Canada’s Defence AI Procurement Gap: From Pilots to Production

Canada has world-class research, a clean-power heritage, and companies already shipping immersive training and decision tools to defence, aerospace, healthcare, and heavy industry. You can see the promise in digital twins of ships and aircraft, in simulators on shop floors and in classrooms, and in early AI copilots that shorten learning cycles.

Yet too many capabilities remain pilots and slide decks rather than tools in daily use. We need to think like builders and measure supply, timelines, and results. If we want to lead, we need a single pipeline that turns electricity into compute, compute into software, and software into deployed capability.

Procurement Friction as an Innovation Killzone

Peak AI will be gated by electrons as much as algorithms. Cloud remains useful, but many missions need lower latency, predictable cost, and tighter data control than pure cloud provides. At the same time, models are getting smaller and more efficient, which shifts intelligence to the edge, in headsets, on vehicles, and inside secured facilities.

The opportunity is to match this startup culture with a procurement and export machine that rewards speed and outcomes. Today, it does not. Selling to National Defence typically requires 12 to 15 steps before a file even reaches Public Services and Procurement Canada. These span the Statement of Work and cost breakdown, the Minor Project Synopsis, DR Prog authority review, chain-of-command approvals such as RCAF endorsements, ADM(Mat) signatures, a return to DR Prog and dispatch to the relevant base for the financial form, base-level initiation, completion of the SRCL and Form 9200, Controlled Goods verification when required, technical checks such as CTAT, Central Registry intake, and Central Allocation routing. Once at PSPC, pre-queue intake, queue placement and negotiation, final review, and activation in SAP Ariba commonly add three to six months.

Start with a procurement vehicle in April, and a contract can arrive in January, leaving roughly two months before the March 31 fiscal year end. For small and mid-sized firms, that cadence is an innovation killzone.

What Canada Must Do Now:

Owning Compute to Control Cost and Risk

Canadian SMEs should own on-prem GPU clusters because ownership controls cost, protects data, and improves resilience as models shrink. Accelerated write-offs, investment tax credits on hardware, and repayable contributions for racks, cooling, and networking would let teams train, fine-tune, and ship without being trapped by monthly cloud bills.

Modernizing Industrial and Technological Benefits

Industrial and Technological Benefits also need an AI-era upgrade. Count what moves the needle: production deployments, shared compute that strengthens ecosystems, useful datasets, and exported products. Create a national ITB matchmaking portal with explicit AI criteria so primes and innovators can find each other quickly. Broaden access to the 9x multiplier so more qualified companies benefit, and make it simple.

Governments should not sit on the sidelines while the private sector runs matching programs. They should help create an ITB bank and an ITB investment framework that channels capital to where it compounds. Let ITB finance more than product, including regional R&D centers and buildings, since Canada is living through a housing crisis and needs to expand its industrial footprint outside city cores. Reward adoption and exports more than paperwork, and give credit when a prime helps a Canadian SME place a working capability in a squadron, a ship, a hospital, or a college.

Buying Outcomes, Not Perfect Requirements

The most effective modern defence companies do not wait for perfect requirements. They sit with operators, invest their own capital, and ship finished capabilities that update continuously. Canada can mirror that posture by buying outcomes off the shelf, testing on representative ranges in days, and scaling in months once users confirm value.

Include the departmental approval phases before PSPC in any serious red-tape reduction plan. Enable validated SMEs, including those under Pathway to Commercialization, to move through an accelerated or unified track, and avoid repetitive cycles once a solution is already recognized as innovative and suitable for adoption. Extend program windows beyond three years because real contracting often consumes most of that time. Aligning with NATO’s defence spending goal should come with a commitment to move faster for SMEs. Business-to-government contracting represents a significant share of the economy, often estimated in the 15% to 20% range, so when procurement is slow, one in every five scaling firms risks stalling, and when it is fast, those same firms hire, export, and strengthen national security.

Turning Domestic Adoption into Exports

Exports should operate like a system. The Canadian Commercial Corporation, working with ISED and the Trade Commissioner Service, can form a joint team focused on AI and simulation, leveraging government-to-government pathways and existing MoUs with NATO and Five Eyes partners. Introduce Canadian SMEs through defence departments and labs abroad, bring mission users to Canadian demo ranges, sign small pilot task orders quickly, then convert pilots into multi-year programs. Domestic adoption should be the springboard to allied exports, not the finish line.

From Pilot Theatre to Production

The spirit here is practical and very Canadian. Build clean power, own critical compute, modernize ITB for adoption, buy what works, export with allies. Swap process for progress. Move from pilot theatre to production. Keep our values, privacy, and security, while competing at the speed of technology. The window is measured in quarters, not decades. If we act now, we will deter rather than react, grow rather than drift, and keep our best people building here.