Last updated: July 23, 2026
Is Your Business Leaving Up to $5 Million on the Table?
If your company is building, testing, or deploying AI — and you haven’t applied for Canada’s AI Compute Access Fund — there is a real possibility you are funding expensive cloud infrastructure entirely out of pocket when the federal government is offering to cover 50% to 67% of those exact costs.
That is not a rounding error. For a Canadian AI startup spending $500,000 per year on cloud compute, that represents up to $335,000 in non-repayable federal support. For a scale-up spending $3 million on GPU clusters and inference infrastructure, the number approaches seven figures.
The AI Compute Access Fund (ACAF) is one of the most targeted and generous funding programs the Canadian government has introduced for the technology sector in recent years — and it is frequently overlooked because it sits outside the traditional grant ecosystems that most business owners are familiar with. This guide explains exactly what it is, who qualifies, what you can recover, and how to position your company for the next intake.
Why AI Compute Costs Are the Bottleneck Canadian SMEs Ignore
Building an AI product is fundamentally different from building traditional software. The compute bill — the cost of renting GPU time on platforms like AWS, Azure, Google Cloud, or Canadian providers like Borealis AI infrastructure — is often the single largest line item in an early-stage AI company’s budget, frequently exceeding payroll in the most compute-intensive phases.
The situation becomes particularly acute for Canadian companies for a structural reason: the most powerful GPU clusters (NVIDIA H100s, A100s) are predominantly hosted on American hyperscaler infrastructure. Every dollar spent on those clusters flows out of Canada, eroding the domestic AI ecosystem over time.
The ACAF was designed specifically to address this bottleneck. Announced December 5, 2024, as part of Canada’s Sovereign AI Compute Strategy, the fund allocates up to $300 million to offset compute costs for eligible Canadian AI companies — with a deliberate incentive structure that rewards using Canadian-based cloud infrastructure.
The key insight: This is not a research grant. This is operational cost recovery for AI companies already building and scaling products. If you are paying cloud compute bills today, this fund was designed for you.
Funding Details, Eligibility, and the Two-Tier Structure
The ACAF uses a tiered coverage model based on where your compute infrastructure is located — a deliberate policy choice to strengthen Canada’s domestic cloud capacity while still supporting companies that need to use global hyperscalers.
- Canadian cloud compute providers: Up to 67% (2/3) of eligible costs covered
- Non-Canadian cloud compute (AWS, Azure, GCP, etc.): Up to 50% (1/2) of eligible costs covered, available until March 31, 2027
- Project size: Minimum $100,000 — Maximum $5,000,000 per project
- Total fund size: $300 million nationally
Funding can be structured as non-repayable contributions (effectively grants), conditionally repayable (repaid only if certain revenue milestones are hit), or repayable contributions depending on the applicant’s commercial stage and the nature of the project.
Who Qualifies?
The eligibility requirements are more accessible than many applicants assume. To qualify, your company must be:
- Incorporated in Canada as a for-profit entity
- Fewer than 500 full-time equivalent employees (standard SME threshold)
- Revenue-generating OR Series A funded or later — this is the key commercial traction requirement
- R&D team must be Canada-based — remote employees working from Canada count
- Actively developing AI products or services — not simply using AI tools in operations
What Costs Are Eligible?
The fund covers compute infrastructure costs directly related to AI development, including:
- Cloud GPU rental for model training, fine-tuning, and inference
- Storage and data pipeline costs tied to AI workloads
- Costs from Canadian sovereign cloud providers (e.g., Cohere, Vector Institute partnerships)
- Networking and data transfer costs associated with AI compute
General SaaS subscriptions, marketing tools, or operational software do not qualify — the expenditure must be directly tied to AI compute workloads. For official eligibility details, refer to the ISED Program Guide for the AI Compute Access Fund.
Step-by-Step Action Plan: How to Position for the Next Intake
The current intake for the AI Compute Access Fund is closed. However, ISED has signalled additional intakes as part of the broader $300M deployment. Here is how to prepare your company now so you can move quickly when the next window opens:
- Register for updates on the ISED portal. Go directly to the ISED AI Compute Access Fund page and sign up for program notifications. Intake windows can open with short notice.
