Evaluation of the Pan-Canadian Artificial Intelligence Strategy 2.0

Table of Contents

• Program context
• Evaluation context
• Findings
• Summary
• Recommendations
• Appendices
  • Appendix A: Acronyms
  • Appendix B: Other federal AI programming
  • Appendix C: Logic model
  • Appendix D: Methodology
  • Appendix E: Challenges and mitigations
  • Appendix F: Key terms

Program Context

Program Overview

Help build capacity and facilitate the application of AI across various sectors in Canada.

The Pan-Canadian Artificial Intelligence Strategy (PCAIS) was designed to harness the potential of AI to enhance the lives of Canadians and strengthen Canada's economy. PCAIS operates within a broader federal AI ecosystem that delivers programming to support AI research, compute, and adoption/commercialization, which collectively contribute to enhancing Canada's AI capability and performance (refer to Appendix B: Other Federal AI Programming).

A first phase of the strategy was launched in 2017, with an investment of $125 million from 2017-22, to strengthen Canada's talent base and global competitiveness in AI research.

A 2022 evaluation of CIFAR, which included funding under PCAIS 1.0, found that PCAIS 1.0:

• Enhanced national coordination and collaboration via the NAIIs;
• Attracted and retained AI researchers and supported training for new researchers;
• Promoted interdisciplinary and cross-sector collaboration, as well as the dissemination of AI research and responsible AI practices; and

The government launched phase two of the strategy in 2021 (PCAIS 2.0), investing $443.8 million from 2021–31 to build on the gains of the first phase and support the responsible adoption and commercialization of AI.

PCAIS 2.0 encompasses three key pillars:

Commercialization: fostering commercialization and use of AI to ensure that Canada's research strengths are translated into real-world applications that can help grow the Canadian economy and improve Canadians' daily lives.

Standards: advance the development and adoption of AI-related standards aligned with Canadian values and interests.

Talent and Research: renews the research- and talent-focused activities under the first phase of PCAIS.

Governance

Roles and Responsibilities

Innovation, Science and Economic Development Canada (ISED) is the lead federal department responsible for the Strategy's development, oversight, and coordination.

The Canadian Institute for Advanced Research (CIFAR) is responsible for administering the talent and research initiatives under the Strategy.

PCAIS is primarily delivered via Contribution Agreements with intermediary organizations, including the Global Innovation Cluster (GIC) funded cluster organizations, the National Artificial Intelligence Institutes (NAIIs), CIFAR, the Digital Research Alliance of Canada (DRAC), and the Standards Council of Canada (SCC). The funding provided to these organizations is governed by multiple Contribution Agreements (CAs) established between the third-party organizations (TPOs) and ISED for the various components of the Strategy.

Program Delivery

The Government of Canada is collaborating with partners from across the country to implement PCAIS 2.0.

National Artificial Intelligence Institutes (NAIIs): comprised of Amii in Edmonton, Mila in Montreal and the Vector Institute in Toronto, the NAIIs are helping Canadian businesses develop and use AI and turn AI research into real-world applications.

Digital Research Alliance of Canada (DRAC): helps to ensure that AI researchers in Canada have access to advanced research compute capacity.

Canadian Institute for Advanced Research (CIFAR): expands the Canada CIFAR AI Chairs program and continues research, training and collaboration programs.

Standards Council of Canada (SCC): furthers the development and adoption of standards related to AI.

Global Innovation Clusters (GICs): the five clusters which comprise the GICs (Digital Technology, Protein Industries, Advanced Manufacturing, Scale AI, Ocean) are promoting the adoption of made-in-Canada AI solutions in key industries.

Program Costs and Resources for PCAIS 2.0

DRAC was provided funding (2021-26) to enhance dedicated computing capacity for AI.

CIFAR was provided funding (2021-22 to 2030-31) to help increase the number of researchers appointed as Canada CIFAR AI Chairs and support collaboration and academic research and training at the NAIIs.

SCC was provided funding (2021-22 to 2025-26) to support the launch of two streams: the AI Intelligence Conformity Assessment Program and the AI Standardization Collaborative. Since the SCC is a Crown Corporation, the Standards pillar was not part of the evaluation.

GICs were provided funding (2021-22 to 2025-26) to support AI commercialization and accelerate economic growth in Canada's most promising industries.

NAIIs were provided funding (2021-22 to 2025-26) to grow capacity of Canadian businesses and other organizations to develop or use AI, turn AI research into real-world application, and encourage responsible development and use of AI.

Figure 1: Program Resources for PCAIS 2.0

Evaluation Context

An evaluation of the PCAIS 2.0 is required in accordance with the Financial Administration Act and the Treasury Board Secretariat Policy on Results.

The objective of this evaluation was to assess the relevance, performance and efficiency of PCAIS 2.0.

The scope of the evaluation included all PCAIS 2.0 activities, excluding the Standards Pillar. The evaluation covered the period from November 9, 2020, to March 31, 2025.

The evaluation was conducted by the Audit and Evaluation Branch at ISED. A results-based approach was used to examine the achievement of expected outcomes for the PCAIS 2.0, as identified in the logic model (Appendix C).

Evaluation Methodology

The following lines of evidence were used (details in Appendix D):

• Document and Literature Review
• Data Review
• Case Studies
• Online Surveys
• Interviews

Evaluation Questions

Relevance

1. To what extent is there a demonstrable need for PCAIS 2.0?

Performance

Immediate Outcomes

2. To what extent have collaboration and capacity building engagements to advance AI commercialization projects increased?
3. To what extent has investment in Canadian AI commercialization projects increased?
4. To what extent has support for advanced and interdisciplinary AI research, training and learning increased?
5. To what extent has collaboration between AI researchers and different organizations increased?

