Al Vigier: Canada's AI strategy shouldn't include secret Palantir bills

Instead, buy domestic product, and out in the open.

By: Al Vigier

Three weeks ago, the Canadian federal government released “AI for All” and named itself the “strategic anchor customer” for Canadian artificial intelligence. The pitch is that Ottawa will buy from sovereign Canadian businesses and drag a lagging country forward.

Only about 12 per cent of Canadian businesses use AI, and barely eight per cent of small firms do. Ottawa’s new strategy wants 60 per cent by 2034. Government buying is supposed to lead the way.

The inconvenient part is that the government is already a serious AI customer. It just buys American, and it buys quietly.

Consider Palantir, one of the bigger U.S. firms in the space.

The Department of National Defence signed a contract with its Canadian arm in March 2020, starting at $14.4 million. It was never disclosed. More than a dozen amendments later, the value had climbed to about $44.4 million by last October, with roughly $46.8-million actually spent. A separate $3.7 million Defence contract surfaced only after a Conservative MP pressed the government on its AI spending.

The Ontario Provincial Police have run Palantir’s Gotham platform since
2015. These are precisely the data-fusion and decision-support systems the new strategy says Canada must own. Ottawa is buying them. From a foreign vendor. Behind a wall of “not for public disclosure.”

So when a strategy promises to anchor sovereign AI, the fair question is not whether the government will buy. It already does. The question is whether it will buy in the open, from Canadian firms, under rules it is willing to show the public. Nothing in “AI for All” forces that. The document is loud about building a sovereign AI base and silent about the foreign systems already running inside Defence and policing.

Look at what the strategy actually funds: $500-million to take equity stakes in promising businesses, $700 million more for compute power, a Trusted AI Certification program, and a missions program starting in health. Notice the common thread. Not one of those is a purchase order.

A government reaches for equity and certification schemes precisely when buying is hard, because writing a clean contract to a small sovereign vendor is the one thing its procurement machinery is built to slow down. Grants and equity are what you offer when you cannot, or will not, simply buy the product. Even the one procurement promise arrives wrapped in process: the strategy routes faster AI buying through a new Office of Digital Transformation: an office, not a mandate to write cheques to Canadian vendors.

The equity idea should worry company founders who might find themselves approached by a government looking to invest. A state that takes a stake in your company takes a seat at your table and plants a doubt in the mind of every allied investor who might otherwise back you. Government-built national champions have a long record of becoming wards of the state rather than competitors in a market. Canada does not need to own its AI companies. It needs to buy from them, in daylight, the way it has quietly been buying from Palantir for years.

Then there is the choice to launch the missions program in health, the most privacy-bound, risk-averse, slowest-procuring corner of the state. If you wanted to prove that public AI buying can work, you would start where decisions are already high-consequence and where the law already demands an audit trail. There are the courts, public safety, and defence. Beginning in health is choosing the venue most likely to stall in pilots for three years, then call the delay a lesson learned.

The test of an anchor customer, as Ottawa wants to be, is binary. Did it buy, in the open, from Canadian firms? There are other things it can do. Make the source lists usable by companies under 50 people instead of treating a security questionnaire as a moat for incumbents. Set a hard floor with a fixed share of every department’s AI budget spent on Canadian-controlled products, not foreign platforms or Canadian-flagged consulting. Hold those Canadian systems to the same auditability the public never got to see in the Palantir files. Then publish the receipts every quarter.

I run Caseway, a Vancouver company that builds the kind of auditable, sovereign decision-support software this strategy claims to want, so weigh my motives accordingly. I also not that I am not asking Ottawa to fund me. I am asking it to do the harder, less photogenic thing it keeps avoiding. They should buy Canadian AI the way it already buys the foreign kind, only out loud. Until then, “AI’s first customer” is a slogan stapled to a country that has been someone else’s customer all along.

Al Vigier is the founder and CEO of Caseway, a Vancouver-based automation and AI company, and a veteran of the Canadian Army, where he was medically released after getting shot.

---

The Line is entirely reader and advertiser funded. No federal subsidies, no bailouts. If you value our work, please consider supporting us by subscribing or making a donation. Donations are not subscriptions and do not unlock paywalled content, but they help keep The Line independent

---

To contact The Line with a general inquiry or comment, please email info@readtheline.ca. For other ways to connect with us or to follow us on social media, please see our LinkTree.