Technology companies are fundamentally rethinking where to build massive AI data centers, moving them away from traditional coastal locations toward power generation sites to secure the enormous amounts of electricity these facilities demand. Governments and energy providers are now actively encouraging this shift, recognizing that the infrastructure needed to supply electricity to data centers cannot expand fast enough to meet demand in traditional tech hubs (Source 1, 2, 3).

Why Are Data Centers Moving Closer to Power Sources?

The challenge is straightforward: AI data centers consume staggering amounts of electricity. Some facilities use more than one gigawatt of power, equivalent to the electricity consumption of roughly 850,000 homes. The International Energy Agency reported that capital spending by the largest technology companies exceeded $400 billion in 2025 and is expected to rise another 75% in 2026. This explosive growth has created a bottleneck.

Transmission projects take years to permit and build, utilities face growing queues for large-load connections, and equipment shortages can delay new generating capacity. Rather than waiting for the grid to catch up, companies are increasingly prioritizing what they call "energy certainty" by securing their own dedicated power sources. This has led to a fundamental shift in how data center locations are chosen.

"It will be much more economical if you go closer to where the power itself is generating," said S Krishnan, Electronics and IT Secretary in India, speaking at Microsoft's launch of its fourth cloud region in Hyderabad.

S Krishnan, Electronics and IT Secretary, Government of India

India's government is actively pushing this strategy. Krishnan noted that building data centers near power sources is easier and more economical than locating them near coastal cable landing stations like Chennai, where companies traditionally built facilities to connect to international undersea cables.

How Are Governments and Companies Implementing This Strategy?

• Bring Your Own Power Requirements: Saskatchewan has implemented a new data center framework requiring new projects to supply their own electricity, preventing additional pressure on the provincial grid. Bell Canada's planned expansion to a 1.2-gigawatt AI infrastructure hub in Saskatchewan includes up to 900 megawatts of additional capacity under this "Bring Your Own Power" principle.
• Co-Located Power Facilities: Microsoft and Chevron have entered a 20-year agreement for a co-located natural gas power facility in West Texas expected to provide approximately 2.67 gigawatts of dedicated capacity to a Microsoft-operated data center.
• Dedicated Generation Partnerships: Meta partnered with Pembina Pipeline to build a natural gas-fired electricity-generating station to power its first Canadian data center. Until that project comes online in 2030, Capital Power will provide 250 megawatts of electricity to the site.

What Does This Mean for Canada and Alberta?

Meta's C$13 billion ($9 billion) data center investment in Alberta has positioned the Canadian province as an attractive destination for hyperscalers, companies that operate massive-scale cloud computing and data center infrastructure to support artificial intelligence. The province offers abundant, cheap natural gas to generate electricity, available land, and a cold climate that makes operating supercomputers more cost-efficient.

Capital Power, an Edmonton-based energy company, is in talks with several data center developers and expects multiple hyperscalers to build large-scale facilities in Alberta beyond Meta. "I do feel strongly that we've got the potential to be a multi-gigawatt market," said Avik Dey, CEO of Capital Power, noting that multiple hyperscalers have been evaluating Alberta for the last 18 months.

Alberta's approach differs from traditional models. Rather than requiring data centers to connect to the existing grid, the province is giving new proponents the option to build their own electricity sources to maximize their power capacity. This encourages companies to invest in new electricity supply rather than relying on shared infrastructure.

What Are the Broader Implications of This Shift?

This strategy represents a fundamental reimagining of energy infrastructure. For more than a century, the modern energy system has followed a familiar pattern: extract fuel, transport it, generate electricity, transmit that electricity, and finally consume it. Each step requires infrastructure, capital, and time. By locating data centers at or near power generation sites, companies can dramatically shorten this chain.

The model could extend far beyond data centers. Industrial manufacturing, fertilizer production, critical infrastructure, military installations, and remote communities all require reliable power. Sufficiently large developments could potentially contribute electricity to the broader grid, creating a more distributed and resilient energy system.

However, this shift is not without challenges. A report released by the Pembina Institute, a clean energy think tank, warned that allowing data centers to draw from the provincial grid before their own generating capacity is running will strain supply and could raise electricity costs for consumers. Additionally, an Angus Reid poll in July found that 68% of Canadians would oppose a large data center near their home.

In the United States, data centers are facing increasing pushback from communities and lawmakers over power, water, and pollution concerns. Alberta's measured approach of encouraging data center proponents to invest in new electricity supply rather than relying on the grid reflects an attempt to balance economic growth with environmental and community considerations.

As AI infrastructure continues to expand globally, the question of where and how to power these massive facilities will shape not only the technology industry but also energy policy and infrastructure development for decades to come.