European startups are leading the way in the Generative AI revolution. Unlike past technological shifts dominated by the US, startups on the continent are pioneering market-leading foundation models and B2B applications that will underpin next-generation GenAI tech stacks. We believe that the trifecta of good regulation from the outset, superior technology talent and outstanding fundamental research will mean Europe will emerge as the winner over time.
European governments’ focus on privacy and GDPR compliance is likely to establish an important standard for the world in AI regulation. Most European companies and governments are thinking critically about how to make AI “ethical”, and over time these standards will structurally affect how applications are built in the space. In the long run, European Generative AI startups will benefit, with such standards being built into offerings from day one, providing a competitive advantage.
Furthermore, deep technical and research talent in Europe will contribute to advancement on the AI front and there will be European winners that go on to compete and win globally. For example, there are several market leading foundation models based out of Europe, including Stability.ai, an open source LLM that was originally ideated at LMU Munich. Aleph Alpha, are building a large LLM in Germany as well, and are a strong contender in the space. Hugging Face, the world’s largest machine learning model hub, is a Paris-based business. There is, in fact, a potential geopolitical advantage to having local European LLMs in order to ensure reliability at all times, coupled with the reality that Europe has a better grid and infrastructure than the US and as such, can support a large amount of GPUs.
We have identified key areas of opportunity in the GenAI tech stack, and a handful of companies we believe are on stellar trajectories.
These key areas are:
B2B Applications: Innovative startups are embedding GenAI into products in specific verticals, tapping into large markets with quick proof of concept. Examples include our portfolio companies Omi and Swimm, plus Supernormal, Raycast, Synthesia, Colossyan, Hour One AI, ElevenLabs, Corti, and Cradle.bio.
LLM Infrastructure & Ops: Many European companies are developing infrastructure to support AI-driven products, addressing challenges including limited model knowledge, access to AI, costly inference, and AI governance. Notable emerging players include Nuclia, Deepset, Weaviate, Qdrant, MindsDB, Humanloop, dust.tt, klu.ai, Qevlar, TitanML, Nebuly, Holistic AI, Context, and Enzai.
Foundation Models: These underpin the GenAI tech stack, and market-leading models are currently being scaled by European startups including Aleph Alpha and Mistral.AI.
We are conscious that all companies innovating in these areas face challenges. For example, those developing B2B Applications face intense competition and need a ‘data moat’ – a large, unique proprietary data set – to help prevent easy replication of features by competitors. Meanwhile, LLM Infrastructure & Ops companies face huge pressure to reduce running costs, optimise inference, and build community for their products in order to create long-term defensiveness and trust. Foundation Models present a huge opportunity, but are also expensive and time-consuming to produce and take to market. Moreover, hallucination continues to be a concern for many LLM applications, and existing solutions like vector databases aren’t fit-for-purpose for all data sets.
The companies we identify in the market map below are making significant strides across all areas of the emerging GenAI tech stack through creating revolutionary software products and robust infrastructure solutions.
- If you’re a founder innovating in this space, please get in touch with Haakon Overli (haakon@dawncapital.com), Norman Fiore (norman@dawncapital.com), Henry Mason (henry@dawncapital.com), Shamillah Bankiya (shamillah@dawncapital.com) and Zoe Qin (zoe@dawncapital.com).
- For our piece last year on the landscape – and our take on why it is far from all ‘hype’ – please click here.
