Inside Mistral’s monster fundraise and Palantir-like repositioning

0:01Hello and welcome to the Sifted podcast. I'm your host Freya Pratty, Associate Editor here at Sifted. Joining me today is Senior Reporter Daphné Leprince-Ringuet. Where are you joining us from today Daphné? From Paris. Nice and I am in South London as I usually am for the pod. Today we're going to be discussing a deal that has got the ecosystem really chattering. It's kind of the big one at the moment that we're looking at. This is French AI darling Mastral's latest funding round. Mastral is reported to be raising a€3 billion Series D round at a valuation of roughly€20 billion. So it's a big one.

0:40Last week, Daphne, you and I broke the news that the EU's new scale-up fund, this is a€5 billion pot managed by EQT, is in talks to lead or co-lead this deal. There's a lot of discussion about which company that fund will pick first. So this is exciting from both kind of a Mistral side and from the scale up fund side. First off, just give us a bit of background on Mistral and what we know about the fundraise so far. Sure. So Mistral is a French company, as you mentioned, that launched three years ago in 2023. The pitch at the time was to provide a European alternative to the huge AI labs that are based in the US, OpenAI in particular.

1:19so initially the company focused on making what is known today as frontier AI models and since then it has evolved to also include AI infrastructure and compute so essentially Mistral is actually now also building data centers for AI. A lot of European hopes to compete in AI are riding on Mistral. They've been very successful in many ways they've really established themselves as kind of this symbol of European tech hopes. So they've raised around$3.5 billion of both debt and equity in just three years. They're now, as you said, presumably on track to raise another€3 billion at a reported€20 billion valuation.

1:59So that's really remarkable for a three-year-old European company. It's definitely not the kind of story we hear every day. What else about Visrael? They've previously said they're targeting €1 billion in revenue in 2026. They have a thousand employees. They have presence in the US, in Europe, in Asia. So again, all of these are very kind of positive indicators of how well Ms. Troll is doing. There are increasingly kind of doubts about whether they are able to compete with big US labs, such as OpenAI, as I mentioned, but also now Anthropic, which is becoming kind of the main competitor in that race.

2:39um we've also got obviously that the the ai labs within big u.s uh tech giants like meta and google we've also got increasingly uh challenges coming from china which are building open source models that um increasingly european organizations are keen to deploy because they are open source and not proprietary um so yeah there are a lot of questions still about where mr is headed and how successfully it's going to accomplish its mission. But it is very impressive what they've done in three years for a European company. Yeah, the aspect you just brought up about sovereignty and Mistral being very central to having a European model is particularly relevant at the moment, which is handy for their fundraise.

3:23Talk us through what's happened around that debate in the last couple of weeks. Yeah, definitely. So Mistral have always pitched themselves as a very sovereign player and was a huge part of the pitch back in 2023 have a European as I said European alternative to US monopoly and dominance in the field I think over the past few years this thesis has only strengthened due to various different factors and most recently due to increasingly tense geopolitical relations with the US which have meant that Europe feels like it can no longer rely on the US like it did before and some of these relationships might be slightly up in the air and so it kind of has prompted this this realization in Europe that we rely on the US for a lot of our critical technologies AI increasingly being one of these critical technologies you mentioned recent weeks I'm assuming you're referring to the recent order that came from the US government for Anthropic to ban some of its top performing models for non-US nationals so this was the first example of what's known as a digital kill switch by which the US could essentially restrict or even completely suspend access to key technologies for European countries.

4:46We discussed this in a previous pod by the way and I think it was a real wake-up call for Europe that there is a need to have a level of autonomy and resilience in the face of these increasing geopolitical threats and to technology being a leverage in geopolitical relations really what happened with anthropic just over a month ago was a real proof in favor of mistral's theory and interestingly in france just last week some parliamentarians published a report on french digital sovereignty with a number of recommendations for the government And one of these recommendations, very interestingly, was that the government should take a stake in Mistral.

