Technology in-house roundtable examines Canada’s competitive position

In July 2026, Canadian Lawyer’s Leaders network convened its inaugural technology in-house roundtable, bringing together in-house legal leaders from Canada’s technology industry. Participants gave their views on and experiences working in Canadian tech, including what makes a national AI champion; talent, compute, and infrastructure; data sovereignty; the build vs. buy debate; IP protection; and what government should do to ensure Canadian companies can succeed. Quotes have been edited for length and clarity. 

Participants 

Topic 1: Building a Canadian AI Company – Independence vs. Foreign Capital 

Q (Tim Wilbur): If you had to choose between two options – (A) building a globally competitive Canadian AI company that relies heavily on foreign capital and infrastructure, or (B) building a smaller Canadian AI company that remains independent and Canadian-controlled – which would you pick? 

Melissa Reiter 

I’m a “why not both” person at heart, and I have a hard time with that dichotomy. What came to mind is: why can’t we use foreign capital and infrastructure while keeping data and IP rights anchored in Canada through contractual provisions and the way we negotiate those deals? Looking at it through a contracting, risk, and governance lens, independence comes from how you negotiate, not necessarily from where your capital originates. 

I’m not sure it’s realistic to avoid global players entirely, though it would be interesting if someone could structure something that exists in a very contained way. Being plugged into global accelerators could be very useful for the right type of company. That’s where Legal earns its seat at the table – navigating the negotiations that help keep the crown jewels within the Canadian sphere. 

Sahil Razdan 

The first thing I would add is that I am speaking on my own behalf and that these are my opinions, not necessarily the views of Samsung Electronics. 

I’m inclined toward Melissa’s position. Done correctly, you’d get the best of both worlds. But if I truly had to choose one at this juncture, I’d lean toward Option B – building a smaller Canadian AI company that remains independent. 

The primary reason is that it lessens the impact of foreign investment and the agendas foreign investors might bring, including potential misunderstandings of Canadian law and societal expectations around AI, data privacy, and responsible use. We’re seeing this play out in a different context right now with the FIFA example, where an entity with foreign financial interests can influence actions in ways that may not serve the broader group’s best interests. It’s a different situation, but it illustrates what can happen when someone with foreign skin in the game can push actions that don’t necessarily serve what you’re trying to build. So at this juncture, based on where AI is in this country, I’d choose Option B. 

Laure Fouin 

I’m similar to the others – “why not both” if possible. My view is slightly different, however, on whether it’s still possible to build a globally competitive Canadian AI company. I’m not convinced we still have that capacity right now. 

From my vantage point at Coinbase, I work across multiple jurisdictions. What I’ve observed is that the countries that get specific and tailored attention from a global company like ours are the ones that have both high revenue potential and clear laws with regulators who actually enforce them. Canada has a Coinbase Canada entity focused on Canadian users and laws – that’s because we have both strong revenues and strong regulators here. 

Extrapolating to AI: if we want Canada to have a stronger global position, we need stronger, enforceable rules about where data sits. Commercial negotiation, as Melissa noted, only works when you have the stronger hand. The most effective way to achieve that is when you can tell your counterpart, “This is not my choice – it’s the regulation I’m subject to. You have to build a data centre here.” I don’t usually advocate for more regulation, but when it comes to imposing Canadian requirements on global players, it is the most effective tool available. With the current government focused on competitive positioning, I think there is an opportunity to advocate for this. 

Robert Tremblay 

I agree with Melissa. The question reads as: Is it better to be completely independent and Canadian-controlled, or to also leverage foreign resources? I think the answer is balance. Not engaging with foreign talent at all, or not using hyperscalers and global accelerators, would kneecap you. Canada has amazing talent, but leveraging foreign resources where appropriate is a strength. If you didn’t engage with the major players, you wouldn’t be able to build a company that scales fast enough. But that doesn’t mean you have to sacrifice what makes you Canadian – building talent and a company here. 

Topic 2: Defining a National AI Champion 

Q (Tim Wilbur): When you hear the phrase “national AI champion,” what does that mean in practical, concrete terms? 

Robert Tremblay 

A company demonstrating leadership and innovation – developing internationally competitive, leading-edge capabilities – while still serving as a flag bearer for Canada with a real focus on building here. That means building jobs in Canada, performing innovation here, and creating an ecosystem around it. The talent pool gets deeper as a result of having that kind of company here. And beyond the technology itself, it’s about building all the surrounding capabilities: infrastructure, the opportunity to keep Canadian data in Canada. All of those things go together. 

