marking their 33rd global data center campus. And it’s not just about tech; the construction phase is expected to crank out 3,000 building jobs, with another 300 long-term, full-time roles once operations kick off.
To keep the lights on, Meta says they’ll foot the bill for custom power generation facilities and plug them straight into Alberta’s grid, leaning heavily on natural gas to keep things humming. When finished, this center will be the engine room for AI model training and inference, pumping out the compute power needed for the Meta AI assistant, plus content recommendation engines and generative AI features across Facebook and Instagram. Why Alberta? Well, aside from keeping land and energy costs in check, the region’s chill energy regulations and massive natural gas reserves make it a sweet spot. Meta’s already teamed up with several Canadian energy players—like Greenlight Limited Partnership, Altalink, Capitol Power, and the Alberta Electric System Operator—making sure they “plan and meet energy needs years in advance.”
Honestly, this isn’t even Meta’s first $10-billion-class data center splash this year. Back in February 2026, they broke ground in Lebanon, Indiana, on the colossal “Project Domino” data center—another $10 billion-plus bet with a 1GW capacity, ranking as one of their biggest single AI infrastructure plays at the time. Then in March, Meta cranked up the dial on their El Paso, Texas data center, ballooning the initial $1.5 billion budget to a whopping $10 billion, aiming to get 1GW of power online by 2028.
And let’s not forget the Hyperion project in Louisiana and the Prometheus project in Ohio. The Hyperion campus alone is slated to hit a mind-boggling 5GW of compute capacity, with just the supporting power plants, transmission lines, and energy storage setups costing close to $27 billion—industry insiders already call it one of the globe’s biggest AI training clusters. At this point, Meta’s publicly committed AI infrastructure investments have stacked up to nearly $180 billion. Talk about going all in.
Behind this aggressive spending is a serious financial comeback for Meta. Peek at their Q1 2026 earnings: total revenue hit $56.31 billion, up 33.08% year-over-year, blowing past market expectations. Net profit attributable to shareholders soared to $26.77 billion—a massive 60.86% jump—pushing the net margin up to a record-breaking 47.54%. Thanks to AI-driven ad automation tools making ad targeting way more efficient, their ad business alone churned out $55.02 billion, making up over 90% of the total revenue pie.
At the same time, Meta’s capital expenditures are scaling up fast. In 2023, their full-year capex was $28.1 billion. By 2024, it leaped to $39.2 billion, and in 2025, it hit $72.2 billion. Now, for 2026, they’ve announced a massive hike in full-year capex to somewhere between $125 billion and $145 billion, throwing the bulk of that cash at AI compute clusters and large model R&D. That’s roughly double what they spent in 2025.
Meta has completely shaken off that rough patch from 2022. Back then, taking a beating from over-investing in the metaverse, slipping ad revenues, and the Fed’s rate hikes, their stock tanked to around $88 a share, wiping out over 70% of their market cap. Zuckerberg caught a lot of flak for betting the farm on the metaverse. But then, they pivoted hard: slashing over 20,000 jobs, trimming non-core businesses, and shifting the strategic spotlight from the metaverse straight to AI. Operating efficiency bounced back beautifully. Pair that with an ad business recovery and the AI hype wave, and their stock rode a three-year uphill climb. As of the latest trading day, shares have smashed through the $600 mark, pushing their total market cap near $1.53 trillion—a staggering 5x bounce from those dark historical lows.
Here’s the kicker: just a week before this Canada data center news dropped, Meta announced they’re opening up their idle GPU resources to enterprise clients as a compute leasing service, stepping right into the ring to brawl with cloud heavyweights like Amazon AWS and Microsoft Azure. By some estimates, about 35% of Meta’s current AI compute is just sitting there doing nothing; they’ve built up such a massive reserve that their internal operations can’t digest it all in the short term. This little revelation actually sent AI compute stocks in both the US and Asian markets tumbling hard.
The day that news hit, the market saw it as a fresh cash cow for Meta’s hundred-billion AI spending, pushing their stock up 8.8% in a single day. But on the flip side, the AI hardware supply chain got slammed with a collective sell-off. Folks are genuinely worried that this massive expansion of industry compute supply could totally flip the old supply-demand logic for AI chips on its head. That same day, the Philly Semiconductor Index took a noticeable dive, with storage and optical communication—the core AI hardware sectors—leading the market drop.
A recent research report from Everbright Securities spells out the underlying logic: Meta’s march into cloud computing is all about forging new monetization paths for their monstrous AI infrastructure investments. It’s a clear sign that cloud services are currently the fastest, most no-nonsense way to recoup those massive upfront costs and boost current returns. Essentially, this outward monetization muscle is now the cash flow lifeline letting Meta keep cranking up their AI bets. And make no mistake—Meta’s long-term master plan to keep stacking AI compute power hasn’t shifted an inch.
