Meta's Canadian 1GW campus was sited and grid-secured years before public announcement, per grid-first real-estate strategy.
Meta's first Canadian data center will rise in Sturgeon County, Alberta, representing a US$9 billion (roughly C$13 billion) investment in a 1 GW AI campus. The project is expected to employ roughly 3,000 construction workers at peak and more than 300 permanent employees once operational. The company plans to invest approximately $42.3 million in local road and water infrastructure.
The announcement ranks among the largest AI infrastructure investments disclosed this year. But the most revealing detail lies in how Meta described its preparation. The company worked with Greenlight Limited Partnership, AltaLink, Capital Power, and the Alberta Electric System Operator "to plan for and meet our energy needs years in advance" of the data center coming online.
This approach offers a window into how hyperscalers now approach AI infrastructure. Rather than announcing campuses and then pursuing power, operators increasingly secure generation, transmission capacity, and regulatory alignment before making projects public. As campuses scale from hundreds of megawatts to gigawatts, locking in electricity years in advance has become a key advantage in site selection and delivery.
"The hyperscaler race is moving from site acquisition to power-path control," said Neil Osnato, founder of Persistence Analytics Group. "For a 1 GW AI campus, land and GPUs are not enough. The advantage belongs to operators that can prove generation, transmission, interconnection, regulatory approvals, and community durability years before the project comes online."
Grid interconnection and transmission upgrades are measured in years, and in some jurisdictions the queue itself forms a bottleneck before permits are filed. "That's why you're seeing developers negotiate power agreements before they've finalized site plans," said Balaji Tammabattula, CEO of BaRupOn. Gigawatt-scale campuses increasingly require developers to coordinate with utilities, transmission providers, regulators, and power suppliers well before projects become public, shifting competitive advantage from land acquisition to executable power strategy.
According to Tammabattula, such conversations are beginning much earlier than historically. "We've seen that engagement move from what used to be a six- to 12-month conversation right before construction, to something that now starts three to five years out, sometimes earlier for gigawatt-scale projects. Developers who show up early with a credible plan and real commitment get a seat at the table. Developers who show up late get put in line behind everyone else."
Meta devoted unusual prominence to explaining how the Alberta campus will be powered. The company plans to add enough clean energy to Alberta's grid to match 100% of the facility's annual electricity use and said it pays "the full costs of our data centers' energy use so other consumers aren't negatively impacted." Meta stated it is "fully funding new generation and grid infrastructure in Alberta to support our data center," with investments that "will improve reliability across the entire Alberta grid and benefit all consumers."
These statements directly address one of the most contentious questions facing AI infrastructure: whether rapid data center growth shifts infrastructure costs onto other utility customers.
Alberta's competitive electricity market and capacity to accommodate large industrial loads have increasingly drawn hyperscaler attention as alternatives to more capacity-constrained US markets. Beyond the campus itself, Meta's investment extends to road and water infrastructure. The company also plans "strategic network infrastructure investments" intended to help the region accommodate future large-scale developments.
The campus will use a closed-loop liquid cooling system with dry cooling, eliminating operational water use for cooling. Heat is rejected without evaporative water consumption under normal operations. Water will instead be used for construction, building operations, and fire protection systems.
The significance of Meta's announcement extends beyond its entry into Canada. It illustrates how the largest AI campuses are now being developed. While public announcement is the first visible milestone, the underlying work to secure generation, transmission, and regulatory support takes place years in advance. As AI facilities push toward gigawatt scale, competitive advantage is increasingly defined by the ability to align generation, transmission, permitting, and regulatory approvals well before a campus is revealed. For the largest projects, the race is no longer simply about deploying more GPUs—it's about delivering the power systems that make those GPUs viable.
"Power capacity cannot be built on the timelines this industry is moving at," Tammabattula said. "The companies willing to build or secure their own generation, and do the unglamorous work of transmission planning and regulatory coordination years ahead of need, are the ones who will be able to scale when everyone else is stuck waiting for interconnection."