AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: Canada's Grid: The Hidden Engine Behind AI Innovation on ThorstenMeyerAI.com

Age 18–24?Offer from Amazon

Prime made for students and young adults

  • Fast, free delivery for dorm and study essentials
  • Prime Video and Amazon Music included
  • Member-only deals
Try Prime for Young Adults Free trial for eligible 18–24 year olds
As an affiliate, we earn on qualifying purchases.

TL;DR

Canada’s hydro-rich provinces are restricting new power procurement for data centres, complicating the country’s role as a key energy supplier for AI innovation. This challenges previous assumptions about Canada’s abundant, cheap clean power and highlights infrastructure constraints affecting AI growth.

Canadian provinces, notably Quebec and British Columbia, are imposing restrictions on new power procurement for large data-centre projects, challenging the narrative that Canada offers abundant, low-cost clean energy for AI innovation. These restrictions are affecting Canada’s ability to serve as a major energy hub for AI development, especially as global demand for data-centre capacity surges.

Quebec has been limiting new power procurement for large data centres since 2024, with Hydro-Québec proposing a higher tariff of 13 ¢/kWh for data centre loads above 5 MW—a rate roughly double the current large-industrial tariff of 6.82 ¢/kWh. The province’s regulatory authority has yet to decide on this proposal, which is contested by a coalition of data-centre operators. Meanwhile, British Columbia has allocated only 400 MW over two years, capped at 145 MW per project, far below the needs of major campuses like the 200 MW Lübbenau site in Germany.

In late 2025, Canada’s total live data-centre power demand was approximately 1.4 GW, significantly lower than the 40.6 GW in the United States. The limited capacity and ongoing restrictions mean Canada cannot currently meet the energy demands of large-scale AI infrastructure, contrary to earlier assumptions based on its hydro resources. The constraints are not only regulatory but also infrastructural, with existing power grids and procurement policies limiting expansion.

At a glance
reportWhen: developing; restrictions enacted since…
The developmentRecent restrictions by Canadian provinces on power procurement for data centres are limiting Canada’s capacity to supply energy for AI development, contradicting earlier assumptions of abundant, cheap hydro power.
Energy Is the AI Policy — Reality Check
AI Dispatch · Reality Check · 18 September 2026

Energy is the AI policy: why Canada’s grid matters more than its labs — and why it isn’t free

Almost all the coverage leans on one assumption: Canada has abundant cheap clean power and Europe doesn’t. That assumption is about to be wrong, and the evidence is already public. Europe isn’t being offered a reservoir. It’s being offered a queue — already contested, already being repriced.

◆ The brochure — and it’s real
  • >78 GW installed hydro; ~60% of national generation
  • Lowest unit system costs: Quebec C$76/MWh, Manitoba C$91, BC C$100
  • Cold climate cuts cooling load; Ontario nuclear expanding
  • Ottawa: double capacity by 2050, non-emitting, plus an intertie programme
vs
✕ The reality, current and documented
  • Quebec has halted new large data-centre power procurement since 2024
  • BC: 400 MW over two years, capped at 145 MW per project
  • Alberta: 1,200 MW cap vs a >10 GW queue — a 1-in-8 hit rate
  • Canada live capacity ~1.4 GW vs the US 40.6 GW
⚠ The price of Canadian AI power is being set in a provincial regulatory proceeding — not in Strasbourg
6.82 ¢
/kWh · current large-industrial
→ ~2× →
13 ¢
/kWh · proposed >5 MW data-centre class
Hydro-Québec filed with the Régie de l’énergie on 19 Feb 2026. Eight months on, undecided — partly because a Coalition of Data Centres (six operators, 23 Quebec sites: QScale, CSquare, Equinix, eStruxture, Vantage, Cologix) is contesting it. A proposal, not a rate in force.
Four provinces, four different ways of saying “not so fast”
Québec
Rationing + repricing

Procurement restricted since 2024. Data centres are the largest new line item in the supply plan; consumption forecast to rise ~7× by 2035 (200 MW → >1,000 MW).

British Columbia
400 MW / 2 yrs

Capped at 145 MW per project from Feb 2026. For scale: Lübbenau’s first phase alone is 200 MW.

