AI talent in Canada has become a federal economic priority. The latest Express Entry data point in the other direction.
Immigration, Refugees and Citizenship Canada opened a consultation on August 4 asking employers, industry groups and the public how category-based Express Entry selection should work in 2027. The consultation closes September 1.
The timing matters. Canada says it wants faster entry and clearer permanent-residence paths for skilled AI workers. Yet IRCC’s own consultation data show the number of STEM-eligible candidates who received an invitation to apply fell from 24,084 in 2023 to 5,942 in 2025.
That is a 75% drop in two years.
TL;DR
- IRCC is collecting input on 2027 Express Entry priorities until September 1, 2026.
- STEM-eligible candidates who received invitations fell from 24,084 in 2023 to 5,942 in 2025, according to IRCC.
- Canadian AI employers should submit specific occupation, vacancy, wage and project-delay data. General statements about needing talent will carry less weight.
What happened
The 2026 Express Entry consultation asks how Canada should balance three goals in 2027:
- Long-term labour shortages
- Attraction and retention of skilled international workers
- Francophone immigration outside Quebec
IRCC names Canada’s national AI strategy as one policy that could shape next year’s categories. It also points to an expanded Global Talent Stream for skilled AI workers and a planned accelerated route for some H-1B visa holders.
No dedicated AI category has been announced for 2027. The consultation is input, not a program decision.
That distinction matters. Employers should treat this as a short policy window, not a finished recruitment channel.
Why it matters in Canada
Canada’s national AI strategy aims to raise business AI adoption from 12% to 60% by 2034. Adoption at that scale needs people who can connect models to real operations.
The hiring need reaches beyond research scientists. A working AI program may need data engineers, security specialists, product managers, systems integrators, process owners and staff who can test model output inside a regulated workflow.
The Express Entry figures show a policy mismatch worth examining. STEM-eligible invitees represented 28.7% of all Express Entry invitations in 2023. That figure fell to 18.4% in 2024 and 5.8% in 2025.
These numbers do not tell us how many invitees were AI specialists. They also do not prove that Canada admitted fewer STEM workers overall. Candidates can qualify under several categories, and the table counts category eligibility rather than one isolated selection stream.
Still, the direction is hard to ignore when Ottawa is also calling AI talent a strategic asset.
Business impact
For employers, the problem is not simply the size of the labour pool. Timing and role fit matter.
A six-month delay in filling a senior data engineering role can postpone a customer-data project, an automation rollout or a security review. A general software developer cannot always replace someone with production experience in model evaluation, retrieval systems or agent controls.
Immigration policy also cannot carry the whole load. Canadian firms need stronger internal training, better entry-level roles and more disciplined workforce planning. Importing senior talent without building junior pathways creates another bottleneck later.
The better policy mix connects four pieces:
- Faster recruitment for scarce senior skills
- Permanent-residence paths that help firms retain those hires
- Paid training for current Canadian staff
- Clear early-career routes for graduates and career switchers
What leaders should do next
1. Submit evidence before September 1
Use the IRCC survey as an employer submission. Include job titles, National Occupational Classification codes, time-to-fill data, wage ranges and the business work delayed by each vacancy.
2. Separate scarce skills from convenient hiring
Name the capabilities your local hiring market cannot supply at the required level. Avoid treating every technical vacancy as an AI shortage.
3. Pair immigration with a training plan
For each senior international hire, identify one or two Canadian employees who can gain adjacent skills through project work and formal training.
4. Audit retention barriers
Compensation matters, but so do project quality, management, compute access and a credible route from pilot work to production. Recruitment will fail if skilled people spend their time making demos with no deployment authority.
5. Ask industry groups to publish their evidence
Aggregated vacancy and project-delay data would give Ottawa a stronger basis for category design. It would also let employers test their assumptions against the wider market.
Numbers worth knowing
- 24,084: STEM-eligible candidates who received an Express Entry invitation in 2023.
- 13,749: The comparable figure in 2024.
- 5,942: The comparable figure in 2025.
- 75%: The calculated decline from 2023 to 2025, rounded to the nearest whole number.
- September 1, 2026: The deadline for the current consultation.
All Express Entry figures come from IRCC’s consultation data. IRCC notes that a person can qualify for several categories, so category figures should not be added together as distinct individuals.
The skeptic’s view
The drop in STEM-eligible invitations does not prove that Canada has rejected an AI talent strategy. Express Entry priorities changed across the three years, and many technical candidates may qualify through Canadian Experience Class, French-language or other routes.
There is also a fair argument that employers use “talent shortage” too loosely when they really mean they cannot hire an experienced candidate at their preferred wage.
That is why a useful employer submission needs hard data. Vacancy counts without pay, seniority, location and time-to-fill evidence say very little.
Closing analysis
Canada has set an aggressive AI adoption target. Its workforce policy now needs the same level of measurement.
Employers have three weeks to show Ottawa where the real shortages sit, what those gaps cost and which skills Canada can build at home. If business groups submit vague support for “more AI talent,” the 2027 categories may be built on the same vague evidence.