5:31So there were two companies cited in the report. One was Mistral, one was Chaps Visions, which is a big data company, essentially a European version of Palantir. Both were considered to be kind of the two most critical companies for European sovereignty, for French technological sovereignty. One of the recommendations was that the French state should be able to take a 5 % stake in Mistral. So I think this really kind of is another example of how this sovereignty debate is now completely taken over the discussion when it comes to tech in Europe at a very political level. So yes, that fundraise really has to be put into that context and perspective.

6:12And that leads us very nicely into the news last week that we wrote about that the EU scale-up fund is in talks. we have to stress in talks the deal is still up in the air still being finalized um in talks to lead or co-lead this fund and that fund has a very clear mandate to back sovereign solutions tell us what we learned last week yep so last week we learned um that eqt which is a swedish investment firm is in talks to lead or co-lead mr else fundraise um via the scale up europe fund as you mentioned, which is a 5 billion euro pot of money that is dedicated to addressing Europe's lack of late stage capital.

6:53And it definitely, as you said, intends to back scale ups in strategic fields for sovereignty. So mostly deep tech, fields like AI, quantum computing, chips, etc. And so just to clarify how this kind of mechanism works, EQT, the investment firm, won a competition to manage this fund, the Scaled Up Europe Fund, which is an investment vehicle that was proposed by the European Union. And so presumably if the talks go through, they are not finalized yet, things can still change. But if they do go through, this would be Scaled Up Europe Fund's first investment, an investment that has been a waste with a lot of anticipation.

7:31What else do we know about this? So Scaled Up Europe Fund, we know plans to lead or co-lead investments. They plan to write checks between 100 and 500 million euros. So we understand the talks with Mistral are in mind with this thesis. So State of Europe Fund would be leading or co-leading the Series D. And I guess that's pretty much it so far. It's a very secretive fundraise. It's been difficult to have any access to information, really. Mistral are known to be very private and confidential about these things. So that's about as much as we know so far. that uh deal from the scale up fund would be really interesting because we wrote a couple of weeks ago about how the uk wants to get involved in the scale up fund which is a european union vehicle but they're trying to to get involved and we had heard and reported that france were the ones that were trying to block them and i think the fear the suspicion is there's a fear that the money would seep out of the eu into the uk because we have here a lot of uh companies at the right level to receive those growth funds and the talks with Mastral kind of show perhaps what they worry the money will move away from and into the UK if they've got this sovereign model play that they think should be bolstered by these funds.

8:44I thought that made that reporting make a bit more sense on why France would be so against money potentially coming out of it into the UK. There's also been a lot of other reaction to the reporting on EQT and the scale-up fund potentially backing Mastral. What have people been saying? Yeah, there's been mixed perception, I'd say, on social media. Some people are saying it's great news. Others are saying it's a big mistake. One, and I'm quoting, says this makes no sense at all. Another one says this is a mistake. Mistral is not the right horse. And then it goes all the way to someone who said we need to give the EU's best AI, which is Mistral, billions and a nuclear plant.

9:23So, you know, there's a whole spectrum of reactions. I guess the interesting question here is why would people be against that move from EQT this is speculation on my part but I think there's different reasons for that the first one is probably that people don't believe in Mistral anymore as I said you know the company is competing against huge competitors in the US in some ways they are lagging behind we can maybe come back to that later and so I think there is a whole category of people who don't think Mistral is where Europe should be placing its best for technological sovereignty and perhaps they think that the State of Europe Fund should be placing bets instead on where Europe has a chance so not Frontier AI Labs kind of maybe lost that fight but we do have very good companies in applied AI we have very good defence tech companies we have very good quantum computing companies ship companies so um i think perhaps there's a sense that this this money should have gone to back these um and i also think it it perhaps reflects um a frustration with a kind of mistral centricism um in the sense that mistral gets a lot of attention it gets a lot of funding it always has um so there's maybe that feeling that once more mistral is capturing everything to the detriment of other contenders for the fund and that perhaps Europe should be focusing on also bringing about other champions to fight for that sovereignty in tech.

