The AI LLM Battle, an EU perspective

Posted by

This title might seem strange, as all the attention for Artificial Intelligence Large Language Models goes to the US-based OpenAI, Google, and Microsoft, with their GPT, Bard, and GitHub Copilot respectively. But something is also brewing in Europe, but nobody is sure about what the strategy might be.

The European Union is working from different angles when it comes to Artificial Intelligence. On one side, there is an effort to regulate AI through the “EU AI Act” which the European Parliament calls “the world’s first comprehensive AI law”. The EU is a master is legislation and regulations, and its approach to AI from the perspective of risk, although correct in essence, has got a lot of AI experts and industry vendors running scarred, with claims that it will only “hinder innovation” and with the US Congress potentially being more of an advising government body on the AI matter, rather than a regulator.

At the same time, the EU is determined to invest billions of Euros in AI in the upcoming years through its Horizon Europe research grant scheme, or the Digital Europe program that provides funding for supercomputing, artificial intelligence, and cybersecurity among other themes. European academic institutions are heavily involved in these grant schemes, with consortiums being able to secure funding across multiple related projects, one example being AI on Demand, which is a continuation of other projects such as AI4Copernicus, ELISE, TAYLOR, HUMANE AI NET, and others. The funding the EU put into Artificial Intelligence in the past 10 years resulted in a plethora of research projects, research centers, innovation hubs, testing and experimentation facilities, ethical use frameworks, guidelines, and post-doc programs. But as you guessed it, no AI Large Langauge Model has resulted from any of these funding schemes.

The most notable European AI LMM providers, Aleph Alpha (Germany) and Synthesia (UK) are not beneficiaries of any EU public grant funding. According to Wikipedia, Synthesia raised $3.1 million in seed funding in 2019. In April 2021, the company raised $12.5 million in Series A funding.[9] In December 2021, it raised $50 million in a Series B funding round led by Kleiner Perkins and GV. As part of a Series C funding round in June 2023, Synthesia raised an additional $90 million from the likes of US fund Accel and Nvidia, granting the company a total valuation of $1 billion (thus making it a unicorn). Aleph Alpha, a direct competitor to OpenAI and Google, raised €23 million/$27 million in a Series A funding co-led by Earlybird VC, Lakestar, and UVC Partners. Following a seed round of €5.3 million from LEA Partners, 468 Capital, and Cavalry Ventures in November 2020, Aleph Alpha has now raised a total of €28.3 million.

And then there’s Mistral.AI from France, which raised over 100M EUR in their seed round, before the company had any staff (except for the 3 French founders), a product, and anything else for that matter, except a single-page website. The project is evidently designed to become a counterpart to the US-based OpenAI and Google, and industry voices are claiming this project is the French Government’s pet, as President Macron has clear intentions to find AI for “french speaking people” ( Again, no EU funding whatsoever.

As conclusion, Europe is definitely trying to catch up with the US when it comes to Artificial Intelligence and NLP (Natural Language Processing) more specifically. But it is doing it through private investment funds and venture capitalists, not through the European Union grants designed for AI. France’s strategy of taking ownership of AI research and advancements directly and through its own funding schemes shows just how far behind the real-world are the EU Digital Europe and Horizon funding programs are.