Ontario
You pay the marginal cost

Connection-asset payments, expansion deposits, locational marginal pricing. Shifts the cost — doesn’t remove the constraint. Nuclear expanding.

Alberta
Most welcoming

Federal MoU suspends Clean Electricity Regulations obligations; encourages made-in-Canada data centres. But 1,200 MW capped through 2028.

◆ The scale gap nobody sizes properly — live data-centre capacity vs European ambition
United States — live capacity, early 202640.6 GW
Canada — entire live fleet~1.4 GW
Mistral’s 2030 compute target~1 GW
Schwarz Lübbenau — first phase200 MW
One European champion’s 2030 target is comparable to Canada’s entire current data-centre fleet. Canada isn’t somewhere Europe offloads its compute demand — it’s somewhere incremental capacity can be added, supplementing rather than substituting.
◆ The tension energy forces on sovereignty

Energy economics push European AI compute out of Europe. Sovereignty rules push it back in. SecNumCloud requires EU-only storage; CADA’s assurance levels turn on data residency; the Digital Trade Agreement would prohibit “unjustified” localization. Three instruments, three directions. The workable answer is to tier the workloads: classified and DORA-bound work stays on EU soil regardless of price; pre-training runs and synthetic-data generation with no personal or classified data can sit where the electrons are cheap. Not all compute is sovereign compute — treating it as one undifferentiated resource is what makes the trade-off look impossible.

✓ What Europe should actually negotiate for — none of it in the current framing
1Interconnection priority, not price. The scarce good is a grid connection. Ask for queue position.
2Co-invest in interties — Alberta–BC, Alberta–Sask, Sask–Manitoba, Atlantic. Buys headroom better than any single campus.
3Nuclear & SMRs are the long game — hydro is largely allocated. EDF, Framatome, Siemens Energy, Rolls-Royce SMR make this a contribution, not a request.
4Keep critical minerals in the same instrument — grid buildout, storage, transformers and cabling run through the same chains.
5Arrive financing generation, not requesting megawatts. Projects bringing ownership, Indigenous participation, waste-heat reuse and grid investment clear. Others don’t.
The take

The sovereignty debate has been conducted as a legal argument — ownership caps, adequacy, assurance levels. All of it matters. But the binding constraint of the next five years is physical, measured in megawatts and queue positions. On that measure Canada is genuinely the best partner on offer: real hydro, a nuclear programme, cold climate, critical minerals, a government building sovereign compute. The alliance logic holds — at a smaller scale and higher price than the enthusiasm implies. Buy queue position, co-finance generation, put the sovereignty-bound workloads at home and the rest where the electrons are cheap, and tie it to interties and SMRs rather than one campus. Because Lübbenau’s lesson crosses the Atlantic: the scarce thing was never the model — it was the connection to the grid.

Sources: Hydro-Québec’s 19 Feb 2026 Régie de l’énergie filing (~13 ¢/kWh >5 MW class vs 6.82 ¢ industrial), its pendency and the Coalition of Data Centres challenge via The Concordian & ConstructConnect; Quebec’s post-2024 procurement restriction and 7×-by-2035 forecast; BC’s 400 MW/145 MW caps, Ontario’s marginal-cost regime, Alberta’s MoU and AESO 1,200 MW cap vs >10 GW queue, and Canada ~1.4 GW vs US 40.6 GW via BLG & NES Fircroft; provincial unit system costs via C.D. Howe; >78 GW hydro, double-capacity-by-2050 and interties via NES Fircroft & Data Center Frontier; crowding-out analysis via the Canadian Climate Institute; global 59→96 GW and Virginia’s 7-year waitlist via TD Economics; European load, hub congestion, E.ON 6 GW and Ember’s diversion warning via S&P Global; Mistral and Lübbenau as previously reported here. The Régie proceeding is unresolved; the tariff is proposed, not in force. Not investment advice.
thorstenmeyerai.com

Implications for AI and Global Data Centre Markets

This situation underscores that Canada’s reputation for cheap, clean hydro power is being challenged by recent policy restrictions, which could hinder its ability to attract AI investment. As global competition for data-centre capacity intensifies, Europe’s and the US’s more accessible energy markets may become more attractive. Canada’s constraints reveal that energy availability, not just resource abundance, is a critical factor in AI infrastructure development and global competitiveness.

