OTTAWA — OVHcloud is expanding in Canada as the French data centre firm seeks to capitalize on growing demand for the processing power to run artificial intelligence applications. 

The cloud arms of tech giants like Amazon, Google and Microsoft have banked billions by selling computing capacity to developers that need to train large language models (LLMs) and other machine learning systems. OVHcloud is targeting companies that are building products on top of open-source LLMs. “We’re convinced this market exists,” said newly-installed CEO Benjamin Revcolevschi.

OVHcloud’s footprint in Canada is big, and growing. In March, it opened a new data centre in Cambridge, Ont., pledging to spend $145 million over eight years to build out the site. The firm already operates eight facilities in Beauharnois, Que. It also has a factory there to assemble servers—“a computer in the form of a pizza box,” Revcolevschi calls them—for its North American data centres. 

The company has 250 employees in Canada, according to Revcolevschi, and is always looking to expand data centre capacity as it fills up. OVHcloud also sells access to “local zones”—smaller clusters of servers in areas it doesn’t have facilities but where customers need to process their data to meet regulatory requirements or to ensure applications run fast. It’s considering setting up some in western Canada in 2026.

The company sells dedicated servers to some clients like IT management firms Welcome Networks in Langley, B.C. and S3 Technologies in Montreal. Some customers, like Toronto-based blockchain startup Figment, buy access to pooled capacity.

The company describes itself as Europe’s biggest pure cloud provider, but it’s significantly smaller than the so-called hyperscalers that dominate the market. OVHcloud reported €993 million ($1.49 billion) in revenue in its latest fiscal year ending in August. Amazon Web Services, by contrast, had US$27.5 billion in net sales in the third quarter alone.

Still, publicly-traded OVHcloud is growing rapidly, particularly in overseas markets. Founded in 1999 and headquartered in the northern French city of Roubaix, the firm now has 43 data centres and 16 local zones worldwide. 

ChatGPT’s ascent has boosted the stocks of cloud providers and chipmakers providing the infrastructure for AI. Many commercial model providers have received funding from tech giants, and are in turn using their cloud arms to train and run their technology; OpenAI with Microsoft, Anthropic with Amazon.  

Revcolevschi says he is not concerned about such deals. “It’s a way to attract investments, and generate growth,” he said—the startups get funding, and the cloud giants get revenue. “But this is not the route we’re taking.”

OVHcloud is instead focusing on inference, the stage when an already-trained model produces outputs like text or images. Its clients are typically using open-source products from developers like Meta or Mistral rather than commercial products from OpenAI and Anthropic. Customers want to update the model with their own data, then develop immediate applications, Revcolevschi said. “That’s where we see demand.” 

The firm’s AI inference clients include Canadian IT consulting giant CGI and its French peer Capgemini, machine-translation tool Reverso and transcription service Gladia. OVHcloud wins clients by being cheaper than the hyperscalers, by promising privacy, and because compute is its only offering, according to Revcolevschi. “We don’t compete with our customers,” he said.