- Document your compute expenditures now. Pull 12 months of cloud billing data from AWS, Azure, GCP, or your Canadian provider. Separate AI-specific compute costs (GPU instances, training jobs) from general hosting. This documentation becomes your application evidence.
- Assess your eligibility tier. Are you revenue-generating, or Series A+? If you are pre-revenue and pre-Series A, you may not qualify — but this is a good prompt to revisit your commercialization timeline and consider parallel funding options like NRC IRAP, which does not require revenue.
- Evaluate switching to Canadian compute providers. The 67% reimbursement rate for Canadian cloud is 17 percentage points higher than the 50% for foreign providers. Companies spending $1M+ on compute could recover an additional $170,000 simply by shifting workloads to Canadian infrastructure. Run the numbers — it may justify migration costs.
- Prepare your project description. ISED will want to understand the AI product or service being developed, the business model, and how compute spending directly enables product development. Draft a concise 1–2 page project overview now, framing the compute costs as essential R&D infrastructure.
- Stack this with other programs. The ACAF does not preclude claiming SR&ED tax credits on the same R&D activities. Work with a tax advisor to identify which expenditures can be claimed under both programs — the stacking potential is significant.
Common Mistakes Canadian AI Companies Make With This Fund
Based on patterns across federal innovation funding programs, here are the pitfalls that derail otherwise strong applications:
- Waiting until the last minute. Federal intake windows are competitive and can close within weeks. Companies that spend months “getting ready” often miss the window entirely. Prepare your documentation now, in advance of the next announcement.
- Conflating AI usage with AI development. Using ChatGPT to write marketing copy does not make you an AI company for ACAF purposes. The program targets companies whose core product is an AI system — not companies that use AI as an operational tool.
- Underestimating project scope. The minimum project size is $100,000 in eligible compute costs. Some early-stage startups assume they are too small. If you have been spending $8,000–$10,000/month on compute over the past year, you are likely above the threshold.
- Not separating AI compute from general hosting costs. A common application failure is submitting blended cloud bills without isolating AI workloads. Your cloud provider’s billing console can generate cost allocation reports by service — use them.
- Missing the Canadian-compute premium. Many companies default to international hyperscalers out of habit. The 17% cost-recovery premium for Canadian providers is a material financial incentive worth actively evaluating before your next infrastructure decision.
How Enno Consulting Can Help You Access the AI Compute Access Fund
Navigating federal innovation funding — from understanding eligibility to preparing a compliant application — is rarely straightforward. The ACAF sits at the intersection of technology, corporate finance, and federal policy, which is exactly where many founders hit a wall.
At Enno Consulting, we work with Canadian technology companies, particularly those led by Chinese-Canadian entrepreneurs in British Columbia, to identify and access government funding programs that match their stage and sector. Our work includes:
- Funding landscape assessment — mapping all applicable federal and provincial programs to your company’s profile (ACAF, NRC IRAP, SR&ED, SDTC, and others)
- Eligibility pre-screening — honest assessment of whether you qualify before you invest time in a full application
- Application preparation support — project descriptions, financial documentation, and review for federal submissions
- Program stacking strategy — identifying legal combinations of grants, credits, and loans that maximize your total recovery
The ACAF alone can recover up to $5 million in compute costs your company is already spending. That capital, reinvested in product development, hiring, or market expansion, represents a meaningful competitive advantage. The question is not whether to pursue it — it is whether you have the time and expertise to do it well.
Ready to find out if your AI company qualifies for the AI Compute Access Fund?
Contact Enno Consulting for a free eligibility consultation. We will review your company profile, compute spend, and commercialization stage — and give you an honest assessment of your funding potential across all applicable programs.
Explore more resources for Canadian business owners:
- Read our guide: NRC IRAP: Non-Repayable Funding for Technology Innovation
- Read our guide: Canada Small Business Financing Program (CSBFP) Complete Guide
Have questions about the AI Compute Access Fund or other Canadian SME funding programs? Leave a comment below or contact our team directly. We respond to every inquiry.

