Canadian Telecom BCE to Build Large Network of AI Data Centers

Canadian Telecom BCE to Build Large Network of AI Data Centers

(Bloomberg) — BCE, Canada’s largest telecom company by revenue, will set up a network of artificial intelligence data centers across the country, mirroring the aggressive AI infrastructure build-out happening in the US and abroad.

Bell Canada plans to invest hundreds of millions of dollars to build data centers in six Canadian cities, said chief executive officer Mirko Bibic in an interview. The first site, in Kamloops, British Columbia, is expected to come online in June, with US-based AI chip startup Groq as an anchor tenant. Other facilities are planned in the province, as well as in Manitoba and Quebec in the coming years, Bell said.

Once completed, the company expects the project – called Bell AI Fabric – to have 500 MW of capacity, a sizable amount to support AI applications, though far less than the multi-gigawatt facilities under development in the U.S., Middle East and Asia. Groq will be the exclusive inference partner for the new Canadian data center network, said Jonathan Ross, the startup’s founder and CEO.

A growing number of countries are working to expand their homegrown AI networks and capabilities, a push called sovereign AI. This month, Saudi Arabia – also a Groq partner – and the United Arab Emirates announced plans to work with US tech firms, including Nvidia and OpenAI, on local AI infrastructure. The moves are fueled by surging demand for computing resources to run models, as well as by a desire for AI systems customized to meet local needs and languages.

Related:Canada Invests $2B to Fuel AI Innovation

“We want to make sure that when you’re talking about cloud and AI that in our country we have the capabilities operated and managed by Canadians,” Bibic said. “We don’t want Canada to be turned off. Should there be any geopolitical tensions, we want to make sure we continue to have access to the technology that’s required to power the Canadian economy.”

Though Canadian researchers like Geoff Hinton and Yoshua Bengio pioneered many of the breakthroughs underpinning the current artificial intelligence boom, the country has lagged in the push to commercialize the technology, Bibic said.

“It’s a classic Canadian thing,” he said. “We always lead on the research and innovation side and then in terms of scaling up and commercializing, we are always a little behind.”

BCE isn’t the only Canadian firm trying to better position the country for the AI era. Telus Corporation, another large telecom provider, said this week that it’s investing more than C$70 billion ($50 billion) over the next five years to expand and enhance its network infrastructure and operations across Canada. Part of that is a bid to support data centers, or what Telus calls “sovereign AI factories.”

Related:Power Availability Now Drives Data Center Site Selection

Canada’s push toward more local AI data centers also taps into growing demand for inference, or the process of running an AI model after it’s been trained. While data centers focused on training AI can be placed almost anywhere with enough land, the location matters much more for inference as it’s all about speed.

“Sovereign has been a big part of our business,” said Groq’s Ross, whose company focuses on chips for inference. 

Groq now has more than 1.6 million developers who use its services, Ross said, with the option to select an AI model and a data center location to process their queries. Currently, Groq’s drop-down menu offers a choice of two US locations or the Middle East. Once the Kamloops site is operational, he said, Canada will be added to the list.

Comments

No comments yet. Why don’t you start the discussion?

Leave a Reply

Your email address will not be published. Required fields are marked *