Julien Billot has been CEO of Scale AI, the federally funded, Montreal-based global innovation cluster, since its creation in 2018. Scale AI describes itself as “the central pillar of Canada’s AI ecosystem,” dedicated to building the next-generation value chain and boosting industry performance by leveraging AI technologies.

Billot also is an adjunct professor of management at HEC Montreal and the Montreal lead for two transformational programs aiming to launch and grow startups in artificial intelligence leveraging Montreal tech and business ecosystem, NextAI and the Creative Destruction Lab.

He has extensive experience in the marketing, media and mobile industries, with a track record of successfully executing print to digital business transformations.

Prior to Scale AI, Billot was president and CEO of Yellow Pages Group Corporation in Montreal. He also serves as a director in various companies and non-profit organizations. He is currently a board member at the Canadian Partnership Against Cancer, chairman of Montreal-based company Jogogo Media, a board member for DistrictM and a French media group, Groupe Paris Turf. 

Under Billot’s leadership, Scale AI has achieved some impressive publicly announced key performance indicators:

• 120 AI projects supported.
• $600 million co-invested in industry-led projects. 
• $5.2 billion in direct value generated.
• 5 small and medium-sized businesses participating per project. 

Billot talked with Mark Lowey, Research Money’s managing editor about the benefits for SMEs working with Scale AI, the accomplishments he’s most proud of, the biggest challenges Scale AI has faced, how Scale AI is helping companies leverage AI, whether Canada can be a global player in AI, his longer-term vision for Scale AI, and the potential impact on the global innovation cluster if there’s a change in the federal government in this year’s election.

R$: One of the main goals of the federal government’s global innovation cluster initiative was to engage more Canadian SMEs in innovative commercial projects with large corporations, to help SMEs scale up and expand their market reach. Has Scale AI been able to achieve this goal during the six years the innovation cluster has been in operation and, if so, how? Can you provide a couple of examples.

JB: When you look at the list of projects from Scale AI, you can have the feeling there are a lot of big companies. But the reality is more than 70 percent of the money we’ve spent went to SMEs. At the end of the day, we are doing two types of projects at Scale AI. One is what we call adoption projects. That’s typically Ravel by CF, Canadian Tire, Bombardier, Metro [and other big companies] leveraging AI in their own processes. Initially, most of the [AI] adopters were big companies. Now we’re seeing  more SMEs adopting AI.

But when we support the project for Bombardier, for example, most of the money doesn’t go to Bombardier. Most of the money goes to the people delivering the solution for Bombardier and these companies are SMEs. Even if our customers are big customers, when we reimburse a project, we direct reimbursement mainly to the companies building the solution and they are SMEs. Typically we will always refund 50 percent of the costs of a service company building the AI solution. We only pay between 20 and 40 percent of the internal costs of the adopters.

Even in the adoption projects, even if sometimes it’s big companies [involved] – and it’s less and less big companies, it’s more and more middle-range companies – most of the [Scale AI] money goes to small companies building the solution.

The second stream of what we do is commercialization. Commercialization is really helping companies build a new suite of projects leveraging AI and then selling it. In this case, that’s mostly SMEs. The reason why 70 percent of the money we’ve spent since [Scale AI’s] inception has gone to SMEs is three-fold. One is more and more SMEs adopting AI. Second, even if big companies are adopting AI, most of the expenses go to the SMEs providing the solution and service. And third, our stream of commercialization is mainly SMEs. So think of companies like Vooban, IVADO Labs, Moov AI, Mely.ai, Lemay.AI, Nuvoola AI and others on the service side. On the products side, companies like OVA, Plusgrade and BrainBox AI. We see more and more medium-sized companies adopting AI. In our recent portfolio, we had companies like MappedIn, Avenoir, Aéroport Montreal and others that are really adopting AI.

R$: Other than the benefit of getting funding from Scale AI, what sort of benefits does an SME get when it plugs into a project with a big company? How is intellectual property handled?

JB: When you are an SME, what you want to build is knowledge and references. Of course you want to have money. But most importantly you want to have references and knowledge to be able to resell your solution or your competence. Typically we’ve seen companies like Moov AI, Vooban, IVADO Labs and others going from companies with five or 10 people to now being 100- 150- and 200-people companies. We have not funded all their projects, so it’s not because of us directly, but because indirectly we [help] them by building their competence, building their knowledge, building their reputation. And then they’re able to leverage that in finding new customers.

Every time we have SMEs involved in projects we ask that SMEs keep at least [the IP they’ve bringing to the project]. So the intellectual property belongs to the small company building [the solution] or at least is co-shared with the company building it. The company building the solution has full ability to reuse either knowledge or IP for different customers after they’ve built the solution.