Intermediate Outcomes

6. To what extent do engagements increase the capacity of partnering organizations, Canadian small and medium enterprises (SMEs) and their employees to develop, adopt, use or commercialize AI technologies?
7. To what extent is knowledge related to AI increasingly shared and mobilized in Canada?
8. To what extent is Canada producing increasingly more highly qualified personnel in AI?

Ultimate Outcomes

9. To what extent is Canada's AI technology sector leading to the generation of economic and social benefits?
10. To what extent has Canada's international profile, capacity and AI research competitiveness been maintained?

Efficiency

11. To what extent is the PCAIS 2.0 funding model an efficient and effective approach to address the commercialization and talent and research pillars?

The evaluation produced 12 findings, supported by multiple lines of evidence, and led to 3 recommendations.

Findings

Finding 1: The second phase of Canada's Pan-Canadian Artificial Intelligence Strategy (PCAIS 2.0), and the funding it provides, were foundational for enhancing AI competitiveness and innovation in Canada. Through its commercialization-focused initiatives, PCAIS 2.0 has been helping to address key challenges, gaps, and barriers to the scaling, adoption, and commercialization of AI technologies in Canada.

Need for PCAIS Funding

There was broad, near-unanimous agreement among interviewees and literature review findings that public funding was important for supporting the full innovation pipeline in AI – from groundbreaking research, to talent retention and training, to commercialization.

Almost all Canada CIFAR AI (CCAI) Chairs (84%) surveyed viewed PCAIS 2.0 as being extremely important. Survey respondents said that Canada was an early leader in AI through its national strategy. Interviewees said that PCAIS 2.0 has helped institutions to compete globally in attracting and retaining top talent and is widely credited with enabling academic excellence, global competitiveness, and international recognition.

However, it was felt that maintaining global leadership requires renewed investment and more international engagement. It was also said that there's a pressing need to evolve the strategy as AI commercialization accelerates to ensure Canada's businesses can continue to compete internationally.

GIC Program: Most GIC ultimate recipient (UR) survey respondents felt that the funding and support provided by the clusters met the needs of their project (82%) and their organization (75%). As noted in the graph below, the top three needs being addressed via cluster projects were providing funding, reducing project risk, and collaboration opportunities:

Figure 2: GIC ultimate recipient needs addressed by PCAIS funding

Need for PCAIS 2.0 Funding

Interviewees, the literature review and surveys identified some key challenges that PCAIS 2.0 initiatives address:

Awareness Issues: Many businesses lack understanding of AI applications. Statistics Canada's 2021 Survey of Digital Technology and Internet Use found that 69% struggle to identify a business case for using AI, and 28% of non-adopters don't know what tools are available.^{Footnote 1}

Low AI Adoption: As of 2024, only 4.7% of Canadian firms had adopted AI, which was below many other countries, as Canada ranked 28th amongst OECD countries.^{Footnote 2} Adoption is concentrated in large firms.^{Footnote 3} Public funding can boost adoption, especially through collaborative projects.^{Footnote 4}

Commercialization Gap & IP Loss: Canada faces more challenges to commercialize AI compared to peers. Many firms sell their IP or move abroad.^{Footnote 5} U.S. companies dominate acquisitions and investments of Canadian AI IP.^{Footnote 6}

Commercialization Risks: AI projects are inherently risky. Funding helps de-risk and incentivize investment, especially for early-stage innovators. NAII survey respondents also noted unclear return on investments (25% of respondents) and high implementation costs (29% of respondents) as major challenges to AI commercialization. Similarly, the GIC survey found that the high costs of AI commercialization was a significant barrier (37% of respondents).

Venture Capital Deficiency: Canadian AI firms face financing challenges. Canada's venture capital environment is much smaller and more risk-averse than the U.S. Most venture capital investment is foreign (62% in 2022–23).^{Footnote 7} The GIC UR survey found that access to venture capital funding was the top commercialization barrier (53% of respondents). As such, public funding is needed to catalyze private investment.

Skills Gaps: Skilled talent is scarce and competitive internationally, especially given the number of large AI companies in the U.S.^{Footnote 8} 34% of the NAII survey respondents noted that lack of a skilled workforce was one of the biggest challenges. The GIC UR survey also found that difficulties attracting/retaining AI expertise was a top barrier (37% of respondents).

Limited Compute Capacity: Canada's reliance on foreign cloud platforms is problematic for data sovereignty and IP retention. Domestic AI compute infrastructure lags below demand, which discourages local development.^{Footnote 9} There is federal funding for new AI compute infrastructure, but it has "yet to hit the ground," so R&D is still inhibited by insufficient Graphics Processing Unit access.

Finding 2: Canada has built a strong AI foundation through early investments, including federal funding for initiatives such as PCAIS 2.0. However, gaps in coordination and cohesion remain, especially across federal initiatives and AI institutions.

PCAIS 2.0 within Canada's AI Ecosystem

Canada's AI Landscape

Canada has a well-established AI ecosystem, which as of 2019 consisted of 800+ startups and enterprises, 60 public research labs, 75 incubators and accelerators, and 60 investor groups, clustered in Toronto (Vector Institute), Montreal (Mila), Waterloo, Edmonton (Amii), and Vancouver.^{Footnote 10} In terms of investments, as of 2021, most Canadian AI businesses – 293 of 340 – had received equity-free investments, typically from government or accelerator programs.^{Footnote 11}

Federal Programs that Support Canada's AI Ecosystem

A wide range of federal programs support Canada's AI ecosystem, such as the Canadian Sovereign AI Compute Strategy, the Strategic Innovation Fund, NRC programs, Tri-Agencies, Innovative Solutions Canada, and Regional Development Agency initiatives. Together, these initiatives support AI research and talent, expand access to compute infrastructure, and enable business adoption and commercialization (refer to Appendix B: Other Federal AI Programming).