Melissa Reiter 

I like the “flag bearer” framing because it can embody a number of different ideas. You’re out there representing what AI means for Canadian companies at scale. What Laure said about leveraging regulation is also really interesting – so much of what it means to be a champion depends on your positioning. If you have the ability to set the terms, you can accomplish a lot. 

I think some of it is also about leading the conversation around what “good” looks like in this space – not just domestically, but internationally. Being a national AI champion is not just about getting the technology into people’s hands. It’s about building it in a way that meets consumer needs while also aligning with the government’s desire for safety and soundness in the system. 

Sahil Razdan 

I initially read “national AI champion” as referring to a single entity that pushes AI governance and regulation in Canada, but I’m not convinced that’s the right model. For AI to work well and responsibly, we all need to be on the same page about how it works, how we contract, and how we set industry standards. 

I think all public and private stakeholders should be consulted and represented. Anybody or legislation that governs AI in this country should be as universally accepted as possible. It won’t satisfy everyone, but it should incorporate private and public companies, government, and all stakeholders, each with different use cases and understandings of what responsible AI looks like. That’s what a national AI champion looks like to me: a universally understood and accepted framework that takes everyone’s needs into consideration. 

Laure Fouin 

Canada had a genuine advantage in the crypto space for years because we had our “FTX moment” in 2018, not 2022 – four years earlier than the US. I could tell counterparts in other countries that we’d been dealing with securities regulators for years, that we had regulatory clarity. That was a real competitive advantage for a long time. 

My view is that the quicker Canada can reach the stage Sahil described – a common understanding of what doing AI well means and how it should be implemented in Canada – the greater the opportunity to become a leader. As a user of AI rather than a provider, what would make us choose a Canadian company over an international one is if it were more plug-and-play for us: if it allowed us to use more sensitive data types, including personally identifiable information, under Canadian regulations. Some international tools simply can’t support that for a Canadian financial institution. Clarity – wherever it comes from, whether industry consensus or regulatory enforcement – is the key. 

Topic 3: Canada’s Competitive Position – Talent, Compute, and Infrastructure 

Q (Tim Wilbur): What’s the environment in Canada right now in terms of capital, compute, talent, and data – and where are the gaps compared to other jurisdictions? 

Sahil Razdan 

We have all the ingredients and all the talent to be globally competitive. Canada has been a leader in AI for a long time – we’ve been telecom leaders, point-of-sale leaders, and leaders in data privacy protection. The godfather of AI is Canadian. The talent is here. The unfortunate history is that we’ve allowed a lot of it to go south of the border, to places with deeper pockets and bigger wallets. We haven’t always fostered and retained what we have. But I hope that, with the momentum building around AI in Canada, we can change that – keep that talent in the country, grow our homegrown AI companies, and attract global talent and companies here if we do it right. 

Melissa Reiter 

Attracting and retaining talent is one of the biggest challenges – and it’s not a new challenge that came with AI. The best people always have choices. In a jurisdiction like Canada, where the regulatory environment can create more friction than in some other places, you’ll have to demonstrate that you can enable the kind of speed and scale that top candidates are looking for. 

That’s actually a great opportunity for Legal. If you can demonstrate governance practices that reduce friction – really tight, intentional processes that minimize drag within Canadian regulatory constraints – that can be a genuine competitive advantage. I would build that into the story when trying to convince top candidates to embed themselves in a Canadian company rather than elsewhere globally. 

Laure Fouin 

Coinbase has a fully remote, global hiring policy – we hire wherever you are if you’re the best person for the role. What that has led to is that after the US, Canada is our biggest employment base. That’s where we have the most employees, simply because when we select based on technical capability rather than location, we hire most there. I think that’s a very significant data point about the strength of Canadian talent. 

The other thing I’ve noticed is that the deep pockets of AI companies aren’t just pulling Canadians to the US – they’re pulling from every tech company. Crypto companies are feeling that pressure too, from the very high salaries AI companies are currently offering. What Coinbase has found is that we retain talent because of our specific mission, and because people want to stay in Canada if they can. 

Robert Tremblay 

Canada has a great labour market and great researchers. Much of the AI innovation came out of the University of Toronto. When we hire the best person for the job globally, many of those people turn out to be in Canada. So being in Canada can actually be a competitive advantage. 

