There are conditions where reliance becomes a limitation. In those conditions, priority shifts towards control of what's already in place. And from then on, there's no going back. We define what comes next on our own ground, on our own terms, starting right now.
Wow, amazing. So great to be here everyone and thanks so much for coming out. I feel like we have had some incredible conversations already this morning. It's a bit of a tough act to follow, but here we are to talk about sovereignty. And really, if we were having this panel three years ago, I think this would have been primarily a policy conversation. But now in this country, we have some incredible companies building the tech, the products, and really innovating in these areas. We have an entire capital stack set up to support these companies. And you know, there's lots of conversations I have with entrepreneurs and with policy makers about sovereignty, why we care about this. And the thing that stands out to me is that nobody really has a common definition of what we mean when we talk about sovereignty. And so that's kind of what I wanted to start with with this panel is what are we even talking about when we say we're building sovereign tech? We need sovereign tech in Canada. And I thought I'd actually start with you, Elliot, to tell us a little bit more about how you think about that. At Dominion, you've been around for just less than a year. You've raised over $23 million scaling super fast. And you've recently been appointed to the US-Canada committee on economic relations. And so I know you're hearing lots of different perspectives in the space, but yeah, kick it over to you. What does sovereignty mean to you?
Well, thanks for having me. It's super exciting being here and seeing so many technologists and entrepreneurs and honored to be on such an awesome panel. I guess the way the government thinks about sovereignty is really protecting people, keeping them safe and protecting borders. I think the way that Dominion thinks about it comes out of military doctrine, something called the OODA loop. Basically, observe, orient, decide, and act. We're principally focused on the O on the beginning and the A on the end. So, how do you observe? How do you own the assets to observe what the heck is happening in the Arctic? You know, we have 230,000 km of coastline. We really have no idea what's going on. And then how do you act on what you observe? And technology has massive roles to play on both of those. The orient and decide is principally an organizational political thing, but we're focused on the kind of bookends of that.
Great. And Mina, Kepler Communications is doing a lot of innovative stuff in the space sector. How do you, when you think about sovereignty and building a sovereign company, what does that mean to you and how do you think about that?
So I think the way we end up seeing it with a lot of our government customers, sovereignty oftentimes quickly turns into a question of control. Control and freedom of action. Those are kind of the two priority areas. Can I do what I want to do without any incumbrances? And do I have the opportunity to control my assets, control my space? And those are the two kind of bookends that we think about oftentimes from a sovereign perspective. And there's a lot of these concepts from sovereignty that doesn't come down to in fact ownership of physical goods, assets or hardware. And we've seen sovereign mechanisms at play because ownership of everything is just entirely impractical where people go out and have contractual mechanisms. A great example of that is DPAS ratings in the US where you go out and you can prioritize a program. They can hit a button and say this is DPAS rated and you need to prioritize the delivery of any of your goods and services to us first.
And it's interesting like when you look at a company like Kepler, in fact, and we were talking about this backstage, you know, you are building a lot in Canada, you recently launched with SpaceX's Falcon 9 in January. And so we're not, you know, when I look at Kepler, I'm thinking this is a Canadian company doing some amazing things in Canada, but you're not only building in Canada, you are relying on other elements of the global supply chain.
Yeah. It's completely impractical for us to kind of own everything because to replicate the reliability and reuse of a Falcon 9 rocket is just something that's going to take you 10 to 20 years to go do and you'll be prioritizing the whole of your business around that. Or to replicate the incredible lasers and optics that they built out in Germany, that's something that'll take another decade to produce. So you have to be practical about it from a sovereign perspective and really narrow it down to control and freedom of action. And from our perspective, we end up finding that a lot of customers are pretty comfortable. If you can show them control, show them freedom of action, then they get pretty confident. But we have a factory still here in Toronto where we manufacture satellites in downtown Toronto because we need that for our own control and freedom of action.
And Nick, we've got you here with Cohere. Congratulations on the recent deal with Aleph Alpha in Germany. When you think about sovereignty, I'm curious to get your perspective. I mean, AI is obviously a topic that is on the tip of everyone's tongues. When you go to Ottawa and you're talking to policy makers about AI, there's lots of different perspectives. I think a lot of excitement and energy to, you know, support and build up companies like Cohere so that we can have sovereign AI and control over that in our country. When you're looking at it from a builder's perspective and what you're focused on at Cohere, what does that mean to you? Do you view Cohere as a sovereign company, a sovereign tech company in Canada?