At the end of the day, it comes back to our mission at Scale AI, which is twofold. One is helping Canadian companies improve their productivity by leveraging AI. That’s the adoption piece, basically – we help Canada gain productivity. And at the same time, we create a local Canadian ecosystem of service and product providers in AI. This end is the critical thing. Our mission has never been only to help companies adopt AI. It was helping companies adopt AI and to help build that ecosystem, and to build that ecosystem you need companies to develop things and to keep intellectual property in what they are developing.

R$: What are two or three accomplishments that you’re most proud of during your time as CEO of Scale AI? Any new or recently launched initiatives or programs you’re particularly proud of?

JB: On one side, we see clearly AI [being adopted] by smaller companies. And that’s a big achievement. I always say Scale AI will be successful at the end when companies [don’t rely] on our funding, and that’s more and more what we see, in the sense that we have a huge pipeline and range of companies arriving. Most of them are first-time adopters [of AI], never did AI before. That’s really something we are very proud of.

And on the other side, we are very proud to see a lot of great Canadian service and product companies able to build solutions for [these companies adopting AI]. Having any one of [these accomplishments] would have been nice. But we were fighting to have more, which is having smaller companies adopt AI and having Canadian companies building solutions. That’s exactly right now what we have. We’re not the only one to help build that ecosystem. We were at VivaTech (the largest technology event in Europe) in Paris, France last year. We brought 60 companies. We had more than 140 companies applying to go to VivaTech. This year, we are going to have more than 100 companies going to VivaTech 2025. We expect to have between 200 and 300 companies applying – just in AI. So that’s really showing how the ecosystem of service and product providers in Canada has really grown.

Scale AI’s annual AI event, ALL IN, is really something important because ALL IN could not exist without a mature AI ecosystem. We created ALL IN in 2023 for one reason: because we believed the AI ecosystem in Canada was ready to be showcased. We showcased how companies are adopting AI and how solution providers are delivering on AI solutions. That’s exactly the same reason we went to VivaTech six months after [the first] ALL IN event.

Canada can play in this AI space if and only if Canada delivers on different things. The first one, and the most important one, is we need to have a local market. So we need to have companies adopting AI, because if there are no companies adopting AI, there’s no way companies will use or be able to support AI development.

So it’s very important in our mission for Canada to develop a local [market] for supporting AI adoption. Because if you don’t have a local market, then service companies cannot serve local companies. They will have been forced to go outside [of Canada]. And ultimately they will be bought by the U.S. or other countries, because they have no Canadian support. So we need to support local demand. But of course if you support local demand, at the same time you need to grow a local ecosystem creating Canadian IP.

Then, when it’s ready, you can begin to showcase this ecosystem internationally. And companies can export. That’s really the virtuous circle where we are. For example, there’s a very interesting company called Airudi. Airudi was created in Montreal and is doing AI in human resources. We helped them fund a project with the Port of Montreal. Airudi was going to develop their technology and sell it to a local customer. They grew their technology and we showcased them last year at VivaTech 2024, in Hanover Messe 2024 (a trade show for industrial technology). Ultimately they were able to sell their technology to serve the ports in France through their collaboration with SOGET (a French firm that supplies software and services to France’s ports). That’s really a great example of the cycle we need to follow.

R$: What are a couple of the biggest challenges do you think Scale AI has faced?

JB: The biggest challenge is money, obviously. The problem is there’s a gap between the market need and the money we have for funding. We had some extra money arriving in [last year’s] Fall Economic Statement. (The federal government committed $150 million over three years for the five global innovation clusters). The requests for funding right now are much higher than the funds we have, which should be considered as great news for the country because now finally companies are adopting AI and need support. Our biggest challenge right now is not generating demand. It’s really having the money to be able to deliver on the demand.

Initially, building momentum [took a long time]. In our first mandate, we had great traction post-COVID, because COVID was a very difficult time for AI. AI at the end of the day is mainly [utilizing] human resources so during COVID it was tough. But we had great acceleration in 2022 and 2023. Now the momentum is absolutely strong. We are going to do a record year. We already funded in December 2024 more projects than we did [in the previous year]. We have seen a huge, huge pipeline so we have a lot of demand, plenty of new companies. So the momentum is really strong. Our major challenge is having the money to sustain that momentum.

