Eliza Ward: Hey — you see this one come through this morning?
Brian Reed: The Mistral thing? Yeah, I did.
Eliza Ward: Three billion euros. Series D. Closed September 8th — post-money valuation over €21 billion, which is somewhere around $24 billion. And they're calling it the largest equity round ever for a privately owned European tech company.
Brian Reed: That last part — largest ever — I want to sit with that for a second, because that's not a caveat-free claim. But the number itself... Samsung Electronics is leading this round?
Eliza Ward: Samsung and PSG Equity, joint leads.
Brian Reed: Samsung is a chipmaker. That's — hang on, that's not a typical VC move.
Eliza Ward: No, it's not. And Mistral is three years old. Paris-based, Arthur Mensch is the CEO — so this is, I mean, this is a very young company absorbing a very large number.
Brian Reed: The scale of it is genuinely hard to place. Let me see if I've got this right — they're now valued at €21 billion and nobody's announced a revenue milestone alongside it?
Eliza Ward: No revenue number, no model announcement — nothing. And that's actually the story, I think. When you can't find the business reason for a valuation nearly doubling, you look for the fear reason.
Brian Reed: Say more about that. Because "fear" is doing a lot of work there.
Eliza Ward: It's like — okay, imagine a city with one pharmacy. Everyone agrees a second pharmacy would be good, so investors fund it before it's even built, at a price that assumes it's already open. That's this. Europe wants a second option. So they're pricing in the option, not the pharmacy.
Brian Reed: And the Scaleup Europe Fund, managed by EQT — that's the EU's own money coming in as co-lead, first time in Mistral. That's not a VC making a portfolio bet. That's an institution saying the pharmacy needs to exist regardless of whether it's profitable yet.
Eliza Ward: Right — but here's what the numbers actually look like. Mistral has raised roughly €6 billion total since 2023. OpenAI has raised approximately $186 billion. Anthropic, $161 billion. Those aren't in the same conversation.
Brian Reed: Wait — $186 billion? Just OpenAI alone?
Eliza Ward: Just OpenAI. And there is — I mean, there's one unverified report, globalbankingandfinance.com, saying Mistral has ambitions to hit $1 billion in annual recurring revenue by year-end. But that's not confirmed by any source I've seen. Mensch hasn't said it publicly.
Brian Reed: So the valuation nearly doubled — €11.7 billion a year ago to €21 billion now — and the only revenue figure floating around is unverified and comes from one trade outlet. That's the part I genuinely don't know how to square. Either there are business metrics that haven't been disclosed, or this round is pricing something that doesn't exist yet.
Eliza Ward: And we don't know which. That's the honest answer. We actually don't know what moved the number.
Brian Reed: The sovereignty framing is the take that's already out there, and I'm not sure it survives five seconds of scrutiny.
Eliza Ward: No, it doesn't. a16z, NVIDIA, Salesforce Ventures, BlackRock, Advent — those are the existing investors. That's a U.S. firm, a U.S. chip company, a U.S. software giant, and two more American capital pools. That's the majority of the cap table.
Brian Reed: But the Scaleup Europe Fund is a co-lead. That's EQT, EU-backed. Luxembourg's in too. Does that shift anything?
Eliza Ward: It's a minority layer in a U.S.-funded stack. Like, you can put a European flag on the letterhead, but — wait, actually the hardware situation is the sharper version of this — NVIDIA is both an investor in Mistral and the company supplying the GPUs Mistral trains on. That's one entity on both sides of the ledger.
Brian Reed: So Mistral can't build without chips that NVIDIA decides to sell them.
Eliza Ward: Right. And here's the irony — Mensch himself warned that European companies can't rely on Chinese AI models long-term because of upgrade uncertainty and potential export restrictions. He's describing a dependency risk. But that exact structure, a foreign supplier who could restrict access, that partially describes Mistral's own situation with U.S. capital and U.S. chips. DeepSeek gets named as the threat. The analogy is uncomfortably close.
Brian Reed: He's making the case against his own supply chain.
Eliza Ward: More or less. The ASML integration gets cited as proof of real enterprise traction — and that's a legitimate partnership, Mistral is actually running AI in ASML's manufacturing processes. But that doesn't resolve the structural point. The sovereignty pitch is doing heavy lifting for a company that couldn't train a model without Nvidia hardware it doesn't control.
Brian Reed: Which makes the Samsung lead role the thing I actually want to understand better — because if the sovereignty framing is shaky, what Samsung's doing here, and what Mensch means by doubling compute over five years, that's a different story than the one being told this morning.
Eliza Ward: Samsung is the thing. Because it's not just capital — Mensch said explicitly they're going to work with Samsung on AI integration in manufacturing. Same structure as the ASML partnership. So Samsung isn't a passive LP. They're a customer. And a chip company. And a lead investor. All at once.
Brian Reed: Which means the return Samsung is looking for isn't a multiple on equity. It's — hang on, it's something closer to supply chain positioning. They're exposed to U.S.-China decoupling on both sides. They sell into China, they fab in South Korea, they buy U.S. components. A 'neutral' European AI seat is worth something that doesn't show up in an IRR calculation.
Eliza Ward: Right — but is that confirmed or are we inferring it?
Brian Reed: The dual role is confirmed. The geopolitical motive — that's inference. But it's hard to explain otherwise. A chipmaker doesn't lead a €3 billion round into a company running on Nvidia GPUs unless something beyond the cap table is moving.
Eliza Ward: And then Mensch says Mistral will roughly double its own compute capacity over five years. The New York Times frames that as a strategy shift — toward data centers, enterprise infrastructure. But Mensch is also saying they're not stepping back from frontier model training. Those are — I mean, those aren't the same sentence.
Brian Reed: So which one is it? Because doubling compute over five years, that's not the pace of someone trying to catch OpenAI. That's building a platform someone else will use.
Eliza Ward: We don't know. And I want to be clear — that contradiction is genuinely unresolved. The sourcing doesn't close it.
Brian Reed: Picture a procurement officer at a German manufacturing firm — not in three years, right now, this week — deciding whether to run on Microsoft Azure with OpenAI's API or wait for Mistral's European data centers that may or may not exist yet. That's the actual decision this €21 billion is trying to make possible. The question is whether Mistral can deliver before the procurement officer just defaults to Azure.
Eliza Ward: So what to watch: data center location announcements, any enterprise contract they name publicly, or — actually the sharpest signal — any ARR figure Mensch puts on record. Because right now the valuation is a promise. A number would make it a business.
Brian Reed: The valuation is €21 billion and the thing that would actually anchor it, a revenue number, a named customer beyond ASML and Samsung, anything on record from Mensch, none of it is there. I mean, we're not inferring the gap. The gap is just... present.
Eliza Ward: Which means the open question isn't really about Mistral's technology at all. It's whether Western capital — and let's be honest, it's mostly U.S. capital with a European flag on it — keeps funding this indefinitely if the gap with OpenAI doesn't close.
Brian Reed: That's the one I can't answer. Because if this is a geopolitical hedge — Samsung's supply chain positioning, the Scaleup Europe Fund, Luxembourg in the mix — the return isn't a multiple. The bet pays off just by existing. Which means profitability might never be the test.
Eliza Ward: Right — but that's me inferring motive. What's confirmed is the €21 billion valuation with no disclosed revenue to hold it up. Everything else is us reading the investor list.
Brian Reed: So we watch for an ARR figure from Mensch, or we don't get one, and that itself is an answer.