Yeah, I think sovereignty is fundamentally what we provide. But I think about it maybe a little broader than some of the conversation already. Like when I think about sovereignty, really I just mean agency and autonomy. Like we use the word sovereignty a lot, really I just mean agency and autonomy and I don't mean that just for a country. I know like the word sovereignty is also thought of only as something a country has but I think sovereignty is something that can happen at all spatial scales like a person, a group of people, a company, a country and every one of those can have autonomy and agency and technology can either be a force that provides that for them, that gives them more autonomy, more agency, more of an ability to do what they're interested in and like, you know, free up their time and things or it can be a force that locks them down and we've seen technology lock people down and take away their sovereignty a lot over the past 20 years and we've seen technology and hegemony of technology remove sovereignty from companies, remove sovereignty from countries. And I think AI right now like as a technology it can kind of go one of two ways it can either be a technology that empowers people or similarly to, you know, mobile phones and the internet it can be something that a lot of people feel fundamentally disempowers them. So yeah, I really want LLMs to be something that people think empowers them, that companies think empowers them, that countries subsequently think empowers them. But the tech has to be built and deployed in the right way in order for it to go that way. So that's something that we're thinking about all the time at Cohere.
And what is the right way?
For me, the right way is like open. It's one in which the models are released for people to use and to run locally. It's one in which it can be deployed privately and on-prem so that people can run it on real data and it can be controlled by that organization or person that's using it and it can't be shut off by whoever made the model. So for us, we just released our most recent model open source. We've been releasing our models with open weights for the entire time of this company. So if anybody wants to look at the weights of the models we make, they're out there. And we do that because I think that pushes it into the direction of sovereignty enabling for people, for companies, for countries.
Yeah, it is interesting because when I think about sovereignty and sovereign tech, I really, you know, and Toby was up on stage earlier talking a little bit about this around how, you know, we need to fundamentally build globally competitive companies. And I worry sometimes that our conversation about sovereign tech indexes a little bit towards protectionism and building, you know, Canadian tech in Canada only for Canadian companies and sort of protecting that within a bubble. I know that's not the perspective of everyone on stage here, but like let's dive into that a little bit. You know, Elliot, I see you smiling, so I'll kick it over to you. Like where on the spectrum should we land from, you know, protected subsidized sectors to, you know, something totally open and globally competitive like what should we be aiming for across that stack?
I have high conviction about what we need to protect and those are technologies and institutions that protect our right to have a country, protect our fundamental sovereignty and dignity and right to exist. So like just back to this OODA loop of like do we have the capability to observe what is happening in this country on the borders and do we have the ability to act on those actions of those things that we observe. But I actually think that countries that talk a lot about sovereignty have like a fundamental anxiety about keeping it. And one of the things I would like us to go back to and what Dominion is trying to kind of rebuild is we didn't really talk about sovereignty in the '50s, '60s and '70s, but we built amazing stuff. We built the first passenger jet. We built the first two operational drones. We built the first autonomous submarine. We built incredible stuff. And we got our sovereignty through excellence and ambition and actually generated the leverage with not just the Americans but the global community by building incredible stuff and we did that because we had a customer. So I'm not sure if like the protection is the right way to think about it. The minimum thing that we need is a competent capable customer in the government. I'm not asking for protection. I want to beat Anduril. I want to beat Helsing. I want to beat Lockheed but I am asking for a competent capable rapid customer that is like a fundamental requirement I think of getting back to that ambition.
How is it trying to get governments in Canada, both, I mean mostly I think you're working with federal government but maybe provinces as well, how are those conversations, how is it going with, you know, trying to navigate that and convincing, you know, government to be that first customer because I do hear this a lot that, you know, and it's not just in these sectors represented on stage but, you know, a lot of companies I think have a challenging time in Canada because they go and try to sell to government and the first question is well like who else have you sold to globally and so there's this idea that to be successful in Canada you need to go prove yourself globally first and then, you know, you'll be able to get government as a customer here but Mina, like what are you, like how is that going for you and what are you seeing?
