AI InfrastructureEnergy StrategyJul 10, 2026, 7:25 AM· 4 min read· #5 of 5 in ai

Meta Commits $9.1B to Largest Non-US AI Data Center, Securing Independent Power in Canada

Meta will build a $9.1 billion AI data center in Alberta, powered by a dedicated 932-megawatt natural gas plant to avoid straining the local electrical grid.

By Factlen Editorial Team

Hyperscale Cloud Operators 40%Grid Operators & Regulators 30%Infrastructure Investors 20%Energy Researchers 10%
Hyperscale Cloud Operators
Argue that securing independent, 24/7 baseload power is the only viable path to support the exponential compute demands of frontier AI.
Grid Operators & Regulators
Prioritize protecting public electricity grids from massive AI loads, mandating that tech companies bring their own power generation.
Infrastructure Investors
View the physical build-out of AI data centers and power plants as a highly bankable, long-term asset class distinct from software.
Energy Researchers
Highlight the tension between the tech industry's climate goals and the reality that AI currently requires fossil-fuel baseload to scale.

Why this matters

The sheer scale of AI compute is now forcing tech giants to become independent power producers. By decoupling from the public grid, Meta is setting a new precedent for how the next generation of artificial intelligence will be physically fueled.

The central claim emerging from the artificial intelligence sector this week is that the physical constraints of public power grids can no longer support the exponential growth of frontier models. In a move that provides concrete evidence of this shift, Meta Platforms has committed $9.1 billion (13 billion Canadian dollars) to construct its largest non-U.S. AI data center in Sturgeon County, Alberta.[1]

The primary evidence for this decoupling strategy is the facility's energy architecture. Rather than drawing from Alberta's public electricity grid—which provincial authorities acknowledge cannot support multiple hyperscale AI loads—Meta is anchoring the project to a dedicated 932-megawatt natural gas-fired plant known as the Greenlight Electricity Center.[4]

Documentary evidence from Capital Power, one of North America's leading independent power producers, confirms a long-term Energy Supply Agreement to support the Meta facility. The power plant is being developed by a consortium that includes Capital Power, Pembina Pipeline Ltd., Morgan Stanley Infrastructure Partners, and Kineticor Asset Management, with Meta identified as the primary off-taker.

The Sturgeon County facility represents Meta's largest AI infrastructure investment outside the United States.
The Sturgeon County facility represents Meta's largest AI infrastructure investment outside the United States.

Broader empirical data supports the necessity of this off-grid approach. A June 2026 preprint published on arXiv by energy researchers projects that global data center electricity consumption will more than double from 415 terawatt-hours (TWh) in 2024 to 945 TWh by 2030. The researchers explicitly identify AI as the principal driver, noting that single-site power demands now rival those of major metropolitan regions.[2]

The financial evidence surrounding the Alberta project highlights a bifurcation in how AI is funded. While public markets continue to trade software and narrative valuations, private capital is increasingly required to finance the heavy, physical infrastructure underneath it. Analysts note that sovereign capital and private credit are circling these massive power and real asset deployments, viewing them as bankable infrastructure.[3]

Global data center power demand is projected to more than double by 2030, driven primarily by AI workloads.
Global data center power demand is projected to more than double by 2030, driven primarily by AI workloads.

A secondary claim driving Meta's investment is the strategic need to monetize excess compute. Financial filings and analyst reports indicate that Meta is launching a "Meta Compute" cloud service to sell surplus AI computing power to third parties. This moves the company into direct competition with AWS, Microsoft Azure, and Google Cloud, turning its heavy infrastructure spending into a distinct revenue line.

A secondary claim driving Meta's investment is the strategic need to monetize excess compute.

To address the environmental impact of such a massive facility, Meta has introduced claims regarding resource conservation. The company has committed to utilizing a closed-loop cooling system for the Sturgeon County data center. This engineering choice provides strong evidence of water conservation, ensuring that no continuous freshwater is drawn from surrounding local sources once the system is filled.

Furthermore, Meta is investing $42 million directly into regional infrastructure, specifically targeting improvements for local roads and municipal water systems. This localized investment serves as evidence of a negotiated approach with the Alberta government, which is actively prioritizing hyperscale projects that bring their own power and mitigate community resource strain.[4]

Meta will utilize closed-loop cooling to eliminate the continuous withdrawal of local freshwater resources.
Meta will utilize closed-loop cooling to eliminate the continuous withdrawal of local freshwater resources.

However, there is transparent uncertainty regarding how this natural gas-powered facility aligns with the tech industry's broader climate commitments. While Meta has previously partnered with firms like Silicon Ranch to secure 100% renewable solar energy for data centers in places like Georgia and South Carolina, the sheer scale and 24/7 baseload requirement of the Alberta AI facility necessitated a fossil-fuel solution.[1]

The evidence suggests that current renewable generation, even when paired with massive battery energy storage systems (BESS), struggles to provide the uninterrupted gigawatt-scale power required for training frontier AI models. Natural gas, while emitting carbon, provides the dispatchable reliability that hyperscalers currently demand when operating outside the capacity of established public grids.[2]

A second area of uncertainty lies in the project's execution timeline. The data center and its accompanying Greenlight Electricity Center are slated to begin operations in the second half of 2028, with some phases extending into 2030. Given the global supply chain constraints on specialized AI hardware, high-voltage transformers, and cooling systems, the ability of the consortium to meet this deadline remains an open question.

