OTTAWA — The billions of dollars flowing through the field of artificial intelligence are helping advance the field, not cement the advantages of today’s technology giants, those firms are arguing to Canadian regulators. 

But anti-monopoly advocates are calling for the Competition Bureau to take a closer look at the flurry of deal making in the sector and how the technology might itself be used to corner markets.

Here’s what you need to know about the responses to the agency’s recently closed consultation on AI.

Deal discussions: The AI applications businesses and consumers use are built atop a stack of technologies, including machine-learning models; the cloud infrastructure—”compute”—needed to train and run them; and the data and chips that feed both. 

Amazon, Google and Microsoft all sell access to several of these layers, with their cloud services currently bringing in the most AI business. Still, “there is intense competition at all levels,” said Amazon, which argued there’s no need for new rules specifically for the AI space. Google cited the “increasing rate of AI company formation and widespread dissemination of AI tools” as positive innovation signs, and touted contributions like the research paper demonstrating the transformer architecture that underpins large language models.

All three tech giants also have tied up with startups that make foundation models. Amazon has committed US$4 billion to Anthropic, which uses Amazon Web Services. (Google is also an investor.) Microsoft has pledged billions to OpenAI and Mistral, both of which use its Azure infrastructure. Such partnerships help keep the sector lively, Amazon and Microsoft said. 

Startups get the financial resources and “in some instances” compute capacity of the bigger firms, as well as some credibility from being associated with them, the latter said. (“That is why the names of large technology companies are often featured prominently in press releases and announcements of successful fundraising rounds,” per the filing.) 

Meanwhile, the bigger firms get “strategic optionality” and a chance to build on innovation at startup speed. Right now, those large technology companies are opting to partner rather than acquire, Microsoft said. 

By another name: Those partnerships may in fact be “acquisitions by stealth,” charged the Canadian Anti-Monopoly Project. Large firms’ funding can come with “a high degree of indirect control” over the startups they back, the think-tank said. It called for the bureau to examine whether the transactions are really just mergers that should be subject to competition scrutiny.

Software giant Salesforce also expressed concern about the deals, noting they could lock customers into cloud providers in order to use particular models, or make it harder for other tech firms to work with the partnered startups’ products. “Control over key AI components by incumbent cloud infrastructure providers could harm competition,” it said.

Regulators in the U.S., U.K. and EU are already investigating the Microsoft-OpenAI deal, while Amazon and Google’s backing of Anthropic has also drawn scrutiny. All three tech giants have also made agreements in which they’ve hired away the co-founders of high-profile AI startups and licensed their models. Policymakers have accused the larger firms of employing the novel deal type to avoid antitrust attention.  

Not listed: The bureau’s page of published submissions did not include contributions from notable Canadian AI names, but some submissions did cite them. Amazon, Google and Microsoft all credited Toronto-headquartered Cohere among the leading competitors in the foundation model space.

The Canadian startup uses Google’s cloud, and counts tech giants Cisco, Nvidia, Oracle and Salesforce among its investors; some of those firms are using its large language models in their products. But Cohere has positioned itself as the independent option in a world of tie-ups and vertically integrated players.  

What’s next: The bureau’s consultation was an information-gathering exercise, as it looks to head off anti-competitive behaviour in a fast-evolving market. The submissions will “inform the bureau’s future work in this area, and how we engage with these new technologies,” spokesperson Anna Maiorino said. It has not yet announced plans for action, although both government and regulator have signalled a greater focus on the acquisitive practices of large companies, particularly tech firms.

The bureau has also made AI the theme of its September summit, an event that tends to bring together the top lawyers and policy wonks in the competition field.