Why do these companies still need funding? When you are the CEO of a public company, you have two types of decisions. One is deciding OPEX, or operational costs. The other is deciding investment or CAPEX (capital expenditures). It’s always easy to innovate with OPEX because you pay as you generate revenues. Particularly in AI, there are very few solutions ready for most companies. You have OpenAI, Amazon and Microsoft who will claim they have plenty of solutions available, which is partially true. But in most industry projects there’s no off-the-shelf solution, no software-as-a-service (SaaS) solution. So you need to build customized solutions. When you were building a website back in the 2000s, you needed an agency to build your website. And then at one point, some platforms arrived – GoDaddy, Yellow Pages or WordPress – so now you can build a website with a click. AI is still very much in the stage where you need a customized agency approach. So building a project in AI is really a CAPEX investment. You need to invest a lot of money upfront and then possibly you will see a return on investment later.

So the risk profile for an organization is much higher, because you need to invest upfront. That’s why Scale AI is so [focused on] adoption and commercialization, because we decrease the level of risk by co-investing in this investment. The day most of AI is productized and you can sell it as a SaaS service, well perhaps Scale AI is not needed anymore. But that could take some time in the sense that most of AI is far from being productized. It took 20 years to get there with products to build websites, which is a very simple thing [compared with AI].

R$: Studies indicate that many Canadian businesses still lag behind those in the U.S. and other countries in adopting and deploying AI. How is Scale AI helping businesses to adopt and use AI?

JB: Most of our projects now mainly involve newcomers [to AI], because we know the first step is always the most difficult. At the end of the day, AI is really about business thinking. We talk a lot about AI but the way it is for companies is before introducing AI, you need to think about your business. You need to think about your processes. You need to decide how you can improve your processes using AI. So in leveraging AI, you look at your processes, you look at prescription and prediction. If you have a certain business process, if you could improve your prediction, how could you improve your process? If you could improve your prescription, how could you improve your process? If it’s the case, then you can look at implementing AI and then you can do it. It’s not introducing AI for the sake of introducing AI. It’s really a business process first.

When we look at training, it’s not training people about AI technologies. There are many people doing that. We try to train companies to think about their business processes and how they could introduce AI in their business processes. So our priority in education and training is really having people think business and AI into their business, and not think about what is AI and what is the technology, because that’s not the relevant question. The relevant question for us is: You’re a company. You need to think business first. You need to look at your processes, understand what processes could be improved or transformed leveraging AI. At the end, the outcome of any training we are funding is to say [to businesses]: Those are the processes with the most potential for AI introduction and that’s the complexity of AI introduction, and getting companies to prioritize their next step.

For websites, there was better penetration when platforms [to build your own website] became available. It’s really the same for AI. We’re really moving to when small companies, SMEs, will be able to buy AI solutions on the shelf. To really have a much bigger penetration of AI, you need to change in the solution provided, a change in the business model, the existence of horizontal and vertical platforms customized by a task or by a vertical or an industry. (A horizontal platform provides general functionality that can be applied across various industries. A vertical platform caters to a specific industry or niche market). And that will take a lot of time. Yes, AI is touching more and more companies. But when companies are doing AI using Co-pilot and Microsoft, that’s not really using AI. As I said, you use AI when you look at your business processes and decide that you can change your processes by leveraging AI. And very few companies are doing that. More and more are doing it, but still a very small number. And the [penetration] will grow once more on-the-shelf solutions exist.

R$: There’s an ongoing discussion, if not debate, in Canada about whether the country has deep enough pockets to be a global player in AI development. Some argue that we don’t have the money to do competitive AI development in areas like large language models and instead should focus on delivering specific, made-in-Canada AI applications in strategic sectors where Canada could benefit. What’s your view on this? Can – and should – Canada do both?

JB: DeepSeek in China is showing that big is not always beautiful. There is some space for companies that innovate and create something different with much less money. Cohere [in Montreal] is a good example. Mistral AI in France is good example. You can still disrupt this technology. The game is not over. If the game is not over, then you have a chance to play. On the application side, you have to develop your local market, help your local ecosystem. That’s the good way to grab part of the value chain on the application side.

The other play is the infrastructure side. If you have part of the AI services staying at home, that’s jobs at home, that’s investment. Building [local] infrastructure means instead of paying taxes in the U.S., you pay local taxes in Canada. So you need to play at both ends. One is on the application side, what most of the customers are using. And on the other side, you need to play in the hardware and real estate because that’s a margin you can keep. Perhaps we have hopefully in Canada some players that can play globally on the application side, players like Coveo, Cohere, BrainBox AI, Airudi and others. And on the other side, we have to be sure that most of the [AI] traffic stays in Canada because at the end of the day everybody has to pay for traffic or data centres. If these centres are in Canada and not in the U.S., the money will stay in Canada and the taxes will stay in Canada. So we have to play at the two extremes of the value chain – infrastructure on one side and applications on the other side.