Hyperscalers are increasingly forced to secure independent power generation as public grids reach capacity.
Hyperscalers are increasingly forced to secure independent power generation as public grids reach capacity.

Geopolitically, the evidence points to a deliberate diversification strategy. By placing its largest AI asset outside the United States, Meta gains access to Canada's abundant natural gas reserves, favorable climate for ambient cooling, and a regulatory environment eager to establish itself as a strategic AI infrastructure hub.[1][4]

The Alberta investment establishes a clear evidentiary baseline for the future of AI infrastructure. The era of data centers acting as parasitic loads on public utilities is ending; the era of sovereign, self-powered AI infrastructure has definitively begun.[2]

How we got here

  1. 2023–2024

    The generative AI boom rapidly accelerates demand for compute, straining existing U.S. electrical grid capacity.

  2. August 2025

    Meta partners with Silicon Ranch to secure 100% renewable solar energy for its South Carolina data center.

  3. June 2026

    Energy researchers project that global data center power demand will more than double by 2030, driven by AI.

  4. July 2026

    Meta announces the $9.1 billion Alberta facility, pivoting to independent natural gas power to meet scale requirements.

  5. 2028–2030

    The Sturgeon County data center and the Greenlight Electricity Center are slated to begin operations.

Viewpoints in depth

Hyperscale Cloud Operators

Tech giants argue that securing independent, 24/7 baseload power is the only viable path to support frontier AI.

For companies like Meta, Microsoft, and Google, the race to artificial general intelligence is fundamentally constrained by physics and power availability. The evidence from the Alberta project demonstrates that hyperscalers can no longer rely on public utilities to provide the gigawatt-scale, uninterrupted power required for training massive models. By partnering directly with independent power producers like Capital Power, these companies are internalizing their energy supply chains. They argue this approach not only secures their operational capacity but also protects public grids from being overwhelmed by AI's unprecedented electrical demands.

Grid Operators & Regulators

Provincial and state governments are prioritizing projects that bring their own power to protect public infrastructure.

From the perspective of grid operators, the sudden influx of AI data centers represents a systemic risk to energy stability. The Alberta government's stance provides clear evidence of a shifting regulatory environment: jurisdictions are welcoming the capital investment of hyperscale data centers but refusing to subsidize their power needs with public grid capacity. By mandating or heavily incentivizing independent power generation, regulators aim to capture the economic benefits of the AI boom—such as Meta's $42 million regional infrastructure investment—without risking brownouts or forcing residential ratepayers to bear the cost of grid upgrades.

Energy Researchers

Researchers highlight the tension between the tech industry's climate goals and the reality of AI's fossil-fuel reliance.

While tech companies frequently publicize their net-zero commitments and investments in solar or wind, energy researchers point to the Alberta project as evidence of a more complex reality. The 24/7 baseload requirement of AI training clusters cannot currently be met by intermittent renewables without cost-prohibitive battery storage at a massive scale. Consequently, hyperscalers are quietly turning to natural gas as a necessary bridge fuel. Researchers warn that as global data center consumption approaches 1,000 TWh by 2030, this reliance on natural gas will significantly complicate the industry's ability to meet its long-term decarbonization targets, even with water-saving technologies like closed-loop cooling.

What we don't know

  • Whether the consortium can secure the necessary high-voltage transformers and specialized cooling equipment to meet the 2028 operational target amid global supply chain bottlenecks.
  • How Meta plans to reconcile the emissions from a 932-megawatt natural gas plant with its corporate net-zero climate commitments.
  • If the 'Meta Compute' cloud service will successfully capture market share from established hyperscalers like AWS and Azure.

Sources

Source coverage

4 outlets

4 viewpoints surfaced

Hyperscale Cloud Operators 40%Grid Operators & Regulators 30%Infrastructure Investors 20%Energy Researchers 10%
  1. [1]DataM IntelligenceInfrastructure Investors

    Alberta Emerges as a Strategic AI Infrastructure Hub Following Meta's $9.1B Investment

    Read on DataM Intelligence
  2. [2]arXivEnergy Researchers

    Energy and AI: Projections of Global Data Center Electricity Consumption

    Read on arXiv
  3. [3]GRO Money NewsInfrastructure Investors

    Public Markets Trade the AI Narrative, But Private Capital Finances the Infrastructure

    Read on GRO Money News
  4. [4]Government of AlbertaGrid Operators & Regulators

    Alberta Welcomes Meta's $13 Billion Hyperscale Data Center Investment

    Read on Government of Alberta
Stay informed

Every angle. Every day.

Get ai stories with full source coverage and perspective breakdowns delivered to your inbox.