Role of PCAIS

Literature on local cluster development indicates government policies can reinforce the growth of emerging clusters and ecosystems.^{Footnote 12} Interviewees viewed PCAIS 2.0 as the foundational element of Canada's AI ecosystem, especially on the research and talent development side. The strategy has contributed to creating a critical mass of AI talent, interconnected researchers, and infrastructure through its support of the Canada CIFAR AI Chairs program and the NAIIs. This network is seen as essential for attracting researchers and students to Canada. This in turn creates an anchor for industry, as the presence of top-tier researchers and robust research hubs helps draw in industry, making it advantageous for companies to locate close to talent pools.

Federal and Departmental Roles

The federal approach also largely aligns with international jurisdictions, where it is common to establish AI strategies and to fund R&D and commercialization, compute infrastructure, skills and talent development, academic-industry collaboration, and development of standards/regulatory/ethics.^{Footnote 13} Interviewees and the literature review found that ISED promotes a unified national approach to AI policy. The literature review indicates that ISED's centralized role supports a nationally unified approach to AI governance. The focus tends to be on innovation, technological and economic development, and consumer protection.^{Footnote 14}

Coordination Mechanisms

CIFAR acts as a coordinator for the research and talent component, but does not manage the entire AI ecosystem, which retains a broad set of players. The NAIIs, under PCAIS 2.0, work with academia, industry, and government, forming a hub-and-spoke system. This allows for the flow of talent and ideas between sectors and regions. The NAIIs each have regional roles—but their mandates and programming also have standardized national elements to minimize regional inequity. There was some debate amongst interviewees about regional disparities, but most agreed the structure successfully balances the leveraging of regional strengths and national coherence.

Cohesion and Fragmentation

PCAIS 2.0 exists amongst other federal and provincial programs, as well as global initiatives. While there were very few similar provincial or territorial AI programs identified in the literature review, interviewees noted challenges in cohesion and coordination amongst the PCAIS 2.0 initiatives (e.g., GIC Program and NAIIs) as well as other federal initiatives (e.g., ISED's Canadian Sovereign AI Compute Strategy, National Research Council programming, and Tri-Agency programming). This increases the potential for duplication of effort and lack of coordination among organizations, thereby potentially reducing efficiency and slowing progress in the advancement of AI objectives.

Finding 3: Through PCAIS 2.0 funding, the NAIIs have offered impactful training to build both foundational and technical AI capacity, including programs for underrepresented groups. However, industry demand for this NAII programming exceeded the resources available to the NAIIs. GIC clusters have also supported training activities, particularly through 'on-the-job learning' in AI commercialization projects.

Training Programs and Services

NAIIs: Interviewees noted that NAIIs support the transition of talent to industry and help businesses understand AI challenges and opportunities. Program data shows the NAIIs delivered training to 37K to 47K participants per year, including through Massive Online Open Courses, workshops, bootcamps, and more. Over 50% of participants were from equity-deserving groups, exceeding the 30% target. Audiences included students, teachers, Highly Qualified Personnel, and industry.

Figure 3: NAII Training Participants

As shown below, the NAIIs exceeded their performance target of 50-70% of participants developing new skills or knowledge.

Figure 4: Percent of NAII Training Participants that Developed New Skills or Knowledge

Interviewees noted that the commercialization impact of NAIIs is largely focused on lower-to-mid Technology Readiness Level related activities. While this is largely due to the nature of NAII programming (providing access to AI researchers and research), some industry partners expressed a desire for more late-stage commercialization support. Interviewees also noted that there was a high level of demand, as there was insufficient capacity to meet the full extent of SME demand for NAII services due to the volume of requested training and services.

GIC Program: Interviewees explained that clusters have training initiatives (e.g., courses, micro-credentials, coding bootcamps, and applied AI talent programs) and targeted programs for underrepresented groups (e.g., bursaries, scholarships, and workshops). However, a lot of training is embedded in projects via collaboration and "learning by doing", as participants co-develop, pilot, and deploy AI.

Almost half of the UR survey respondents (46%) said their organization (mostly SMEs) provided increased AI training or upskilling to employees and underrepresented groups because of their project (to a great or very great extent), as shown below:

Figure 5: AI training or upskilling to employees and underrepresented groups

Among these UR respondents, about half (52%) found the training to be effective in increasing the knowledge and skills. The training/upskilling provided to project participants included guidance/support (73%), skills/knowledge transfer from partners (63%), training programs (53%), and webinars/workshops (33%).

NAII Industry Partner Case Study

NAIIs provided technical training and structured engagement. This strengthened organizations' foundational AI capabilities and knowledge and improved early-stage product development, via:

• Guidance: NAIIs provided guidance on scoping, model selection, and methods. They helped build essential data foundations, understanding of Machine Learning and fine-tuning of Large Language Models.
• Capacity Building: Addressing internal AI skill limitations by partnering with the NAII on AI for inspection robotics and anomaly detection.
• Learning Format: Sessions involved expert-led instruction—both high-level and technical deep-dives; smaller group work with technical advisors and support via office hours; and tailored guidance.

Finding 4: PCAIS 2.0 funding for the PAICE project aimed to expand AI compute access but faced a slow and complex rollout, with one-to-two-year delays across key activities, and experienced challenges with co-ordination and communication amongst stakeholders.

