We must build AI infrastructure for those who need it the most
“We have the brain. Now we need the mainframe.”
The “mainframe” that the innovation minister, François-Philippe Champagne, was referring to earlier this year to is AI Compute (AIC) infrastructure. It is the computational horsepower needed to build and test state-of-the-art AI algorithms. AI infrastructure is a critical input in advancing Canada’s national AI ambitions; for a country falling behind on productivity, no technology shows more promise.
Since launching the pan-Canadian AI strategy in 2017, Canada has maintained its position as a world leader in AI research and talent. The federal government’s recent $2.4-billion investment announcement to establish a domestic AI computing infrastructure policy is the first of many steps to boost Canada’s lagging productivity in a rapidly changing global innovation economy.
But how the minister disburses the money will determine the impact of this investment. And time is of the essence. Our recent research shows that right now, the United States has 90 times more raw computing power than we do. Adjusting for population and size of the economy, the U.S. still has 7.6 times more than Canada; Italy outdoes us by a factor of nine.
If we don’t act quickly, we stand to fail to attract and retain the talent desperately needed to jump-start productivity, leaving us further behind on the global competitive scale.
How can we ensure the federal government makes good on these investment dollars by building the new infrastructure needed to scale AI adoption while ensuring that all Canadians have the opportunity to benefit from this new technology?
■ Consult with the public. It must include a wide breadth of public and private sectors with a particular focus on groups historically left behind during previous waves of technological adoption.
Of course, this will require tremendous consultation, research and co-ordination efforts. The good news is, the playbook has already been written.
Taiwan’s digital affairs minister, Audrey Tang, has demonstrated stellar leadership in this field. Like Taiwan, Canada should build our AI infrastructure and industry policy with a lens of inclusive prosperity. This approach requires the voices of scale-ups, small-to-medium enterprises, and marginalized communities to be heard in the wake of AI’s growing impact on everyday life.
Invest in public-private partnerships. We envision the federal government convening a consortium of major Canadian capital providers who could build high-performance computing capacity. This capacity could be made available to various private-sector actors such as scale-ups and small and medium-sized businesses at globally market-competitive rates.
AI computing infrastructure support programs will also need to benefit those outside the AI ecosystem, incentivizing purpose-built technology and services endemic to Canada’s long-standing productivity and capital investment challenges.
■ Collaborate further with key trade partners for AI compute. By partnering with countries like the U.K., France, and others within frameworks like the European High-Performance Computing Joint Undertaking, Canada can secure more favourable purchasing power and access to AIC infrastructure procurement and cloud services for its AI ecosystem.
Over the next few decades, establishing international procurement agreements is critical to expanding domestic computing capacity. Shared purchasing power through these agreements will further drive down initial fixed capital investments for AIC, ensuring resilient and sovereign computing expansion strategies are cost-effective.
Too often, benefits from government funding and programs have ended up in the hands of the few, failing to realize the broader economic effects of government intervention.
We must ask, just as one of AI’s pioneers Yoshua Bengio has, how to direct AI toward implementations that benefit all Canadians. Bengio focuses on health care and safety, but we must have input and insight from a broader set of stakeholders across Canada, technologists and non-technologists alike.
At the same time, innovators, businesses and Canadians more broadly will only realize the benefits from this investment if the federal government deploys the capital fast.
We have the brain and, soon, the mainframe. Now, we need the heart.
Too often, benefits from government funding and programs have ended up in the hands of the few, failing to realize the broader economic effects of government intervention.
GRAHAM DOBBS AND JAKE HIRSCH-ALLEN