Europe's AI Startups Just Crossed $140B: The Full Valuation Map

Europe's AI Startups Just Crossed $140B

27 companies. 7 countries. A defence company on top.

Three years ago, most of the companies on this list did not exist. The composition is worth studying, because it looks nothing like the American one.

Tier 1: The Decacorn Race ($10B+)

1. Helsing ($18B) defence AI
2. Nscale ($14.6B) AI data centres
3. Mistral AI ($14B) foundation models
4. Celonis ($13B) process intelligence
5. Poolside ($12B) AI for software engineering
6. ElevenLabs ($11B) voice AI

Six companies above $10B, and exactly one of them is a foundation model lab.

Tier 2: The Challengers ($1.4B to $9B)

Wayve ($8.6B) in autonomous driving. Lovable ($6.6B) in AI app building. Legora ($5.5B) in legal. n8n ($5.2B) in workflow automation. Synthesia ($4B) in video. Black Forest Labs ($3.3B) in image models. Aleph Alpha ($3B) in sovereign LLMs. Parloa ($3B) in customer experience agents. Quantexa ($2.6B) in decision intelligence. PhysicsX ($2.4B) in engineering. DeepL ($2B) in translation. Framer ($2B) in website building. Neko Health ($1.8B) in health scans. Harmattan AI ($1.4B) in defence drones.

Tier 3: The $1B Club

Tines ($1.1B) in security workflows. Owkin ($1B) in medical research. Pigment ($1B) in business planning. DataSnipper ($1B) in audit automation. Cast AI ($1B) in cloud and AI workloads. Stability AI ($1B) in generative media. Tractable ($1B) in computer vision for insurance.

Three Things That Stand Out

Defence AI took the crown

Helsing jumped from $14B to $18B in under a year, and France minted its first defence unicorn in January with Harmattan AI.

This is a genuinely European outcome. Defence procurement is national, relationships are decades long, and sovereignty requirements mean a European military will not buy an American autonomy stack for critical systems. That regulatory moat is worth more than any model advantage.

The UK has the most companies, Germany has the most value

Eight UK companies, the largest count of any country. Germany leads on total value at over $47B, carried by Helsing and Celonis. France owns the top end with two companies at $12B or above.

That spread matters if you sell into Europe. There is no single hub. A go-to-market motion that treats Europe as one market will miss most of it.

The application layer is compounding fastest

n8n doubled to $5.2B in eight months. Parloa tripled. Legora tripled.

None of those are model companies. They are workflow companies that rent models and own a specific job: automation orchestration, customer conversations, legal work.

Why Europe Skewed Applied

The US list is dominated by two foundation model labs holding most of the value. The European list has one model lab in the top six.

Part of that is capital. Training frontier models requires funding at a scale European venture markets have historically not provided in single rounds.

But part of it is a genuine strategic difference. European companies went where domain knowledge and regulation create defensibility: process mining, legal, audit, insurance, healthcare, defence. Those are markets where being close to the customer and understanding the compliance regime beats having a marginally better model.

That is the same pattern visible in the US application layer, just more pronounced because Europe barely competed for the model layer at all.

What This Means If You Sell Into Europe

Three practical implications.

Your buyer may be a European vendor now. For workflow automation, voice, translation, or document intelligence, the incumbent in a European deal is increasingly a European company with local data residency. That changes the competitive conversation.

Data residency is a real requirement, not a checkbox. Aleph Alpha exists specifically because some European buyers will not run inference outside their jurisdiction. If your product cannot answer that question, it will lose deals it should win.

The fastest-growing segment is the one you can join. Application-layer companies with domain depth are compounding faster than anything else on the list. That is a market you can enter without a data centre.

A Caveat On The Numbers

These are latest closed priced rounds, which means several are stale by a funding cycle and all of them reflect what one group of investors agreed at one moment. European rounds are also disclosed less consistently than American ones.

Read the tiers and the country distribution. Do not read the decimals.

Frequently Asked Questions

Why is a defence company the most valuable AI startup in Europe?

Defence procurement is national, contracts run for years, and sovereignty requirements mean a European military will not buy an American autonomy stack for critical systems. That regulatory and political moat is worth more than any model advantage, and it is not available to an American competitor at any price.

Why does Europe have so few foundation model companies?

Partly capital, since training frontier models needs funding at a scale European venture rounds have rarely reached. Partly strategy: European teams went where domain knowledge and regulation create defensibility, which is process mining, legal, audit, insurance, healthcare, and defence.

Which country leads European AI?

It depends on the measure. The UK has the most companies at eight. Germany has the highest combined value at over $47B. France owns the top end with two companies at $12B or above. There is no single European hub, which matters if you sell into the region.

What does data residency actually change for a buyer?

For regulated European buyers, quite a lot. Aleph Alpha exists specifically because some organisations will not run inference outside their jurisdiction. If your product cannot answer where processing happens, it will lose deals it should otherwise win.

Which of these are relevant to a go-to-market team?

n8n for workflow orchestration, which doubled to $5.2B in eight months. ElevenLabs for voice. Parloa for customer conversation agents. Synthesia for video. Lovable and Framer if you build landing pages at volume.

We build AI-native pipeline systems and you pay per qualified meeting, not a retainer. No paying for activity. You only pay when we book you a qualified sales meeting. See if you qualify.

Valuations reflect latest closed priced rounds as publicly reported in mid-2026. Not investment advice.