So I think our story arc is pretty similar. We started the business. We raise capital from the US. We sell to the US as our first customer. Then we go sell to Europe as our second customer. And we come back to Canada and say, 'Okay, now we're a qualified entity.' And even at that point in time, I think Elliot and I wrote a memo together for Build Canada where we note that the average procurement cycle is 16 and a half years. Like so 16 and a half years. So yeah, you know, so if you even think about that conceptually, that's beyond the lifespan of any company starting up to ever care about government as a user. This just becomes like an afterthought that maybe sometime in the far-off future they can become a user. And that procurement reform is live and vibrant in both the US and Europe. We're seeing some early innings of it in Canada where there's willingness to make change. But I think by default because of the nature of the environment we're in, we have to be globally competitive like the US Space Force has to pick us above, you know, local domestic company and it's a performance orientation above anything else. The Europeans who have a well-regulated economy also have to make that same decision. But if you're performant on a global basis, then they're willing to work with you. And we have to see some of that same reform happen in Canada because otherwise, you know, founders in this room are just not going to care about some project that's going to take 16 and a half years. They'll go find another problem that's going to be interesting to them.
So, do you think overall that Buy Canadian policies are helping or hurting?
I think it's way less about Buy Canadian policies. I think it's all around, you know, having these competent customers like Mina mentioned in government who come in and say, 'Hey, I'm willing to buy a good from you' and they have air cover or policy air cover to go make that decision, even though it may work out well or poorly. And it doesn't have to be a large procurement scale. It's got to be with a linear pathway or a step function pathway that somebody can go in and say, 'Okay, you have a product, in our case, like data from space. I'm willing to buy a million dollars worth of it. If it works, I'm going to buy five. If it works, I'm going to buy 10. If it works, I'm going to actually put you on this pathway towards a many hundred million dollar program with us.' And that mentality has to creep in because oftentimes we see there's just a risk aversion like even if I spend a million dollars with you and it doesn't work out well, that's going to compromise my next promotion. And that mentality really has to shift.
Yeah. What are you seeing, Nick, in your conversations around procuring AI within government? Are you seeing a shift in that mentality or is it sort of more status quo?
Yeah, I think 16 years is the average procurement cycle. That's crazy. I didn't know that. That's crazy. We're not seeing that. Yeah. No, I think I'm seeing in the government right now and in public institutions a lot of interest in trying to get the benefits of large language models. And I don't, I mean, this is my first company. This will be my only company. So I don't know if that's normal, but certainly I'm seeing a lot of excitement and interest and a need for speed. And I think that's because the technology is kind of like obviously useful in a way. So we've been lucky to be able to build a tech that lots of organizations feel urgency to deploy.
Yeah. And so like as you're building these incredible sovereign companies in Canada, you're all in industries that require like immense amounts of capital as you scale and grow. I would love to spend a few minutes kind of diving into that. You know, obviously we don't have to get into like confidential details here on stage about your specific trajectory and raises, but in terms of like actually accessing the capital, you know, if you were to need to raise a billion dollars, is that even available here in Canada? Like how do you think about that? And does taking money from outside of Canada impact how you think of yourselves as a sovereign company?
Can I just jump back to the last question? I think we need to Buy Canada. Like I get what Mina is saying, but like every other country in the world that wants to have a defense industry buys their own stuff. Like it's just such an obvious thing to me. The Americans privilege it, the Israelis, the Koreans. The reason the Koreans are selling us stuff is because they bought their own stuff. So we have the capacity to do it, we have the desire, we have now the leadership to do it. I think we do have to buy our stuff and like maybe it's not going to be world class today, but the whole point of buying our stuff is that you generate the incentive systems so that it's world class tomorrow and the day after that. So I think we really do need to buy our own stuff. And that is not a synonym for buying things that are not world class like we have built world class stuff. We can build world class stuff again. On the capital stuff, yeah, we have, I mean the top eight institutional investors here are, it's the same as the Americans, two and a half trillion. Like for the life of me I actually not really sure why we don't force them to buy stuff from Canadians. We should, it's obvious again that we should tell the pension funds to invest in Canadian companies. That's what successful countries do.
So you're supportive of a mandate for...
And like what percentage would you deem like acceptable? Like if we were going to tell pension funds in Canada, hey, you guys need to spend X% of your pools to support Canadian companies. What's the right amount?