11:04Yeah, I guess this idea that we perhaps have missed out or are in the process of missing out on the first wave of models, but we have these companies like AMI or the Neolabs in London building the next wave of models. Maybe, I mean, the London debate is obviously one that's up in the air with regards to scale up fund, but perhaps there is like the second wave that we could win at is something I've seen discussed. Can you go into a bit more detail on the point you made about them lagging behind other model makers? What does that mean for how the business looks at the moment? So, I mean, obviously you look at numbers.

11:36Mr. L is very obviously a smaller company than Anthropic or an OpenAI. Anthropic recently secured$65 billion at a nearly$1 trillion valuation. That was a month after OpenAI announced a$122 billion raise at an$852 billion valuation. Lots of numbers, but the point is, Mr. L's numbers are big. They're big for Europe, but they're still very much behind the numbers we're seeing in the US. But it's not just that. I think there's also increasingly doubts about the quality and the performance of the models that Mistral are releasing. They're quite systematically behind the competition, not just in the US, but also Chinese models, which are increasingly very high performance.

12:21And so I think this has led to a lot of people saying that Mistral has kind of let go of this idea of competing as a frontier AI lab. And it's said Mistral is focusing on bringing models into enterprises. services um this can be seen in that Mr. L has recruited a lot of what we know now as forward deployed engineers FDEs this is a role that has been pioneered by Palantir where you have specialized engineers so FDEs that work hand in hand with customers to deploy the technology as effectively as possible and ensure gains for the customers um so it looks like Mr. L is increasingly focused on those go-to-market functions and specifically on enterprise go-to-markets and so So some say that the company is increasingly evolving to be more similar to a company like Palantir.

13:09Some even describing Kimisro increasing as a consultancy for enterprises and to help enterprises deploy AI rather than a frontier AI lab like OpenAI and Anthropik. So this obviously brings in questions as well around the company's valuation, how much money it should be raising. And I think that perhaps is also behind some of the criticisms for why EQT is in talks to pick Mistral for its first investment via the Scade Up Europe fund. Do you have an idea from sources you're talking to about what the move to being more like a Palantir rival would mean for the pricing of the round evaluation? Which direction it would send it in in comparison to being a frontier lab?

13:53um i think the only way really is to compare to the multiples we're seeing at palantir and to be honest the the multiples we're seeing for mistral are not a million miles away um but i think it does bring in more widely a question of what is the right business model for frontier ai lab have Anthropic and OpenAI figured it out? If not, is it just a question of, you know, these companies are building these models and they're figuring out what's the best way to actually build a company on top of it because being an AI lab is not being a profit-making company. So I do tend to be quite forgiving that Mistral have gone that way.

14:36Perhaps it's the way that works. I think it may be too early to tell at this point. but certainly if they fill their promise to hit 1 billion euros of revenues in 2026, that would be a very important proof point that it's working. So yeah, it's all still up in the earth at the moment, I'd say. And you've said that people within Mastral are tight-lipped about the deal, which I guess we would expect them to be. But what else do we know? What have you figured out? So I think we have both figured out. We're hearing that EQT, if the deal goes through, would be co-leading so that there would be another lead, which also would make sense given the size of the deal.

15:17You can expect that it's going to require very deep pockets. So a co-lead would make sense here. The question is, therefore, who will be that co-lead? There's two questions in that I think. There's where that co-lead will be based and what kind of investor it will look like. in terms of where so as we mentioned Mistral is is seen as a very strategic company for Europe so it has to be relatively careful whose money it takes or rather where the money comes from I remember when when Mistral closed its series b that was in 2024 they closed 600 million euros it was then by general catalyst included a lot of u.s investors like Lightspeed Andreessen Horowitz etc and it caused a bit of a stir at the time and to be honest I can't really count how many times I've heard since that Mistral is essentially controlled by USECs or that you know USEC's interests are determining the company's strategy etc.