AI Research Infrastructure Capacity Development

Pan-Canadian AI Compute Environment (PAICE): Activities & Progress

Interviewees noted that access to compute infrastructure at AI institutes helps attract talent. DRAC's PAICE project aims to prioritize compute access for AI researchers and CIFAR AI Chairs, differing from DRAC's general compute access model, allowing for a significant increase in the specialized computing capacity and related services available to AI researchers. The PAICE funding was viewed by stakeholders as a temporary stop-gap measure. Increased demand—especially due to the emergence of large language models—led to the Government of Canada's $2 billion investment in AI Compute (Budget 2024).

The PAICE project involves several key milestones:

• Coordinating an RFI (Request for Information) for AI compute.
• Launching RFPs (Requests for Proposals) for 3 AI compute systems.
• Handling system selection, datacenter preparation, and resource allocation process development.
• Developing a National AI Platform with associated monitoring and metrics.

There have been some delays in PAICE activities:

• The RFI was delayed by 3 months (from Q3 to Q4 of 2022–23).
• Other activities saw 1 to 2-year delays, with originally planned completion by Q1 2023–24 pushed to as late as Q4 2024–25.

Finding 5: PCAIS 2.0 programs played an important role in strengthening Canada's AI ecosystem through industry engagement, capacity building, and collaborative partnerships. They have exceeded targets for engagement activities and facilitated connections between researchers, academia, and businesses. However, the importance of engagement on regulation and responsible AI development at the international level has increased significantly over the course of the PCAIS 2.0.

AI Capacity Building and Engagements

Capacity building & Support: NAIIs provide various programs to support AI development and adoption (e.g., Amii assigns students to help companies apply AI solutions).

The survey found that Amii's Machine Learning (ML) Explorations Program, Vector's cohort-based programs, and guidance/support/mentoring were some of the most impactful services in terms of improving their organization's knowledge and capacity to advance AI commercialization to a significant or very significant extent, as noted in the graphic below:

Figure 6: Improvement to Organizational Knowledge and Capacity to Advance AI Commercialization

The NAIIs reached 907 engagements in 2024-25, surpassing the 2026 target (450) and reached 713 partners (target: 800 by 2026). The partner types included startups, SMEs, large sponsors, and investors.

Networking

CIFAR has held a variety of networking events, including events to support collaboration amongst the NAIIs (e.g., AICan Symposium and the annual CCAI Chair meetings).

NAIIs facilitated networking amongst researchers, SMEs, and industry sponsors. The most common forms of networking were via events and conferences (65%) and personal introductions (45%, via the Institutes).

Figure 7: Industry Partner networking activities facilitated by the NAIIs

One-third of survey respondents made new connections with private sector businesses (39%) and business support organizations (35%).

Collaborations and Partnerships

The NAIIs and CIFAR facilitated collaborations and partnerships. Interviewees found that PCAIS 2.0 positively impacted collaboration between AI researchers and other organizations (industry, non-profits, public sector institutions), with program data showing they surpassed the target of 250 active partnerships by March 31, 2026, with 357 in 2023-24.

Figure 8: Active Partnerships

The NAIIs also had 14 to 25 collaboration projects per year (2022-23 and 2023-24) which examined the societal implications of AI, exceeding the target of 2 per year.

However, interviewees said there has been an "extraordinary internationalization of the discourse around AI," especially regarding responsible AI development. It was said that there is a need for more international partnerships and coordination to avoid duplication and share resources. CIFAR staff and others indicated that international engagement is now essential but was not fully anticipated or funded under PCAIS 2.0, meaning such efforts are "cross-subsidized" out of CIFAR's own budget, as well as those of participating institutions.

Survey data showed that CCAI Chairs' involvement with CIFAR was helpful in fostering collaborations with academic researchers (83% of respondents), while half also cited collaborations with private sector organizations. Most CCAI Chairs surveyed also had a high level of satisfaction with CIFAR's role in facilitating these research collaborations.

Figure 9: CCAI Chair collaboration and networking

GIC Program: Interviewees noted that collaboration generates network effects, knowledge spillover, and ecosystem density. The UR survey found half of the respondents engaged in 1-4 ecosystem development interactions and one-third engaged in 5 or more. Interviewees said project collaborations provide SMEs/startups with real world product validation and feedback. Their customers, often larger companies, also get early, lower-cost access to innovative solutions that can be customized to suit their operational needs. The program data review found there were 4.3 collaborators per project, while the survey found 5. Of these, almost half (48%) were small businesses.

Figure 10: PCAIS 2.0 GIC Ultimate Recipient Project Participants

In terms of limitations on collaboration, for the GIC clusters, connecting SMEs with large adopters was effective but constrained by the availability of funding for the clusters under PCAIS 2.0, and some SMEs also had difficulties securing investments after initial pilots.

Finding 6: Through PCAIS 2.0 funding for the CIFAR AI Chairs program, Canada's AI research ecosystem has been strengthened, attracting top global talent, retaining researchers, and advancing progress toward key targets.

AI Research Talent Attracted and Retained

The Paulson Institute ranked Canada in 2019 as having the 2nd highest concentration of top-tier AI researchers, with 10% of the world's most elite (top 0.5%), second only to the US. However, the ranking dropped to 6th in 2022, with only 3% of the world's most elite researchers.^{Footnote 15} While Canada continued to grow its AI researcher capacity, other countries also grew their AI researcher capacity at an even faster rate.