R$: How important is training and education to Scale AI and to Canada’s AI industry in general? Can you provide a couple of examples of Scale AI’s initiatives in these two areas?

JB: In phase one [of Scale AI] we had a lot of training in AI, everything about AI. We changed that in 2023. We started coaching how as a company you can introduce AI in your business processes. So less training and more coaching. We are funding it and it’s delivered by third parties. The reason we did that [made this change] is that a lot of people are delivering training and we didn’t think it was really our mission. Our mission is to help companies adopt AI. So helping companies understand how they can identify their critical business processes, how they can introduce AI in these business processes, and how they can do it – in what sequence and what priority. It’s very important to have more companies coming in and delivering projects. So we decided to focus our investment really on this coaching piece, more than just technical training on AI.

When it comes to education, we support STEM programs. We are investing up to $1 million in a STEM Youth Awareness Program [to encourage] young people and minorities to be interested in science and technology. One of our biggest challenges is we don’t have enough money [to do everything]. So we have to choose. Our support for STEM programs is just a small bit of our activity, although definitely we cannot spend the time and the money to [do this in person]. We have a very small team. Our priority is really on the business side and the company side.

 R$: What’s your longer-term vision for Scale AI? Where would you like to see the cluster in 10 years and does your vision include Scale AI becoming financially self-sustaining?

JB: The financially self-sustaining piece is interesting because we don’t need a lot of money just to operate. Our cost of operations is between $5 million and $6 million a year, including our ALL IN event. So it’s not so expensive. The big question is how long will companies need us. That comes back to the availability of product in AI. My feeling is if AI is still customization or on-demand, we have still a need for Scale AI because customization means CAPEX [spending] and companies will need the support to reduce the risk for investment. The day all AI is productized, perhaps we don’t need Scale AI in the same way. So the big question is how long [significant productization] will take. And how long it will take [relates to] avoiding having only American service providers. You don’t want to have only Microsoft, Google, Amazon and others providing AI solutions. So how long do we need to have a strong ecosystem of Canadian companies developing SaaS solutions?

Our mission is supposed to end in 2028. I doubt the fact that our mission will be over [by then]. In five or 10 years from now, it’s maybe a different story. At one point, possibly nobody will need Scale AI for the funding part. Perhaps we’ll need Scale AI for the advocacy part, for ALL In, or helping companies to export. But the funding piece is still very, very critical when AI is still a service technology and not a product technology.

R$: The polls are telling us that we’re likely to see a change in the federal government next year. If that happens, do you think the work started by Scale AI will continue, even if there’s no further federal funding for Scale AI and the other global innovation clusters?  

JB: I’m just going to say I believe AI is important for Canada. I don’t think AI is a Liberal thing or a Conservative thing. I think AI is important for Canada, for two reasons. One is productivity is a Canadian issue, whether you vote Conservative or Liberal. And AI is one way to solve the productivity issue. Whatever government [is in power] should really look at AI and productivity issues in Canada. The second reason is to position to play that [AI] game for the next 20 years or 30 years, you have to take [the steps now]. If you wake up in 10 years, that will be much too late. So it’s now you need to invest in infrastructure. It’s now you need to invest in service providers. If you believe in AI, you clearly understand the game is win or lose in the next five years. Every government should have that in mind, whatever the party.

Then the last question is: What’s the best vehicle to do it? And that’s an open question. You can consider that a cluster or a non-profit organization like Scale AI is the best way to do it. You can say [government] administration is the best way to do it. The only thing I can tell is from a Scale AI perspective our cost of operation is very low. We are operating around eight or nine percent cost on the funds we deliver. I think it’s quite low in terms of cost of operation. So I think we’re an efficient way to help the right projects, the right companies. The question is whether AI is important and I think AI will be important whoever wins an election.

What we have is expertise at Scale AI. We know what are the right application business processes can be influenced by AI. We know the type of return [a business] can expect. We know how to structure a project. We know how to identify if a product has the right assets, the right knowledge, the right people to deliver. We know how to prioritize and we know how to follow the projects and to extract the information to share with the community. That’s what we know. In fairness, I think that’s a unique asset for Canada. There are very few public organizations in the world that have that knowledge. We’ve seen at the end of the day close to 200 projects in AI, just AI implementation in industry. The value of these projects is around $700 million – just software, not hardware. There are very few players that have so much vision and expertise in knowing what is a real AI project in industry. That’s an asset Canadian taxpayers built and of course I think it would be foolish to destroy it. At the end of the day you can use this asset for Scale AI. But any administration, any provincial government, could use Scale AI for its own projects. We have the knowledge to help. It took us six years to build [this asset]. It’s really an incredible knowledge.  

R$