16:13So for the Series C which closed a year ago or less than a year ago there was a lot of speculation around who would need because of how touchy the topic had become and I'd say Mistral like massively pulled it off in a way they got Dutch semiconductor equipment manufacturer ASML to lead the round with a 1.3 billion euro investment. So this is really important because there's not that many investors in Europe who are capable and willing to inject that much cash in a startup. So going for a corporate investor like ASML was really smart. It takes us to the Series D. So, you know, are there many other investors in Europe who can follow on from what ASML did?

16:55Will ASML put more money in? Will it be another corporate? or will it be more of an institutional investor? I've been hearing SAP's name coming up in entirely speculative ways, but SAP is a customer of Mistral. They recently acquired a startup called Prionabs, which built foundational models as well for€1 billion. So kind of like signs that the company has the financial firepower, the interest and appetite for new technologies. But there are not a million options. in Europe. The other option would be that it comes from another geography. I don't think it could come from the US based on Mistral's existing cap table and it's already quite biased in favor of the US.

17:40So, you know, obviously the other region to look to would be Asia or the Middle East. The Middle East is kind of currently complicated with the current conflict. So perhaps Asia makes more sense. Mistral also has an office in Singapore. It has a dedicated team for Asia Pacific. One of its customers is the Singapore Ministry of Defense. Some of its existing investors include Samsung, the South Korean electronics giant. There's a Japanese bank as well, Mitsubishi, that backed the loan that Mistral secured earlier this year to build out data centers. So it has very close size with Asia. It certainly made it clear that it's a region it's interested in and a region where it wants to build up its activity.

18:24So it would make a lot of sense to see something coming out of there. And then to the second point I made, what kind of ancestor will it be? I guess at this level, we're looking for very deep pockets. So very easy to think of sovereign wealth funds, funds like Temasek in Singapore. If we're keeping the Middle East as an option, there's Mubadana, there's MGX. And there's also very good big corporates as well that have deep pockets. with ASML and Mistral went further up the value chain with a partner that makes equipment to make AI chips. There's many options that we could think of of where else it could target in that value chain.

19:06Could it be a chip maker? Could it be something to do with infrastructure and compute? There are many bets that are still on. I am still making up my mind. I have a decent idea of who I think it could be, but nothing that we can share at this point, I think. It will be really interesting if it comes from outside of Europe in terms of they are so angled towards the sovereignty play at the moment. But I think if it does come from some of those names you mentioned, particularly in Asia, it fits that kind of thesis that I hear increasingly around coalition of a middle powers group. And co-led with EQT as well, there's that balance, I think, that would be achieved if that's how it turns out.

19:48Well, we'll keep tracking this deal very closely. That's all we've got time for for today. Daphne, what else are you working on? What can we expect from you this week? Well, you and I are working on a piece that's actually very relevant to what we just mentioned. We're looking at the talent base at Mistral and how it's evolved throughout the years from a more research-focused talent base to a more commercially-focused talent base in that switch from Frontier AI Lab to more go-to-market, enterprise -go-to-market focused company in the Palantir style. that's in the immediate I'm also working on another interesting company that is emerging in Europe which is the Italian AI model maker Domin so Domin has also made big claims as to how much money it's raised and what its ambitions are for Europe it's also building AI infrastructure in Europe so yeah looking a bit more closely at the company, where it's at at the moment, what are the next milestones for it, etc.

20:53So welcoming any views on that, feel free to get in touch at daphne.sifted.eu if you have any information or interesting perspectives on Domain. That'll be great. And I'm working on an analysis of what the latest model release is coming out of China, particularly moonshot status won't mean for the kind of AI bubble scares or what that will mean for companies in Europe. Daphne, thank you so much for joining the pod. Thank you for having me. listeners will drop links to all reporting discussed today in the show notes and if you want to keep following the stride what happens with this deal you should sign up to our daily newsletter and our ai and deep tech newsletter which is written by daphne we'll put links to those in the show notes as well as always please rate review and share the podcast and this episode was expertly produced by maya durampel hornby