Interviewees and program data found that the CCAI Chairs Program has helped to further anchor and grow a strong research base in Canada by providing funding and support that has helped to successfully attract and retain world-class research talent, with the number of CCAI Chairs increasing to 129 in 2023-24 (close to the target of 135 by March 31, 2026), as shown in the graph below:

Figure 11: CCAI Chairs

Survey data supports the role of the CCAI Program in attracting and retaining AI Chairs, as it showed many respondents considered pursuing a research position abroad (61%), as noted in the graphic below:

Figure 12: Role of CCAI Program in Attracting and Retaining AI Chairs

Growth in Student Research Community

Interviewees highlighted the strong international reputation of the CCAI Chairs program and noted that the talent developed is key to Canada's AI success. They said that the CCAI Chairs train hundreds of graduate students yearly, nurturing future AI leaders.

This is demonstrated by program data which shows that the overall research community within the NAIIs continued to grow, with student researchers (graduates, undergraduates) and faculty growing year-over-year.

In particular, the target for the number of CCAI Chair-supervised graduates was exceeded significantly, with 254 in 2022-23 and 310 in 2023-24 (target of 160 by March 31, 2026).

Figure 13: CCAI Chair-supervised graduates

Further, for fiscal years 2022-23 and 2023-24, the percent of CCAI Chair-supervised graduates who identify as a member of one or more equity deserving groups ranged between 58% to 67%, as shown in the chart below:

Figure 14: CCAI Chair-supervised graduates who identify as a member of one or more equity deserving groups

Finding 7: Through PCAIS 2.0, stable and flexible funding has enabled increased interdisciplinary AI research and publications, which have been disseminated to various audiences (e.g., academia, private sector, government, and public).

Research Impact

NAIIs: Interviewees said PCAIS 2.0 is contributing to the enhancement of Canada's AI research reputation and capacity. The target of 150 active research teams at the NAIIs was achieved by 2021-22, with 175 teams. The number of teams grew further to 360 in 2023-24, as shown below:

Figure 15: Active Research Teams at the NAIIs

In 2024, Canada also ranked 2nd in the world on a per-capita basis for the number of AI research papers published.^{Footnote 16} For peer-reviewed publications, NAIIs exceeded the target of 550, with publications ranging from 1,164 to 1,402 per year. Interviewees highlighted key strengths of CIFAR CCAI Chair Funding:

• Promotes interdisciplinary research and networking.
• Five-year stable and flexible funding allows exploration of cutting-edge topics and quick pivots (e.g., AI safety).

CIFAR: 42% of CCAI Chair survey respondents said that without CIFAR's financial support, their research would not have occurred. The other half said scope, quality, or timing would have been affected. Further, the majority of CCAI Chairs (65%) reported that their involvement with CIFAR's networks/programs influenced the direction of their AI research. The most beneficial aspects of the program were the flexibility to recruit researchers/students and interactions with the AI institutes and CCAI Chairs, while other benefits were also noted below:

Figure 16: Beneficial Aspects of the CCAI Chair Program

Knowledge Dissemination

NAIIs: The NAIIs met their target in 2023-24 (12 activities), with 17 knowledge mobilization activities held. The NAIIs held a variety of knowledge mobilization activities, such as conferences, symposiums, science communication events, speaker series, research events, panels, and talks.

CIFAR: Lectures/presentations, peer reviewed publications, conferences, and seminars were the most common avenue for knowledge dissemination activities, and most had an international scope (84% to 100% of respondents). Academic (73%) and scientific communities (94%) were the primary audiences, although other groups were also reached (e.g., private sector, government, and public).

Figure 17: Scope of knowledge dissemination (ranked by the # of survey respondents)

CCAI Chairs were satisfied with the support provided by CIFAR in disseminating their research to various audiences (69% to 82% satisfied) and most perceived CIFAR's knowledge promotion and dissemination activities to be very effective (69%).

Finding 8: PCAIS 2.0 funding has contributed to strengthening Canada's AI ecosystem by enabling innovation and driving impacts in key sectors like healthcare, agriculture, and cleantech, while advancing AI technology, ethics, and safety and contributing to economic growth.

Economic and Social Benefits

The performance target of Canada ranking no less than fourth amongst the Group of Seven (G7) nations on the Tortoise Global AI Index rankings (measures three AI capacity pillars) was met, as it was 3rd amongst the G7 in 2022-23 and 2023-24.

However, the target of Canada ranking 2nd on the Stanford Global AI Vibrancy Index ranking (measures technical progress, economic influence, and societal impact) was not met, as it dropped to 3rd in 2022-23 and 2023-24.^{Footnote 17} Despite being an early leader in AI research, Canada has fallen in the rankings due to comparatively slower industry adoption and commercialization.

Interviewees and survey respondents view the CIFAR, NAII and GIC PCAIS 2.0 components as complementary drivers of AI innovation that deliver broad economic and social value across Canada, as they enable/support:

• Real-world impact through the application of AI in key sectors such as healthcare (e.g., personalized medicine, oncology, neuroscience, disease mapping, detection, and treatment), agriculture, environmental sustainability and aquaculture.
• Broad-level enhancements to AI models and algorithms, as well as AI ethics and safety.
• Economic development by enhancing job creation, productivity (e.g., automation, natural resources extraction, and manufacturing processes), talent retention, and the growth of AI SMEs.

Interviewees noted that the NAIIs were focused on capacity building, early-stage commercialization (e.g., proof of concept and product testing), and talent development, while GICs emphasized sector-specific adoption and commercialization.

These views were supported by a review of program data and client surveys, which found that the NAIIs, GICs, and CIFAR contributed to:

1. IP creation;
2. AI capacity development;
3. Generation of new or improved products, services, and processes;
4. Operational efficiency enhancements; and
5. Job creation and economic growth.

