AI on the Horizon: October 2, 2025

Canada’s latest AI news on the economy, society, and policy. In this issue, we examine AI computing infrastructure and expansion plans in Canada. Computing infrastructure is the backbone of AI adoption, innovation, and data sovereignty. As we’ve discussed previously, Canada has the talent and the leadership, but lacks the tools needed to compete in the current wave of AI innovation.  

The AI Compute Competition Heats Up 

In 2024, Canada responded to the growing compute race by announcing a $2 billion CAD investment over five years, beginning in 2024–25, specifically to expand compute for Canadian researchers and AI companies. More recently, PM Carney signalled his intention to develop a Canadian sovereign cloud, amongst other nation-building projects such as liquid natural gas pipelines, nuclear reactors, and the expansion of mines.  

AI semiconductor firms have seen a meteoric rise in value since the rise of generative AI. Intel and NVIDIA, which account for 70 and 90 per cent of their respective markets, recently announced a collaboration to develop data centre and PC products for enterprise, hyperscale and consumer markets. This joint combination of AI-specialized and central processing incumbents was announced lockstep with NVIDIA’s investments in OpenAI (to the tune of a cool $100 billion USD). Ultimately, investments like this are keeping pace with the rapid acceleration in AI infrastructure ambitions, including Meta’s more than $30 billion USD investment in the first half of 2025 alone and Oracle’s $300 billion USD deal for compute power, which is set to begin in 2027.  

Despite the promise of these investments, Canada’s public computing infrastructure currently ranks last among G7 nations, and almost 90 times less than the US. In the private market, while we host nearly 300 data centres, American companies own nearly a third of them, raising questions about data ownership within our borders.  

Canada’s Telecoms Enter the Race  

In the summer of 2025, Telus and OpenText partnered to announce the launch of the OpenText and TELUS Canadian Sovereign Cloud, providing enterprise-grade cloud computing and AI capabilities with complete Canadian data sovereignty. Meanwhile, Simon Fraser University partnered with Bell to connect the Cedar Supercomputing Centre with the future Bell AI Fabric Site at Thompson Rivers University (TRU). But this wasn’t Bell’s only collaboration of the season, as they also partnered with Cohere to provide AI tools for Bell’s government and enterprise customers. This move is combined with Bell’s May announcement that it intends to build six AI data centres in British Columbia that run on hydroelectric power, and Buzz HPC’s announcement of the acquisition of a 7.2 megawatt data centre in the GTA in partnership with Bell Canada and its Fabric ecosystem.  

Chip manufacturers, such as Ranovus, are planning expansion, notably a $100 million CAD development to help scale and secure Ontario’s position in AI and global innovation. And in the Province of Alberta, energy firms Pembina and Kineticor are working together to fuel data centres and AI-related projects, touted to attract more than $100 billion CAD worth of projects to Alberta in the next five years. While expansion plans and promises are high, some are concerned that Alberta won’t be able to match the level of scale required to draw large investors. 

On the Horizon: Can Canada Keep its Computing Ecosystem? 

Just last week, Evan Solomon, the Minister of AI and Digital Innovation, announced the AI Strategy Task Force, which will conduct a series of consultations in October 2025. The consultations will be presented in November, with the opportunity for the public to share their perspectives over the same period. Infrastructure is one of the eight broad AI-related themes detailed. Minister Solomon has promised to keep up with the computing powerhouses, but the magnitude of investment is not the only challenge. 

The trend of Canadian AI and chip manufacturing startups being acquired or moving operations south of the border continues. For instance, Canadian semiconductor player Tenstorrent has shifted its operations from Canadian to American bases, maintaining a smaller presence in Canada. And AI chip and hardware development teams, like Untether and CentML, have been acquired by AMD (the third largest firm in AI chip production) and NVIDIA, respectively. Unfortunately, these moves and acquisitions aren’t a new trend. Let’s not forget that AMD’s core AI division was Canadian many decades ago. 

To learn more about our research on Canada’s AI economy, please see our recent issue briefing and summary.