On compute: access to computing resources – particularly GPUs – is highly competitive globally right now. There is a finite amount of hardware and a huge shortage, which is part of why you see companies making a concerted push to build AI infrastructure in Canada. But that also speaks to the earlier point about leveraging foreign resources when appropriate. If you tried to rely exclusively on Canadian infrastructure, you’d be holding yourself back. The point isn’t to avoid foreign resources entirely – it’s to maintain Canadian identity while strategically accessing what’s available globally. 

Topic 4: Data Sovereignty 

Q (Tim Wilbur): Is data sovereignty – where data lives and under whose control – something your companies are actively managing, and how does it factor into your decisions? 

Melissa Reiter 

The majority of our customers are in the United States, but we have a large number in Canada, and we are a Canadian company, so the question of where things live is always present. I think Legal is the right owner of this question: what are we giving away to grow quickly? 

Trying to balance maintaining control over certain types of data with the imperative to grow and scale at considerable speed feels like it’s worth asking: Will there be a cost to doing something a particular way later? Rather than taking a hard line – everything must be here – we look at overall cost and the flexibility we retain in making those decisions. Having ownership of key technology and key intellectual property resident in Canada, and maintaining flexibility in our service providers, is critical to our ability to shift as international environments change. It’s always a balance between what we need to do now and the longer-term implications. 

Sahil Razdan 

A lot of our decisions are made through a global lens, often from our headquarters in Korea, and they have to consider the most efficient approach across all subsidiaries worldwide. But local privacy laws and governance must always be considered. If the best approach – meaning the cheapest and most efficient – is to keep data in-country, that’s the decision that gets made. If there are other business factors at the global level, those come into play as well. 

Laure Fouin 

This hasn’t become a major issue for us yet. The way we’re currently using AI at Coinbase, we’ve been particularly careful – no personally identifiable information is involved in our current AI use cases, so we’ve worked around the data sovereignty question for now. But I think it will become important faster than we expect. When it does, we’ll need to move very quickly. Being able to find providers at that point who can offer the right level of security will be important. 

Robert Tremblay 

Cohere’s Canadian identity is attractive to customers concerned about AI sovereignty. There are only a few AI companies in the world that aren’t either American or Chinese, so being a “middle power” option resonates. It’s not just data residency – it’s also about who created the models. A recent public example saw a leading model removed from the market due to export controls in one of the major jurisdictions. That kind of thing alarms customers. They look to us as an option that gives them more control, something not subject to foreign powers. Depending on the use case and risk profile, that sovereign option is increasingly the key driver of Cohere's attractiveness. 

Topic 5: Build vs. Buy 

Q (Tim Wilbur): When your companies are deciding whether to build AI capability internally or procure it from an external provider, how do you make that decision? 

Melissa Reiter 

It comes down to cost and time. If something is easier to plug and play and the cost isn’t prohibitive, we might look externally. But our engineering team is constantly building and improving the product, so if it makes sense to build something internally, there wouldn’t be an aversion to that – as long as we can continue to deliver on the quality promise customers expect from Jobber. Ultimately, everything we do is driven by the desire to offer customers something that actually helps their businesses operate better and faster, whether that’s internally or customer-facing. 

Sahil Razdan 

Not much different from what Melissa described. It’s a mix of both, and it really comes down to cost and efficiency. We probably skew toward building internally first – that would be the ideal situation. But if plug-and-play is cheaper and faster, that works too. Whatever option is the cheapest and most efficient is ultimately what we choose. 

Laure Fouin 

We don’t build models ourselves. We build on top of models that we buy, using tools such as Cursor, LibreChat, and Glean, which plug in using MCPs and different model types depending on the task. Within that framework, we do build in-house. Every employee has been tasked with automating aspects of their own work. We start internally, and only if we reach the limits of what we can do in-house – even with support from our dev team – do we go out and procure. 

Robert Tremblay 

Like the others, it often comes down to time and cost. If we need to add a certain type of functionality to our software platform, the question is: do we use internal talent and take six months, or do we acquire that capability from someone else? The build-versus-buy conversation has become more prominent for us recently. A few years ago, the focus was closer to just building models. Now we have a software platform built on top of models that companies can use to build agents. Extending that platform’s functionality is increasingly the focus, and it’s a more traditional software consideration – build vs. buy comes up with increasing frequency. 