Finding 9: PCAIS 2.0 programs have contributed to strengthening Canada's AI commercialization landscape by helping industry partners build technical capacity, improve AI readiness, develop new or enhanced AI solutions, and generate new IP. Specifically, GIC clusters are driving strong applied, sector-focused AI commercialization, the NAIIs are focusing more on building foundational AI capabilities, and CIFAR is focusing on the creation and transfer of AI knowledge.

Intellectual Property (IP) Creation

NAIIs: A small share of NAII industry participants reported creation of IP (26%) and a few reported licensing IP (6%).

CIFAR: Survey respondents said patent applications were the main form of IP created, with 23% being granted patents, and 11% licensing their IP. The majority of this CIFAR AI Chair generated IP (68%) was successfully filed in international jurisdictions.

GICs: Slightly more than half of UR respondents (57%) said their organization created IP, and one-third reported sharing or licensing IP between partners.

GIC Program: Enhanced Commercialization Capacity

Interviewees noted that the GIC Program's PCAIS 2.0 funding provided to the clusters increases commercialization capacity, provides access to opportunities that would not otherwise be available, and helps businesses to better understand and address sector-specific AI adoption challenges and needs.

These views are supported by UR survey findings, which show that most project leads surveyed (79%) found federal AI funding and support programs helpful in reaching their AI commercialization goals. More specifically:

• 66% felt that the support and funding from the clusters improved their ability to develop, adopt, use, or commercialize AI technologies to a moderate/great extent

Figure 18: Development, adoption, use, and commercialization of AI technologies

• 57% reported that partnerships and ecosystem development interactions improved knowledge and capacity to advance AI commercialization to a moderate/great extent

Figure 19: Improved knowledge and capacity to advance AI commercialization as a result of partnerships and ecosystem development interactions

NAIIs: Enhanced Commercialization Capacity

Interviewees felt engagements were increasing commercialization capacity via training (e.g. one-on-one coaching for companies), applied projects, and talent placement. Program data supports this, as Amii, Mila, and Vector exceeded their targets (50-70%): in 2024-25, Amii reported 100% of industry organizations were better equipped to develop, adopt, use or commercialize AI, while Mila reported 76% and Vector reported 79%.

Figure 20: Improved ability to develop, adopt, use or commercialize AI

The NAII surveys also found there was a significant improvement in AI capabilities after engaging with the institutes (79% of respondents), with AI model development and data collection and processing improving the most.

Figure 21: Improvements to Industry Partners' AI Capabilities

The engagements also improved the majority of respondents' AI knowledge (74%), ability to take on new AI-related tasks and projects (53%), and confidence integrating AI into business processes (51%).

Figure 22: Improvements to Industry Partners AI knowledge and skills

New or Improved Products, Services or Processes

NAIIs: Program performance data from 2024-25 found that Amii, Mila, and Vector met their target range (30-50%) for the percentage of partners reporting new or improved products, services or processes, with Amii reporting 100%,* Mila reporting 67%, and Vector reporting 79%.

In contrast, the survey of NAIIs found more moderate commercialization impacts, which is in line with the earlier stage commercialization focus of NAII activities, as only 65% of respondents felt their involvement with the NAIIs had a significant impact on the development or improvement of AI solutions.

• 53% of respondents (average of 3 new solutions) developed at least one new AI solution following their engagements.
• Fewer reported improving AI solutions (28% of respondents).

While NAIIs supported early-stage startups and SME adoption, some interviewees noted a gap in support for mid-sized and scaling companies.

Figure 23: New or improved AI solutions developed by NAII partners

GIC Program: Interviewees said that the GIC UR projects are driving real-world commercial outcomes, especially when SMEs are matched with large industry partners or "first customers." Interviewees cited impacts in various sectors, including healthcare, agriculture, and manufacturing. Most UR survey respondents (87%) reported having developed new or improved AI products, processes or services for their projects, with an average of 3 solutions per organization. However, interviewees noted it was still early in project implementation and that the overall funding envelope was relatively small for PCAIS 2.0-related GIC projects.

CIFAR: Of those who collaborated with the private sector (48%), two-thirds surveyed indicated it led to new or improved AI-related products, processes or services, with an average of 3.25 products, processes or services.

There were various applications of the research, such as university curriculum and indirect and direct technology/knowledge transfer, as shown below:

Figure 24: Applications of CCAI Chair Research

There was a large variety of potential beneficiaries identified as well, with private sector (83%), public and semi-public institutions (63%), and federal government (61%) being the most frequently cited.

Finding 10: PCAIS 2.0 programs have contributed to improved company performance and growth. Businesses reported enhanced productivity, reduced costs, and faster decision-making due to AI-driven enhancements. Engagements supported job creation, particularly of Highly Qualified Personnel, and helped organizations establish, grow and compete.

Operational Efficiencies

NAIIs: Program data for 2024-25 found that Amii (86%), Mila (62%), and Vector (75%) met their target (30-50%) for the percentage of industry partners reporting improved outcomes (e.g., productivity gains, cost savings) because of the engagements.

The AI driven improvements to service delivery or production most frequently cited by survey respondents were data-driven decision-making (40%), operational automation (40%), and reduced manual workloads for employees (44%).

Around half (58%) reported moderate or significant productivity enhancements. In terms of the specific productivity enhancements, process automation, reduced manual workloads, and faster decision-making were the most frequently cited areas.

Figure 25: Industry Partner Operational Productivity Enhancements

GIC Program: Most UR respondents (83%) reported increased efficiencies as a result of their cluster project, as well as decreased costs (37%) and increased revenue (50%).

Job Creation and Growth

NAIIs: Most respondents (86%) felt that their engagement with the NAIIs had a positive impact on their overall growth and competitiveness.

