The EU Is Betting $56 Million On Open Source AI


For years, the European Union has been faster to regulate artificial intelligence than to invest. Now, with the Battle for AI supremacy warming with a new powerful open source model from Deepseek from China built for a fraction of the price of its American competitors, it leaves the touch. This morning, Brussels announced its intention to develop its own Open Source model, with $ 56 million in funding to do so.

The investment will finance the best researchers of a handful of companies and universities in the EU countries, because they develop a large language model which can work with the 30 languages ​​of the block. The project will also operate the supervisors of companies like Mare Nostrum from Spain and Leonardo from Italy, who both received funding from the EU.

The budget is a tiny fraction of money side in the main American laboratories like Openai, which would have risen to 25 billion dollars to an assessment of $ 300 billion, or even one of the hottest IA unicorns in Europe, Mistral, which has raised $ 640 million Last year. But given the breakthrough of Deepseek, which once sounded like a change of chump in the space of AI could go much further: the support of the EU for the project is 10 times what Deepseek claims to spend on S ‘ lead one’s own R1 model. “I think Europe should now be considered in the race,” said project coordinator Jan Hajič, professor of computer linguistics at Charles de Prague University.

The project aims to finance an open source LLM that European companies and governments can rely on, assured that it has been built with the values ​​of the cooked EU. gardens. “There is a need for open source models that are aligned in terms of languages, values ​​and with society as a whole,” explains Peter Sarlin, co -founder of Finnish AI Lab AMD Silo. (Mistral, which is based in Paris and was estimated at more than $ 6.2 billion last year, made a similar argument, but it does not participate in the project.)

Europe has good reasons to build its own LLM: Meta,, OPENAI And other American companies have slowed down or blocked the release of tools in Europe due to legal risks from its AI Act, which aims to prohibit the “high -risk” uses of AI and come into force in February. The founder of Meta, Mark Zuckerberg and the co -founder of Spotify, Daniel Ek, warned that “incoherent and complex regulations” could lead Europe last year.

Since the adoption of the law last summer, the EU has made some movements to strengthen its technological sector. Its annual budget for research grants has been increased by a quarter to $ 1.5 billion. He will also invest $ 1.5 billion in upgrading the continent’s supervisor, and last week, the president of the Ursula von der Leyen commission revealed plans intended to stir up growth among its 27 member nations.

For some, the project recalls a failure project of $ 550 million to build a European search engine in rival google. Why invest in public funds, when venture capital has argued to support private projects? “Signing money in a fragmented manner is more a question of public relations than to conduct real results and impact,” warned the Rasmus Rotheri Capital technological investor.

More FO

ForbesThe Chinese AI company Trump says it is an “ alarm clock ” for the Silicon ValleyForbesWhy these startups IA Chip are delighted with the Freakout DeepseekForbesWhy AMD spent $ 665 million by buying a small AI research teamForbesOpenai believes that Deepseek “ distilled ” his data for training – there is what you need to know about the technique

Leave a Reply

Your email address will not be published. Required fields are marked *