OpenAI is actively exploring strategic partnerships to develop large-scale data center infrastructure in Canada, citing the nation’s abundant energy resources and available land as primary catalysts for the expansion. The announcement was made by George Osborne, OpenAI’s managing director and head of countries, during an investment summit hosted by Mark Carney in Toronto. This pivot toward the Canadian market signals a calculated shift in OpenAI’s international infrastructure strategy, particularly as the company navigates the complex regulatory and energy landscapes that have previously stalled its European ambitions.
The summit, which took place on September 14 and 15, served as a high-profile platform for global financiers. Organized by Mark Carney—the former Governor of the Bank of England and current UN Special Envoy on Climate Action and Finance—the event convened roughly 100 of the world’s largest investment firms. The overarching objective of the summit was to mobilize up to $1 trillion in private capital over the next five years to fuel the next generation of industrial and technological infrastructure.
The Strategic Pivot: Why Canada?
For OpenAI, the move into Canada is driven by the fundamental requirements of hyperscale AI compute: power and space. The “Stargate” model, which requires massive clusters of GPUs to train increasingly complex large language models, demands a level of energy reliability and grid capacity that has become a bottleneck in many developed economies.
George Osborne, who served as the United Kingdom’s Chancellor of the Exchequer from 2010 to 2016, joined OpenAI in December to lead the company’s international engagement arm. His appointment reflects OpenAI’s desire to bridge the gap between private sector technological needs and the public policy frameworks of foreign governments. At the Toronto summit, Osborne lauded the Canadian approach to AI, specifically highlighting what he described as “one of the clearer plans for AI adoption of a western government.” He characterized the Canadian federal strategy for medium and long-term technological development as “very impressive,” suggesting a level of policy alignment that is currently missing elsewhere.
A Comparative Analysis: The Stargate UK Stagnation
The pivot to North America stands in stark contrast to the company’s recent experiences in Europe. In April, OpenAI formally paused its “Stargate UK” project, which was initially envisioned as a massive compute cluster in the north east of England. The project, intended to house 8,000 Nvidia GPUs and eventually scale to 31,000, was intended to anchor the UK’s position in the global AI race.
However, the project faced a series of headwinds. Among the primary concerns were the prohibitive costs of British industrial electricity, which currently tracks at roughly four times the rate found in the United States or the Nordic regions. Furthermore, the UK’s energy grid is experiencing unprecedented strain. Data from the UK’s national infrastructure bodies indicate that grid connection requests surged from 41 gigawatts in November 2024 to 125 gigawatts by June 2025, creating a massive backlog that complicates the viability of power-intensive projects.
Beyond the energy math, the Stargate UK project faced reputational and procedural challenges. Reports from The Guardian suggested that the project lacked foundational due diligence, with claims that OpenAI representatives had never visited the proposed site for its flagship British facility. Additionally, the British government had previously publicized a £1.9 billion contract for the project, which it was later revealed had never been formally signed.
Regulatory hurdles further exacerbated these issues. The UK’s legislative framework regarding text and data mining remains in a state of flux, with proposed exceptions facing significant opposition from the domestic creative and publishing industries. Without a clear legal path to utilize copyrighted material for training purposes, the long-term investment required for a physical data center remains a high-risk proposition for companies like OpenAI.
The Nordic Success: A Blueprint for Efficiency
While the UK project languished, OpenAI successfully established a foothold in Norway. The Stargate Norway project, currently in operation, leverages the country’s extensive hydropower network. Situated in Narvik, the facility began with a 230-megawatt capacity and is on track to host 100,000 Nvidia GPUs by the end of this year. The success of the Norwegian site underscores the importance of low-cost, sustainable energy in selecting locations for AI infrastructure.
The Norwegian model provides a blueprint for what OpenAI is now seeking in Canada: a combination of renewable, cost-effective energy and a regulatory environment that supports, rather than hinders, the scaling of compute infrastructure.
The European Union’s Gigafactory Response
Recognizing the competitive disadvantage posed by fragmented energy costs and regulatory uncertainty, the European Union is attempting to accelerate its own infrastructure development. The EU has launched a “gigafactory” initiative, inviting private bids for up to seven sites across the continent to bolster its AI capacity. The call for bids, which closes on November 12, is expected to attract over €20 billion in private investment.
France has already signaled its intent to lead this effort, submitting a €10 billion bid for an AI gigafactory project dubbed “Aion.” This push highlights the growing realization among European leaders that AI infrastructure is not merely a corporate interest but a matter of economic sovereignty.
Implications for the AI Infrastructure Landscape
The competition for AI infrastructure is effectively becoming a competition for electricity and physical space. As energy consumption from data centers rises, governments are increasingly forced to prioritize which industries receive priority access to the grid.
For Canada, the interest from OpenAI represents a significant opportunity to transition from an energy exporter to a digital infrastructure hub. If the federal government can successfully facilitate the partnerships envisioned by Osborne, Canada could secure a central role in the global AI supply chain. However, the move also invites domestic scrutiny. Environmental groups and local municipalities will likely demand transparency regarding the water usage for cooling and the overall carbon footprint of these facilities, despite the reliance on hydroelectric power.
For OpenAI, the strategy appears to be one of diversification. By moving away from the bureaucratic and energy-cost challenges of the UK and focusing on resource-rich regions like Norway and potentially Canada, the company is insulating itself against the volatility of the global energy market.
Timeline of Key Events
- 2010–2016: George Osborne serves as UK Chancellor; initiates professional relationship with Mark Carney.
- December 2023: George Osborne joins OpenAI as Managing Director and Head of OpenAI for Countries.
- April 2025: OpenAI officially pauses the Stargate UK project citing high energy costs and regulatory ambiguity.
- June 2025: UK grid connection requests hit an all-time high of 125 gigawatts.
- September 14–15, 2026: Mark Carney hosts an investment summit in Toronto; OpenAI expresses interest in Canadian partnerships.
- November 12, 2026: Deadline for EU gigafactory bids.
Looking Ahead: The Politics of Compute
The interplay between OpenAI’s corporate needs and the policy agendas of Western governments is becoming increasingly complex. George Osborne’s transition from the UK government to OpenAI serves as a case study in the blurring lines between public policy and private AI development. His influence, combined with Mark Carney’s efforts to catalyze institutional investment, suggests that the future of AI development will be decided as much in the halls of government as in the laboratories of Silicon Valley.
As the demand for compute power continues to outpace current supply, the countries that can offer the most streamlined path to energy access and regulatory clarity will likely emerge as the primary hosts for the next generation of foundational models. Canada, with its vast hydroelectric potential and newfound focus on AI infrastructure, is positioning itself to be a primary beneficiary of this trend. However, as the failed UK project demonstrates, political enthusiasm alone is insufficient; long-term success will require a transparent, reliable, and cost-effective framework that balances the immense power needs of AI with the broader economic and environmental interests of the host nation.