About one-quarter of respondents (25%) reported hiring staff due to NAII programming. Half of the respondents said they hired full-time employees and one-third hired part-time employees, with an average of 3.2 Highly Qualified Personnel (HQP) hired per organization.

NAIIs also supported job creation through their role as talent hubs. They facilitated work-integrated learning, internships, career fairs, bursaries, scholarships, career counseling, and job search and acquisition.

19% of CCAI Chair survey respondents indicated that spin-off companies were created as a result of their AI research, with a total of 18 spinoff companies cited.

GICs: As noted below, two-fifths of UR respondents would not have proceeded with their project without GIC PCAIS 2.0 funding. The remainder would have proceeded with a reduced scope, delays (average 1.7 years), or fewer personnel.

Figure 26: Influence of GIC Funding on Ultimate Recipient Projects

Project Activity Reports estimated there would be 562 jobs created and 1,006 maintained as a result of the PCAIS 2.0 funded UR projects by 2025-26.

Figure 27: Jobs created and maintained

The GIC UR survey showed progress has been slow (likely due to delays), as respondents reported an increase of 3.6 FTEs per project (extrapolated to a total of 162 jobs created for the 45 projects), most of whom were HQPs (89%).

Finding 11: Overall, PCAIS 2.0 contributions to intermediary organizations are being managed in an efficient manner. They met or exceeded match requirements, costs were similar to comparable organizations, and overhead largely remained within expected limits. However, insufficient centralized coordination was seen as a challenge by stakeholders.

Funding Leveraged

Interviewees said the NAIIs, DRAC, GIC clusters, and CIFAR successfully attracted federal, provincial and private sector revenue sources:

• Around 50% of NAII revenue came from non-federal sources (provinces, industry partners, and service fees).

Figure 28: NAIIs – Non-federal revenues (%)

• 5 of 5 GIC clusters met or surpassed the 1:1 industry match requirement. Scale AI and Ocean were only required to match the GIC funds allocated toward non-community stream AI commercialization projects.

Figure 29: GICs – Non-federal funding leveraged (ratios)

• 30% of CIFAR funding came from non-federal sources (around 60% when PCAIS 2.0 contributions are excluded).
• DRAC's PAICE project was not advanced enough for significant matching funds to have been generated.

Operational Cost Comparison

PCAIS organizations' funding profiles are largely comparable to other ISED-funded third-party organizations (TPOs).

• PCAIS 2.0 organizations spend a higher share on wages (61% vs. 41%), largely due to the NAIIs and CIFAR being research-focused organizations.
• Operations and Maintenance (O&M) expenditures are slightly higher for PCAIS 2.0 organizations (18% vs. 14%), largely due to initial setup/overhead costs for GIC PCAIS 2.0 streams.*
• PCAIS 2.0 organizations have marginally higher private sector funding, likely due to relatively higher match funding requirements.

Figure 30: Operational Cost Comparison

Note: Organizations are not fully comparable as there are some differences in their characteristics, such as the organizational maturity or types of activities undertaken.

Funding Model and Centralization

PCAIS 2.0 model viewed by interviewees as effective, especially compared to international models/approaches to funding AI activities.

Interviewees said that having multiple smaller organizations deliver the program allows for different areas of specialization and ensures a more equal distribution of projects. However, this creates some duplication of effort and requires more money for management/oversight.

CIFAR's mandate is, in the case of PCAIS 2.0, convening and coordinating across the research and talent part of the ecosystem. However, interviewees noted that CIFAR doesn't have a mandate to coordinate the entire AI ecosystem—some confusion exists among stakeholders.

Challenges and Areas for Improvement:

GICs & NAIIs: Need expressed by interviewees for better coordination among federal and provincial funders, longer federal funding horizons, and more federal spending flexibility.

Overall: Lack of centralized coordination seen as a major gap by interviewees; coordination among PCAIS 2.0 programs is mostly informal. Interviewees called for ISED to play a stronger role in coordination, reducing fragmentation, and breaking silos between sectors/organizations.

Operations and Maintenance (O&M)

Program data review found that O&M costs were largely within the required limits. For some third-party organizations, there was higher O&M in early project stages, but this ratio is expected to decline over time as spending ramps up (e.g., GIC projects).

Figure 31: GIC Clusters – O&M Expenditures to Date

Figure 32: NAIIs – O&M Expenditures to Date

Finding 12: PCAIS 2.0 delivery partners made meaningful progress on EDI through strategies and initiatives. Contribution Agreements are generally well-managed, though staff turnover and some reporting gaps presented minor challenges. Due to delays in the rollout of some initiatives, spending was less than anticipated for some organizations.

Equity, Diversity, and Inclusion (EDI) Initiatives:

• DRAC, the GIC clusters, CIFAR, and the NAIIs have taken measures to incorporate EDI into their organizations.
  • DRAC has an EDI framework, outreach programs, and a proposed equitable resource allocation model for its PAICE systems, benefiting underrepresented researchers.
  • The GIC clusters engage in EDI initiatives, including scholarships, training, and participation in the 50-30 Challenge.
  • CIFAR implemented an EDI strategy, including balanced peer review committees, support for parental leave, and scholarships for Black and Indigenous students.
  • The NAIIs developed EDI strategies by 2022, formed EDI committees, and joined the 50-30 Challenge. Each NAII also runs various programs that are targeted to support underrepresented groups (e.g., AI4Good Labs, bursaries, mentorships).

Management of Contribution Agreements:

• Interviewees explained that the Contribution Agreements (CA) under PCAIS 2.0 had processes in place to manage the agreements. Key practices identified include regular meetings, guidance documents, risk assessments, and claim processes.
• The document review found that the reporting requirements specified in the CAs appear to have been largely met. Some specific gaps in reporting (e.g., KPIs and financials) were noted, but this information was largely available from other sources.
• One challenge noted by interviewees in managing CA requirements was high turnover of ISED operational staff.

