This week we are looking at the trends that are affecting all of us in Europe and globally. If you've been buried in your own build and haven't had time to catch up with the news - this is your chance.

I'll be honest: this wasn't a quiet week. Between Mistral's massive infrastructure bet, a very public reckoning over who gets to touch NHS data, defence founders quietly packing their bags for the Bay Area, a high-profile AI compute deal imploding, and China formally declaring tokens the new global commodity - there's a lot to process. What strikes me most isn't any single story. It's what all of them, taken together, are saying. Europe is at a genuine inflection point. We're building more than we ever have. We're also losing more than we should.

These stories aren't just headlines. They're decisions that will shape what's possible for founders here over the next five to ten years. Let's get into it.

See below to continue…

I built Rare Founders by accident. Showed up to events, talked to people, stayed in touch. That's it.

But I was also losing track of half the conversations I was having. Forgot who I met where, never followed up properly, missed the moment when someone actually needed what I had.

SoPhy is the tool I wish I'd had. Contact management, context notes, follow-up reminders - built specifically for people whose deal flow, career, or community runs on relationships.

Waitlist is open. Won't be forever.

Get early access →

Continue…

Mistral AI has raised $830 million in debt to build a new data center near Paris powered by Nvidia chips. This is its first-ever debt raise, and it signals a clear strategic shift: Mistral is no longer just a model company. It's going after the hyperscalers on European soil.

The facility is set to become operational in Q2 2026 and will power the training of the company's AI models while delivering inference services. Mistral is targeting 200 megawatts of total capacity across Europe by 2027, putting it in direct competition with global cloud leaders like Microsoft, Google, and Amazon.

For founders, there are two ways to read this.

One: it's genuinely exciting that a European company is swinging this big.

Two: even with $830 million in debt financing, Mistral is still operating at roughly one percent of what the US hyperscalers are deploying. The scale gap is real and enormous.

What this bet does signal, though, is intent. Mistral is trying to become the sovereign compute layer for European governments and enterprises that don't want to depend on American infrastructure. That's a real market - and if they execute, the opportunity to build on top of it as a European founder could be significant.

This connects directly to why Mistral's infrastructure bet matters - because while Europe is scrambling to build sovereign compute, China is already running a vertically integrated token economy at a scale that should make every founder stop and think.

At the 2026 China Development Forum, China formally adopted "ciyuan" as the official Mandarin term for "token" - declaring it a value anchor for the AI era and a settlement unit linking technological supply with commercial demand. Think of it as OPEC, but measured in inference rather than barrels.

Daily AI token usage across China has surpassed 140 trillion this month - a more than 1,000-fold increase from the 100 billion tokens processed daily at the start of 2024. Chinese AI models run at roughly one-sixth to one-quarter the cost of comparable American systems, with DeepSeek's pricing sitting at approximately $0.028 per million tokens.

The real-world proof of concept: Cursor - a $50 billion company - chose a seven-week-old Chinese open source model over US alternatives because it was eight times cheaper. An Andreessen Horowitz partner estimated that 80% of US startups now use Chinese base models for derivative development.

For founders building AI-powered products, this creates a genuine strategic fork. Chinese models are cheap, increasingly capable, and fast. They also come with documented security risks - US government researchers found Chinese models significantly more susceptible to agent hijacking attacks - which matters the moment you're selling into enterprise or regulated sectors. The token economy is coming either way. How you position around it is a choice you should be making deliberately, not by default.

The Mistral and China stories frame a bigger truth this section makes concrete: control over compute is existential, not optional.

Poolside - the AI startup developing software to automate coding securely enough for government use - is now in talks with cloud providers including Google to revive its Texas data center project after its previous agreement with CoreWeave fell apart. CoreWeave walked after Poolside couldn't stand up its first GPU cluster on time - and the $2 billion Nvidia-led funding round collapsed with it.

The lesson isn't specific to Poolside. It's systemic. As co-founder Eiso Kant put it when the deal was first announced: "To compete at the frontier you need to be vertically integrated from dirt to intelligence." He was right. The problem is, building that vertical stack requires capital, anchor tenants, and execution capacity most startups simply don't have.

This is also Mistral's bet in microcosm - the reason they're raising $830M in debt to build their own data center is precisely because depending on third parties at scale is existential risk. For European founders building AI products, this is a structural lesson about the infrastructure layer beneath everything you're building. Treat it accordingly.

The infrastructure dependency story playing out in AI has a very human parallel in British healthcare right now.

NHS staff - both clinical and otherwise - are reportedly boycotting Palantir's Federated Data Platform, part of the £330 million contract awarded in 2023 to connect NHS systems into a single searchable database to help clear care backlogs. The pushback centres on ethical concerns over Palantir's ties to US defense and intelligence agencies, and deep unease about what happens to the most sensitive health dataset in the world. Government figures are now considering triggering a break clause ahead of the formal 2027 contract review.

Here's the uncomfortable truth this surfaces for founders: when you don't build sovereign capability early, someone else fills the gap - and later, when the politics shift, you pay a very different kind of price. The NHS built a critical dependency on a single external platform with no domestic alternative waiting in the wings.

The question for health tech founders in the UK and Europe isn't whether to build here. It's whether this moment creates enough urgency - and enough political capital - to finally back the companies trying to do so. Building in regulated sectors like healthcare is hard. The sales cycles are long. The trust threshold is high. But right now, the trust gap in the existing incumbent is wide open.

If you've been tracking the UK defence tech space, this story won't surprise you - but it should still sting.

UK defence tech founders are leaving. Not in a dramatic wave. Quietly. Methodically. Relocating to the Bay Area, to Germany, to wherever contracts actually close and capital actually moves. The reason is blunt: while most European governments have ramped up defence spending, Germany stands out as having visible pathways from prototype to procurement for startups that many other European markets still do not provide.

The UK's Strategic Defence Review paid heavy lip service to opening up procurement for startups. The follow-up Defence Industrial Strategy promised a plan. That plan still hasn't been released. In the meantime, Europe lags in spacetech, AI chips and processors - two sectors identified as leading global investment growth in security and resilience tech.

The gap between rhetoric and reality is costing us founders. A seed-stage drone company shouldn't have to choose between a three-to-five-year UK procurement cycle and a four-week raise in California. The fact that this is still the choice being made tells you everything about where the structural problems are. Germany is doing something different. The US is doing something different. The UK is talking about doing something different.

Step back from each of these stories and a single thread connects them all.

The NHS-Palantir story is about trust - what happens when an organisation builds its critical systems on a supplier its own staff fundamentally don't believe in. The defence brain drain is about trust - founders not trusting that the UK government will follow through. The Poolside collapse is about trust - the kind that needs to be earned between infrastructure partners before billion-dollar bets get made. Mistral's infrastructure play is, at its core, about building a trustworthy alternative for governments and enterprises who no longer want to be entirely dependent on American tech. And China's token economy is forcing every founder to ask a harder question: do you build on cheap, capable infrastructure that your enterprise buyers may never accept - or do you pay more for the trust that gets you through the door?

For European founders, this is actually good news - if you can see it clearly. The companies that will win in the next cycle aren't necessarily the ones with the most advanced technology. They're the ones that build systems their users, their buyers, and their government counterparts actually believe in. In healthcare, defence, infrastructure, and AI tooling alike, trust is the moat that's hardest to replicate. It's also, right now, the one most conspicuously absent from the incumbents.

Build with that in mind.

✅ Know someone building in Europe right now? Forward this their way. These trends affect all of us - the more founders who see them, the better.

(👉 Vote now — we’ll share the results in next week’s issue. All votes are anonymous.)

(Last week’s poll results)