Global AI App Race Heats Up: Why Europe Still Has a Shot at the Top Layer
The global competition in AI might seem dominated by the U.S. at first glance — and for foundation models, it largely is. But according to a recent report by Accel, the battleground is shifting. While the U.S. maintains a sizeable lead in large model development, the story is different when it comes to the application layer — and that gives regions like Europe and Israel a real shot at playing a bigger role. ([TechCrunch][1])
Europe and Israel are narrowing the gap
Accel’s 2025 GlobalScape report points out that European and Israeli cloud & AI application startups have raised about 66 cents for every U.S. dollar in 2025 — a strong improvement from a decade ago when Europe raised only one-tenth of U.S. levels. ([TechCrunch][1]) Accel partner Philippe Botteri attributes this to the emergence of a software ecosystem in Europe that finally understands how to build scalable companies and has been operating as a “flywheel” for about 10 years. ([TechCrunch][1])
Founders across sectors — from legal and healthcare to manufacturing and marketing — are now combining deep technical talent with domain expertise. That means Europe is no longer just a talent or staffing pool for Big Tech, but increasingly a place where full-fledged AI-native application companies are being built. ([TechCrunch][1])
Apps are moving faster than ever
One of the surprising take-aways: some AI-native applications are reaching US$100 million in annual recurring revenue (ARR) in just a few years — a pace much faster than traditional enterprise software companies. ([TechCrunch][1])
What’s driving this is product-centric design, strong go-to-market execution, and high revenue efficiency (revenue per employee) — especially in companies that build with AI at their core. Botteri says this efficiency “is the highest we’ve ever seen for software companies.” ([TechCrunch][1])
Models vs. Applications vs. Data — models aren’t the only game
Even though models (foundation or large language models) dominate headlines, the report argues that focusing only on “model vs. app” is a false dichotomy. According to Grove Ventures managing partner Lotan Levkowitz, data is undervalued in the current wave. He points out that companies building proprietary data flywheels have strong long-term potential. ([TechCrunch][1])
In other words:
- Models need data to train.
- Applications need models and data to deliver value.
- Having unique data and being able to turn it into application-level experiences is a differentiator.
- Meanwhile, building a globally competitive foundation-model company remains a much tougher climb — especially in Europe, which Accel currently views as “not a very target-rich environment” for the big model race. ([TechCrunch][1])
The big takeaway for startups and investors
- For startups: If you’re building AI apps (rather than just models), focus on differentiated product UX, a fast go-to-market, and ideally domain-specific data or workflows. That’s where there appears to be momentum.
- For investors: While the model-infrastructure wave continues to draw attention (compute, chips, etc.), applications and data-driven workflows might offer better return dynamics in the near term — especially in regions like Europe and Israel where competition for apps is intensifying.
- For regions outside the U.S.: This is a moment to lean into your strengths — domain expertise, localised workflows, regulatory understanding, and unique data — to compete in the app layer rather than simply trying to catch up in the model infrastructure race.
Glossary
- Application Layer: In the context of AI, the “application layer” refers to software products or services that sit on top of AI models and infrastructure, providing end-user functionality (e.g., AI-generated video tools, legal automation, marketing assistants).
- Foundation Models / Large Models: Very large machine learning models (often deep neural networks) trained on massive datasets to serve as general-purpose engines (e.g., LLMs). They form a base upon which applications can be built.
- Annual Recurring Revenue (ARR): A metric used in subscription-based business models to measure the predictable yearly revenue from customers.
- Data Flywheel: A virtuous cycle where proprietary data improves the product, the improved product attracts more users, which generates more data, which further improves the product — creating a growing competitive advantage.
Source: TechCrunch — “The global race for the AI app layer is still on”
| [1]: https://techcrunch.com/2025/11/11/the-global-race-for-the-ai-app-layer-is-still-on/ “The global race for the AI app layer is still on | TechCrunch” |
