Good morning, everybody, and thank you so much for having us on stage. So since we only have 20 minutes, I would like to dive right into it. And so, Joelle, Cohere has very recently announced that it has signed an agreement to become the first transatlantic sovereign AI solution. What does that exactly mean?
Well, thanks for asking and lovely to be here with you this morning. This is really in reference to our partnership with Aleph Alpha, a German company champion who has been building models and delivering enterprise solutions for AI, which really matches what we've been building at Cohere, really focused on building AI solutions completely for government and enterprise. And I would say right now there's a tremendous appetite for sovereign AI solutions. And so we're joining forces. Our teams are starting to figure out how we're going to address the need for increasingly large models and the need to find a way to make them viable for enterprises and governments, which have different requirements, whether it's with respect to capabilities or to safety or to compliance in terms of various standards. And so super exciting new era starting for us.
Thank you. And let's find some common ground at the start of this panel first. So since everybody likes to define sovereign AI a little bit differently, maybe one or two persons of you can also tell us what they understand under sovereignty.
Sovereign AI really started with the angle of geopolitics, with countries wanting to control their own supply chain, control their own destiny, and have their own AI models run in their countries and their own data centers. What Uniphore is focused on is, how does an enterprise define sovereign AI? And that's, like, if I distill the message behind sovereign AI, it's about optionality, it's about flexibility. But more importantly now, with a lot of urgency, it's become about control. And especially with the developments of the last week or so, every enterprise, every CIO that we talk to, wants to have control over their AI stack.
Do you want to add anything, Joelle?
So last year I wrote the AI strategy for the Netherlands and it was really the reason why to write it was because we saw the amount of impact that AI was going to have but also the dependence that we as European countries have on these American companies. And we saw earlier in the year that just by the flick of a switch we lost access to one of the best models. And if you believe that AI is going to be a truly competitive advantage for your companies and I personally believe that because with my own company we see the impact that we have for our clients then losing access to the very best models also means that you cannot compete anymore with the companies that do have that access and so we started thinking like okay how do we make sure that even if that AI is not being developed here in Europe, how do we keep access to it? Part of it is the compute. We need that compute here in Europe to actually be able to run either open source models or have those American models be run here and at least then have some control over it. But on the other side, there currently is a lot of dependence from us on them, but very little the other way around. Like we often talk about ASML, but that's just one card that we can play. And there was actually in the news yesterday that ASML doesn't have any customers in Europe. So we can say like, oh, we stopped delivering ASML machines to the US, but then ASML doesn't exist anymore. So it's not really a card that we can play. And so we need to have more technologies here that are being developed and that have a global impact so that at least we can play poker.
And just this morning, you have announced something that is trying to do exactly that.
Yeah, so as part of the AI strategy that we wrote, we identified why is it that the Netherlands doesn't have that many great companies. We are scientifically a powerhouse. Like we produce more papers per head of the population than almost any other country in the world. Like we're in the top three, but we don't convert that to companies. And what we identified is that we are very risk averse. So the research that we fund is actually the relatively safe research that will result into papers, but it's not the high risk research that leads to breakthroughs. Deep Tech Ventures. And that institute was announced last week by Ara King. It's a copy of Sprint in Germany, ARIA in the UK or DARPA in the US. And what we announced this morning is our first program. And our first program is that we're putting 40 million towards the early stages of chip design and how we can utilize AI to design radically new architectures. And so we'll be selecting seven teams who get two and a half million each, non-diluted funding. And after eight months, we look at like, okay, who has the most momentum? We pick three and they get another seven million. So it's almost 10 million per team. And I hope, I hope that that triggers a whole bunch of people to finally go after those risky bets that have always been in the back of their mind, but they have been too afraid to take on or not able to raise funding for or whatever.
And what do you and Joelle, what do you answer customers who are worried about dependencies like that? So can you even be truly independent of, for example, the chip supply from America as a company?
I can start. To some degree, the goal isn't independence per se. The goal is actually to have choice and to be able to build a resilient...
But do you have a choice right now? Because there are only like NVIDIA and maybe AMD right now.
Yeah, so on the chip level right now, there's not a lot of choice. That's the reality and really thrilled to see the initiative that's happening here. It was really interesting listening to you because there's so much commonalities in the gaps you're identifying here in the Netherlands versus what we're seeing in Canada. Amazing talent, amazing research, investments, institutions, but converting that has been difficult and really capital and like an appetite for risk are probably the biggest missing ingredients. And so I think, you know, we need more of these kinds of initiatives that give us optionality and choice. But we also need to figure out where can we collaborate on these. You know, there's going to be a strong chip design program here in the Netherlands, that there's going to be a strong model building program across Canada and Germany. I think there's opportunities to diversify these initiatives.
And Umesh, how do you deal with that?