Program Expenditures:

• The GIC clusters were slow with their initial spending; spending was mostly on Operations and Maintenance in the first year, which is consistent with other funding streams for the clusters.
• DRAC underspent in 2022–23 and 2023–24 due to project delays.
• CIFAR had multiple amendments to its funding timelines, with year-to-year deferrals as well.
• NAII spending is aligned with the CA spending profiles.
• A challenge identified by some GIC interviewees included utilizing funds allocated to projects within the approved funding period.

Figure 33: GIC clusters – Cumulative PCAIS 2.0 Expenditures as of 2023-24

Figure 34: DRAC – PAICE Expenditures

Summary

Conclusion: PCAIS 2.0 has helped to position Canada as a global leader in AI research and talent development. It has also helped to foster innovation, supported commercialization, and advanced equity and inclusion efforts, contributing to technological progress across sectors. At the same time, some key barriers remain. Challenges in commercializing and scaling AI solutions, coordinating national programming efforts, and difficulties building and deploying AI infrastructure in a timely manner are limiting Canada's ability to fully capitalize on its AI investments. Continued leadership, integration, and strategic investment are essential to unlock the full economic and societal potential of AI in Canada.

Finding 1: The second phase of Canada's Pan-Canadian Artificial Intelligence Strategy (PCAIS 2.0), and the funding it provides, were foundational for enhancing AI competitiveness and innovation in Canada. Through its commercialization-focused initiatives, PCAIS 2.0 has been helping to address key challenges, gaps, and barriers to the scaling, adoption, and commercialization of AI technologies in Canada.

Finding 2: Canada has built a strong AI foundation through early investments, including federal funding for initiatives such as PCAIS 2.0. However, gaps in coordination and cohesion remain, especially across federal initiatives and AI institutions.

Finding 3: Through PCAIS 2.0 funding, the NAIIs have offered impactful training to build both foundational and technical AI capacity, including programs for underrepresented groups. However, industry demand for this NAII programming exceeded the resources available to the NAIIs. GIC clusters have also supported training activities, particularly through 'on-the-job learning' in AI commercialization projects.

Finding 4: PCAIS 2.0 funding for the PAICE project aimed to expand AI compute access but faced a slow and complex rollout, with one-to-two-year delays across key activities, and experienced challenges with co-ordination and communication amongst stakeholders.

Finding 5: PCAIS 2.0 programs played an important role in strengthening Canada's AI ecosystem through industry engagement, capacity building, and collaborative partnerships. They have exceeded targets for engagement activities and facilitated connections between researchers, academia, and businesses. However, the importance of engagement on regulation and responsible AI development at the international level has increased significantly over the course of the PCAIS 2.0.

Finding 6: Through PCAIS 2.0 funding for the CIFAR AI Chairs program, Canada's AI research ecosystem has been strengthened, attracting top global talent, retaining researchers, and advancing progress toward key targets.

Finding 7: Through PCAIS 2.0, stable and flexible funding has enabled increased interdisciplinary AI research and publications, which have been disseminated to various audiences (e.g., academia, private sector, government, and public).

Finding 8: PCAIS 2.0 funding has contributed to strengthening Canada's AI ecosystem by enabling innovation and driving impacts in key sectors like healthcare, agriculture, and cleantech, while advancing AI technology, ethics, and safety and contributing to economic growth.

Finding 9: PCAIS 2.0 programs have contributed to strengthening Canada's AI commercialization landscape by helping industry partners build technical capacity, improve AI readiness, develop new or enhanced AI solutions, and generate new IP. Specifically, GIC clusters are driving strong applied, sector-focused AI commercialization, the NAIIs are focusing more on building foundational AI capabilities, and CIFAR is focusing on the creation and transfer of AI knowledge.

Finding 10: PCAIS 2.0 programs have contributed to improved company performance and growth. Businesses reported enhanced productivity, reduced costs, and faster decision-making due to AI-driven enhancements. Engagements supported job creation, particularly of Highly Qualified Personnel, and helped organizations establish, grow and compete.

Finding 11: Overall, PCAIS 2.0 contributions to intermediary organizations are being managed in an efficient manner. They met or exceeded match requirements, costs were similar to comparable organizations, and overhead largely remained within expected limits. However, insufficient centralized coordination was seen as a challenge by stakeholders.

Finding 12: PCAIS 2.0 delivery partners made meaningful progress on EDI through strategies and initiatives. Contribution Agreements are generally well-managed, though staff turnover and some reporting gaps presented minor challenges. Due to delays in the rollout of some initiatives, spending was less than anticipated for some organizations.

Recommendations

Recommendation 1: ISED should explore approaches and mechanisms to enhance cooperation, coordination, and synchronization across the different PCAIS program components as well as other AI-related ISED initiatives.

Recommendation 2: SRS should, in its oversight role, explore ways to strengthen coordination among stakeholders, including working with DRAC to clarify processes and timelines for funding agreements with partners, where feasible and with input from PAICE stakeholders, and apply these lessons learned to future AI compute infrastructure projects.

Recommendation 3: SRS should explore approaches to allocate specific support for CIFAR activities related to international AI partnerships and collaboration in areas such as responsible AI development, while ensuring alignment with other AI-related ISED initiatives.

Appendices

Appendix C: Logic Model

The logic model provided below serves as the PCAIS's roadmap. It outlines the immediate outcomes, the intermediate or medium-term outcomes, and the ultimate outcome.

Figure 35: Logic Model