AI has entered its accountability phase. At Invest Canada ’25, in a session presented by KPMG Canada, participants focused on what it takes to turn AI ambition into measurable portfolio value as adoption accelerates and the gap between experimentation and operational impact becomes harder to justify. The discussion centred on execution, including use-case definition, data readiness, talent constraints, and the risk of Canada’s research strength continuing to translate into commercial scale elsewhere.
At Invest Canada ’25, the conversation around AI shifted toward execution. Globally, generative AI funding has accelerated, and Canada is also putting real capacity behind adoption with the federal government’s $2B sovereign AI compute push announced in Budget 2024. The session titled Driving Value with AI, presented by KPMG in Canada, focused on practical outcomes and emerging friction points as investors seek to convert Canada’s leadership in research into realized value across portfolios.
The mainstage panel featured seasoned voices with direct exposure to AI commercialization, implementation, and scale. The speakers brought perspectives from academic institutions, applied AI labs, venture capital, and portfolio operators. What emerged was a picture of urgency, experimentation, and measured skepticism.
Canadian investors and portfolio companies are no longer debating whether to use AI. They are already deploying it. But the panelists noted that success depends less on the technology itself and more on problem clarity, data quality, and operational readiness. As one participant said, “If you can’t articulate the problem, the algorithm will only make it worse.” This shift from exploratory interest to use-case alignment is visible in how due diligence is evolving. Investors want to know how AI fits into a company’s operations, not just its pitch deck.
A central concern raised in the session was Canada’s lag in commercializing its research. While Canada remains strong in foundational AI science, the link between research hubs and venture outcomes is tenuous. Many of the AI-native companies achieving meaningful commercial traction are doing so outside the country. A few speakers questioned whether Canadian investors are too cautious in allocating capital to AI startups, especially in pre-commercial stages. This hesitancy could be leading to missed opportunities, particularly as large language models and generative tools mature rapidly. Commercialization determines where companies scale, where IP is retained, and where exit value accrues. Canada’s relatively low level of business adoption is part of that gap: Statistics Canada has reported that only 6.1% of businesses were using AI in Q2 2024.
Another theme was data access. Panelists stressed that access to clean, proprietary data is becoming a differentiator. Companies with a tight feedback loop between data, decisions, and product delivery are better positioned to extract value. One speaker observed that model performance is plateauing in many use cases. What matters more now is the speed at which a company can deploy, test, and iterate its AI-driven processes with minimal disruption.
That practicality carried into the discussion of where AI is currently creating the most measurable value. Participants cited verticals like finance, logistics, and healthcare, where time savings and predictive accuracy translate directly into margin gains. But there was a warning against overestimating short-term returns. AI requires sustained investment, often across several product cycles. One speaker emphasized the importance of aligning AI goals with business fundamentals rather than chasing novelty.
A number of GPs in the audience wanted to understand how AI impacts fund-level operations. Several examples emerged. Some funds are using AI internally to speed up deal sourcing by mining public and proprietary data. Others are exploring automated tools for portfolio monitoring. However, the appetite for deploying AI at the fund level remains conservative. One panelist suggested this is due to legal and reputational risk. Until LPs feel confident that these tools are reliable, adoption will likely remain cautious.
Another segment focused on the talent shortage. There was agreement that hiring AI talent remains a major constraint. Even as more students graduate with technical skills, those with deep experience building production-ready models are still rare. There was also discussion around the cultural gap between researchers and commercial teams. Bridging that gap requires strong translation skills and often depends on middle-layer talent who can speak both languages.
Notably, several participants pushed back against the narrative that AI is universally disruptive. For some sectors, adoption has hit limits. AI can improve operations, but it cannot compensate for poor fundamentals or weak business models. The market is starting to sort substance from hype, and investors are recalibrating their expectations.
There was also reflection on regulatory risks. While the Canadian government and others are working toward AI frameworks, uncertainty persists. For investors, this introduces complexity into long-term planning. Several panelists expressed the need for clearer guidance, particularly around data use and model accountability.
When the discussion turned to LP interests, a few observations stood out. First, LPs are asking GPs tougher questions about how they are using AI to drive operational efficiency across the portfolio. Second, funds marketing themselves as AI-savvy must demonstrate more than just AI-themed companies. They need to show evidence that AI is improving outcomes, whether through faster exits, stronger unit economics, or better diligence.
As the session closed, a few participants shared reflections on what investors need to do next. One speaker put it plainly: “Don’t wait for the perfect tool. Start with a real problem. Use AI to solve that, and the rest follows.” That pragmatism captured the tone of the discussion. AI is not a miracle solution. It is a set of methods that can amplify existing strengths when applied with purpose and discipline.
The session offered little room for hype and no appetite for theoretical optimism. Investors want what they always have: better outcomes, less risk, and scalable solutions. AI may offer those things, but only when approached with clear-eyed focus.
Since 1979, Invest Canada has been where Canada’s private capital community comes together to build relationships, close deals, and share real-world experience. It’s the definitive forum for GPs and LPs to connect, collaborate, and uncover new opportunities. In 2026, the Invest Canada conference will be taking place in Halifax, May 26-28. Learn more here.
