Europe's artificial intelligence sector is being held back by a culture of risk aversion and a financing model that favours established industries over frontier technology, according to the CEO of one of the continent's most prominent AI startups. Robin Rombach, who leads German AI image generation company Black Forest Labs, told AFP in Freiburg im Breisgau that Europe needs to shift its mindset to one of optimism and opportunity, not risk and fear.
The warning comes as Europe continues to lag behind the United States and China in cutting-edge AI development, with policymakers and firms scrambling to find ways to close the gap. Black Forest Labs, valued at around $3.25 billion, recently drew headlines when it announced US director Martin Scorsese as an advisor — prompting an angry response from some in Hollywood concerned about AI taking production jobs.
The Capital Problem
Rombach, 33, said that when Black Forest Labs was founded in 2024, there was no real startup ecosystem in Europe, particularly for frontier deep tech. He said that if you want to build a frontier model in a very competitive space, you need to do this quickly, and that certain ingredients are needed for success, such as a lot of capital and computing power. The company has a second headquarters in San Francisco and secured early backing from Silicon Valley venture capital firm Andreessen Horowitz.
Zach Meyers of CERRE, a Brussels-based think tank, echoed some of Rombach's concerns, saying the biggest barrier to AI growth and innovation in Europe was access to capital. He said most European companies rely on financing from banks, which are traditionally more cautious than venture capital investors about backing businesses that may take years to become profitable. Meyers said that is not how technology markets work, particularly with a really nascent technology like AI.
The financing gap has real consequences for Europe's ability to compete. Black Forest Labs, based in Freiburg, has emerged as a prominent developer of AI image-generation models in its second year of operation. The startup has just over 100 staff. Rombach helped develop latent diffusion, a breakthrough that underpins many of today's AI image-generation systems. He and a group of fellow computer scientists founded the AI lab in 2024.
Open Models vs. Corporate Control
Rombach said the debate has shifted to safety in recent weeks after security lapses sparked calls from some industry leaders for development of the technology to be slowed down. But he argued his company's focus on open-weight AI models, some of which can be downloaded by researchers and developers, increases transparency and increases safety. He called for an open approach to AI development, warning it would be fundamentally wrong for such a crucial technology to be controlled by a few companies, rendering oversight more difficult.
The company's FLUX models generate still and moving images from text, and the company is now also expanding into physical AI. Its latest model has been tested at carmaker Audi as part of a project aimed at enabling robots to perform production-line tasks. Rombach said the main thing that can wipe out humanity is stupidity.
Why This Matters:
Europe's struggle to finance frontier AI companies reveals a structural weakness in its innovation ecosystem — one that threatens the continent's ability to shape a technology that will define the next decade. If European startups can't access the capital and computing power needed to compete, the risk isn't just economic. It's democratic. AI development concentrated in a handful of US and Chinese firms means Europe loses influence over safety standards, transparency requirements, and the ethical frameworks that should govern these systems. The debate over open versus closed models isn't just technical — it's about whether AI remains a public good subject to oversight, or becomes a proprietary technology controlled by corporate interests beyond democratic accountability. Europe's regulatory ambitions mean little if it has no domestic industry with a seat at the table.