Amazon

uninterruptible power supply for data centers

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Canadian Energy Policy and Data Centre Growth Limits

Canada’s hydroelectric capacity exceeds 78 GW across provinces like Quebec, BC, Ontario, Manitoba, and Newfoundland & Labrador, with hydroelectricity accounting for about 60% of national generation. Historically, this has supported low-cost power, with Quebec’s 2023 average system cost at C$76/MWh. Ottawa aims to double electricity capacity by 2050, supported by interprovincial links and nuclear expansion in Ontario. However, recent restrictions—such as Quebec’s halt on new large data-centre power procurement and BC’s limited allocation—highlight a disconnect between resource potential and actual supply.

These restrictions have been driven by concerns over crowding out other electrification efforts and rising retail prices. The industry is contesting some measures, indicating ongoing regulatory uncertainty. Meanwhile, Alberta remains relatively open but is capped at 1,200 MW of new large-load connections through 2028, with a queue of over 10 GW of proposed projects.

Amazon

industrial power meters for energy monitoring

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Unresolved Regulatory and Infrastructure Challenges

It remains unclear how quickly provinces will adjust their policies or expand capacity to meet growing data-centre demands. The regulatory decisions on Quebec’s tariff proposal are pending, and British Columbia’s limited allocations may not be sufficient for major projects. Additionally, infrastructure upgrades required to increase interconnection capacity are not yet defined or funded, leaving the full extent of Canada’s ability to support AI growth uncertain.

Amazon

high-capacity backup batteries for data centers

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Next Steps in Canadian Energy Policy and Data Centre Expansion

Regulatory decisions on Quebec’s proposed higher tariffs are expected in the coming months, which could either facilitate new data-centre development or further restrict growth. Provinces like Ontario and Alberta are likely to continue balancing industry demands with grid constraints, potentially exploring new interconnection projects. Meanwhile, industry stakeholders are advocating for policy reforms to unlock more capacity, but significant infrastructure investments and regulatory approvals are needed before large-scale expansion can occur.

Amazon

power grid management software

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Key Questions

How do recent restrictions affect Canada’s role in AI development?

Restrictions limit the availability of low-cost hydro power for large data centres, potentially making Canada less attractive for AI companies seeking scalable, affordable energy infrastructure.

Why is hydro power no longer considered unlimited for data centres?

Provinces like Quebec have implemented policies to restrict new power procurement for large data-centre projects, citing concerns over grid capacity, crowding out other electrification efforts, and rising retail prices.

What are the implications for global data centre markets?

As Canada’s capacity faces limitations, global data-centre operators might shift investments toward regions with more accessible energy markets, such as parts of the US and Europe, which could impact Canada’s competitiveness in AI infrastructure.

Will infrastructure upgrades solve Canada’s capacity issues?

While interconnection projects are planned, they require significant time and investment, and it is uncertain whether these upgrades will be completed swiftly enough to meet future demand.

What should Europe consider in its negotiations with Canada?

Europe should recognize that Canada’s energy constraints are real and ongoing, which may affect the reliability and cost of energy supplies for AI projects, rather than assuming an abundance of cheap, clean power.

Source: ThorstenMeyerAI.com

FALL

Fall Picks

As an affiliate, we earn on qualifying purchases.

You May Also Like

Security And Safety Layers For AI Agent MCP Server Environments

A new proxy-based security layer is being tested for MCP servers to improve permission controls, audit trails, and guardrails for AI agent integrations.

Kimi K3’s Top 3 Placement: A Game Changer In AI Development

Kimi K3 by Moonshot debuts at third place in VigilSAR’s AI benchmark, surpassing GPT and Gemini models, signaling a breakthrough in AI trustworthiness.

Claude Fable 5.1 Tops The AI Index — Analyzing The Cost Line Factors

Claude Fable 5.1 leads the AI Index with a score of 66, but costs about 20% more per task due to increased verbosity. Analysis of factors involved.

AI in Business: Why Diligence Isn’t Enough to Seal the Deal

Live AI experiments show that reading deeply, resisting manipulation, and disciplined prioritization outperform mere diligence — crucial insights for business AI deployment.