According to Dynamix Beating monitoring, Mistral AI co-founder and CEO Arthur Mensch recently made a strong statement at a hearing in the French National Assembly. Faced with concerns from members of parliament about whether the company would be acquired by a U.S. tech giant, he explicitly rejected the idea. Mensch criticized the industry culture in Europe where startup companies are always looking to sell out to Silicon Valley for a quick payday, stating that a successful company should not entertain acquisition offers as that would, in a way, signify failure. He also publicly challenged the U.S. security benchmark Anthropic for the first time, claiming that Mistral's model is fully capable of identifying all the network vulnerabilities discovered by Mythos.
To support its ambitious independent development, Mistral has invested a staggering €1 billion in R&D this year. Mensch revealed that currently 75% of the company's revenue comes from within Europe. In terms of technology strategy, Mistral insists on internally training large-scale models with computational power and then using distillation technology to provide customers with more efficient smaller models. To ensure independent computational power, the company plans to build an 80 MW compute cluster in France next year and aims to reach a target of 1 GW by 2029.
Mensch emphasized that the essence of AI is to convert electrical power into Tokens. Relying mainly on nuclear power, deploying computational power in France can significantly reduce the carbon footprint. Using the construction of a 1 GW data center costing €50 billion as an example, he pointed out that electricity costs actually account for only 10% of the final output value. He warned that if Europe backs down now due to high costs, it will not only completely lose control of the underlying computational power but also face a trade deficit of trillions of euros annually in the future by relying solely on imported U.S. AI services.
