OpenAI has begun discussions to secure data center capacity in Canada as part of its global rollout of Stargate, the ambitious infrastructure platform through which the company aims to exponentially scale AI computing worldwide. Confirmed by Chris Lehane, the company’s global public affairs director, this aligns with Canada’s digital sovereignty agenda and its desire to attract strategic investments that will boost capacity, talent, and supply chains.
The roadmap under consideration is not singular. Lehane describes three potential routes for the country: building its own “full-stack” infrastructure, contracting capacity from established operators, or forming a local partnership that integrates computing, software, and commitments to public programs. In all scenarios, the core idea is for OpenAI to act as a “real partner on the ground” and to serve as a catalyst for the ecosystem: encouraging more investment, GPU deployment, skilled jobs, and serving as a lever for initiatives in sovereign AI that Canada is already promoting.
The Models: What’s Already Announced in Norway and Germany
Lehane compares the potential Canadian deployment to two models that OpenAI has launched this year in Europe:
- Norway (Infrastructure Model). Under the Stargate Norway brand, OpenAI partners with Nscale and Aker to deploy approximately 230 MW of IT capacity and 100,000 GPUs by late 2026, powered by renewable energy. This exemplifies a “gigafactory of AI”: electricity, data centers, and local industrial supply as the backbone of a sovereign AI at a national scale.
- Germany (Cloud Sovereignty Model). In collaboration with Microsoft and SAP, the OpenAI for Germany initiative will provide OpenAI services to the public and regulated sectors via Delos Cloud, a sovereign SAP subsidiary, hosted on Microsoft Azure managed outside of Microsoft’s direct operational control. This is the “sovereign service” pathway: respecting jurisdiction, data residency, and control without building an entirely new physical infrastructure from scratch.
Canada could combine both approaches: physical capacity (either own or rented) and sovereign service guarantees, within public-private collaboration schemes and involving local actors (utilities, hyperscalers, colocation operators, regional cloud providers).
Why Canada Fits into the Stargate Map
- Key Markets. The Toronto, Vancouver, and Montreal corridors host most of the country’s data center capacity, benefiting from established networks, skilled talent, and access to renewable energy.
- Growth on the Horizon. Sector analysis indicates Canada is in an expansion cycle with a notable pipeline of capacity under development and pre-leased, supported by hyperscalers, AI-native companies, and a demand for sovereignty in regulated industries.
- Political Signal. Canada’s federal AI minister, Evan Soloman, named in September the digital sovereignty issue as “the most pressing political and democratic matter of our time,” while emphasizing the importance of international cooperation. This balance — sovereignty and alliances — is the natural space where OpenAI envisions playing.
- Local Ecosystem. Canada already hosts Cohere, a scale-up of foundational models competing at a similar level, which recently announced a collaboration with AMD to use its GPUs. Lehane suggests that OpenAI’s entry could draw capital and accelerate the “flywheel” of compute, models, and applications.
Negotiation Options OpenAI Is Considering in Canada (and What Canadians Are Asking For)
OpenAI is exploring a flexible menu with the federal government — as well as public and private stakeholders:
- Build: a “full-stack” campus similar to Norway’s (energy, land, grid manufacturing, campus scale of tens to hundreds of MW).
- Buy: capacity reservations (long-term contracts) with Canadian operators that already have sites under development.
- Cocodevelop: joint ventures or sovereign agreements where OpenAI commits to consumption in exchange for data controls, residency, and audit regimes.
Canada, on its part, seeks three guarantees:
- Sovereignty & Compliance: data residency, local operational control, and auditable custody chains.
- Energy & Network: joint planning with utilities for new loads (tens to hundreds of MW), renewable PPA contracts, and grid reinforcements.
- Local Spillovers: employment, upskilling, grants, and collaboration with universities and research centers.
Lessons from Norway and Germany for the Canadian Case
Norway demonstrates that physical sovereign AI is achievable if the triangle of land, renewable energy, and anchor clients (here, OpenAI) are aligned. Germany shows that sovereign cloud services (lawful and operational control managed through a national operator) reduce friction for public sector adoption. Canada might hybridize: establishing campuses in Quebec or Ontario (focused on renewables) coupled with sovereign agreements with local operators for government services.
The “Why Now”: Stargate Is Accelerating
“Stargate” is OpenAI’s label for its global deployment of compute, networking, and storage. Launched in January alongside partners like SoftBank, Oracle, and Abu Dhabi’s MGX fund, its goal is to mobilize massive investments over the next 4–5 years. In Europe, Stargate Norway marks the first milestone. In Asia, OpenAI has announced partnerships in Korea (with memory manufacturers and local cloud providers) and is exploring additional locations. Canada would be the next logical step in North America outside the U.S.: a friendly jurisdiction, with mature networks, an attractive electric mix, and an expanding data market.
Key Challenges: Energy, Timelines, and Integration with Local “Sovereign AI”
- Energy & Permits. Loads of 200–300 MW per campus require electric planning years in advance, grid reinforcements, and often PPA contracts to ensure renewable sourcing.
- Timelines. Building from scratch can take approximately 24–36 months (land, substation, civil work, MEP, IT). Purchasing capacity can shorten timelines if the national pipeline already has sites under construction.
- Competitive Coexistence. Canada wants to attract OpenAI while maintaining the strength of local champions (like Cohere) and preserving sovereignty. Designing governance structures that add value: sensitive data and workloads hosted in sovereign clouds; shared R&D; sandboxes for public services.
Impacts on Canada’s Data Center Landscape
The entry of OpenAI — either via campus development, capacity reservation, or sovereign service — would raise the country’s profile among North America’s “tier-one AI markets”. Toronto, Montreal, and Vancouver already dominate the existing stock and ongoing projects; a major contract could trigger new phases of growth, speed up network interconnections, and attract suppliers (optic, liquid cooling, power modules) into the local ecosystem.
Next Steps: Signals to Watch
- Deployment Model: construction, reservation, or sovereign partnership.
- Location: likely corridors include Quebec–Ontario, with attention to electricity costs and grid capacity.
- Size: initial “block” (100–150 MW), expandable to 200–300 MW if demand warrants.
- Data Regime: adopting a “Germany-like” model with a sovereign operator for the public sector.
- Ripple Effect: programs for training, R&D collaborations with universities, and support for local startups.
Frequently Asked Questions
Will OpenAI build its own campus in Canada or buy capacity?
The company considers both options (and a third, sovereign partnership). The choice will depend on timing, available energy, land, and regulatory fit that the Canadian government is willing to provide.
What’s the connection with “Stargate”?
“Stargate” is OpenAI’s global infrastructure program (computing, data, networks) that already includes projects in Norway (~230 MW, 100,000 GPUs by 2026) and alliances in Korea. Canada would represent another part of this distributed map.
How is data sovereignty guaranteed?
The German precedent uses Delos Cloud (SAP) on Azure, managed outside Microsoft’s direct control, with local residency and governance. An analogous agreement could be implemented in Canada with a local operator.
Does Canada have capacity for an initiative of this scale?
The markets in Toronto–Montreal–Vancouver hold most of the installed capacity and the national pipeline. Multiple reports suggest a significant growth in under construction projects and demand; a large anchor like OpenAI could accelerate investments in grid and energy.