I think the math would be the Prime Minister wants to grow the defense industry threefold in 10 years. That's roughly $30 billion. So the pensions should vest like at minimum $30 billion in the Canadian defense industry, which is damn near nothing to them. So like yeah, that's definitely something I pro- I actually haven't had a hard time raising capital in Canada. Maybe it's like a moment in time. But Canadians want to invest in it. Canadians on docks in Halifax, where I'm from, in Victoria, give a damn about their defense industry and care about sovereignty. That is slowly trickling up into Bay Street. But I think the Canadians, you know, the main street really cares about it.
I couldn't agree more. Like I don't know what else to say. I wanted to applaud too. So from our perspective again my story arc was a little bit different. Raise capital from the US, raise, come back to Canada is kind of the story. We're seeing a shift now. There is wildly renewed interest from Canadian investors for the defense sector for sovereign applications. It's something I have never witnessed before. I think you know most of the time I wouldn't have thought about raising capital in Canada. My default and even if you talk to most of your peer groups they tell you if you want to be a capital intensive business raise lots of capital you're going to go to the US but it's changed now we have active dialogues with Canadian based investors and that's not something I would have thought about a year ago.
Yeah. It is actually interesting just to go back to the procurement thing for one second here. That Dominion has like what I would say quite a unique approach to the procurement model and maybe there's other companies doing it as well. I mean you tell me but you know if we were to think about defense procurement in particular and Nick I'm curious on the AI side of things but you know previously what would happen is that the government would sort of like spec out exactly what they wanted in a great amount of detail and companies would bid for that and what Dominion is doing is so innovative because you're actually kind of co-deploying or forward deploying with the Canadian Armed Forces to learn like before any procurement has even happened to learn and iterate in real time. And you know I'm not sure what Cohere is doing on that front, but I would actually just love to kind of have a bit of a conversation around that and hear your perspective on like how is that going and is that like a new model of procurement in this country?
TBD on how it's going. Ask me in a couple of months. On some level, I don't really care. Like I know it's the right thing to do. I'm going to continue to do it. Like the war fighter needs it. The reality is technology is not being developed in a box in Waterloo, Hamilton, wherever the hell it's being developed on the field with the soldier. We went into this 90-day Op Sentinel trial with essentially like a tactical mesh network that went through 200 DOT updates like four fundamental refactors of the back end and it ended up in essentially a common operating picture that now sits somewhere in DND. And I would not have known what they wanted unless I had co-developed them. Unless we had that kind of customer intimacy that I think every other military, if you look what's happening in Ukraine, they're doing software updates, but they're actually doing 50 to 60 hardware updates on like a monthly basis. Like hardware updates. So, you know, we just can't have...
50 to 60 hardware updates?
Hardware updates to like drones. Like if you look at the drone, they have, there's an Amazon marketplace for buying drones. The military can buy drones off their version of Amazon. There are close to a thousand different types of drones. Those drones change every single day. Like it is such a dynamic environment that we're like nowhere close to but we need to get closer to it.
Nick, what are you, like are you co-developing or co-deploying anything or is Cohere's model more like in the traditional procurement?
No. So we use forward deployed engineers as well. And a big part of what we do when we're working with like a big organization is take our model and our agentic framework for using it and we put it securely in their environment, but then it needs to be connected to their data so that it's actually useful, right? So if you just have a large language model and it doesn't have access to like your internal documentation, that's not very useful. So you can imagine like a bank or something, maybe they have like a big HR repository and they have a big database of all the documents about the companies they're tracking or something. And when we deploy into an organization of that size, we'll have some forward deployed engineers come along and write little connectors so that the language model can access whatever data it needs to in that environment in a secure way. And so that's a very standard part of our playbook. That's a little bit like co-developing, right? Like we're figuring out, hey, what connectors are particularly useful for this environment? And then we figure out, hey, like what does the model need to be better at in order to be particularly useful for this administrative task or something? But it's not quite the degree of like, you know, of hardware updates being shipped every day.
And we talked about capital from external sources outside of Canada. You know, if you, and you just did a deal with acquiring a German company, like how do you think about that as you're building a sovereign AI company in Canada? Like how do you think about partnerships globally and capital globally and like does that impact how sovereign you are as a company, do you think?