Topic 6: IP Protection in AI 

Q (Tim Wilbur): How do you manage IP protection when using or building AI – particularly around what a provider can do with your data or proprietary processes? 

Melissa Reiter 

To the extent we can protect our IP – the outputs, the user experience, what gets generated – we’re actively trying to do that. Sometimes you can negotiate those protections; often you can’t. The issues that are live for me include: limiting what a provider can learn on our intellectual property, controlling where it goes, and identifying gaps in IP indemnities. Whether terms can change over time is also a concern. We are actively working to control what we can in those negotiations, even when that’s not always possible. 

Sahil Razdan 

For Samsung, protecting our brand and our IP is paramount given our global presence. That extends across AI tool usage, device functionality, and the outputs those tools produce. It becomes challenging when you’re working with so many different AI providers and platforms – especially when some platforms are built on top of others – but we do the best we can throughout. 

One thing I’ll add, which is more of a personal perspective: I think the major area of focus going forward will be inputs, not just outputs. The more specific and thorough the input, the better the output. If AI models eventually regress to the mean – if most platforms produce similar outputs given similar inputs – then differentiation will come from the inputs themselves. I think inputs will become the protectable commodity in this space. 

Laure Fouin 

The stakes around IP protection are extremely high at Coinbase. The agents that our most valuable employees have built are highly effective with the current tools and current models. What happens if the underlying model is deprecated, or if the person who built the agent leaves? If the input changes because a new employee is trying to use someone else’s agent, is it still as effective? Probably not. Protecting the efficiency gains we’ve made over the last six months is something we don’t yet have a full solution to. 

Robert Tremblay 

This space is evolving quickly. A year ago, conversations with customers were primarily about inputs, outputs, and custom models – companies wanting assurance they own their inputs and outputs. We have a consistent framework for that. But as AI has moved into agentic use cases – software on top of AI, workflows – we’ve had to map those new product features onto existing IP frameworks in real time. What should a customer reasonably own from an agent or a workflow? What do we need to retain ownership over? Getting to a common understanding between our team and customer counsel about what the IP pieces are and who should own what is a significant part of our customer conversations right now. 

Year over year, as AI develops, this conversation gets more complex. We’ve continued to develop our internal approach to reach a consistent, shared understanding with customers. 

Topic 7: What Government Should Do 

Q (Tim Wilbur): If there was one thing you would say to the government about what it should do to ensure Canadian companies can succeed as AI champions, what would it be? 

Melissa Reiter 

Industry consultation is the obvious answer to advocate for, but I genuinely believe it’s important. Keeping that line of communication open will help ensure that regulations don’t become overly restrictive and have a chilling effect on business. Until Canadian regulations are finalized, EU regulations tend to be the primary driver of global compliance, and the American “move fast, break things” approach doesn’t fit everyone either. I always advocate for good governance as a competitive advantage. 

But the government also needs to understand that this is a very fast-moving, rapidly regulated market. How they work with businesses to understand the space – and how realistic they are – matters enormously, because by the time decisions are made, the landscape will have shifted. The goal must be a framework that doesn’t prevent Canada from also being a leader in this space. 

Laure Fouin 

I’d echo Melissa’s first point: industry consultation, and genuinely understanding the actual risks – addressing the real ones, not imagined or inappropriately framed ones. I really hope the government will learn from the crypto experience. Canada started ahead of everyone in crypto regulation, but now that regulation is stifling innovation, and Canadians don’t have access to products that others do because the regulatory focus has shifted to risks that don’t even properly apply to crypto. We tried to fit a square peg into a round hole. 

For AI, the government should consult industry, understand the space, and – critically – hire people who actually understand the industry before writing legislation. If you’re writing legislation about an industry, you should have full-time people on-site who understand it before you begin. That’s what I think is the key to success. 

Sahil Razdan 

Very similar thoughts. As a country, we tend to rely heavily on what the EU, the UK, and to some degree the US are doing. The US sometimes takes a “do first, ask forgiveness later” approach, which doesn’t really work in Canada – we tend to be more risk-averse and more consumer-centric, which aren’t bad things. But the fastest path forward is industry consultation, stakeholder consultation, and truly understanding how commerce is affected: what the pain points are and where there are opportunities for improvement. Getting that full picture from a comprehensive consultation is the fastest and most responsible way to move Canada forward in this space.