The CEO of Mistral, a French AI startup, Arthur Mensch, recently stated in an A16z podcast that every nation should begin developing its own artificial intelligence (AI) infrastructure, as it signals a significant economic transformation on the horizon.
"In the coming years, AI will have a double-digit impact on the GDP of every country," Mensch said in the podcast episode uploaded on Thursday. Gross Domestic Product is a key indicator that measures the value of goods and services produced by a country.
Mensch warned that countries failing to establish independent AI systems may face the risk of capital outflow, similar to those nations a century ago that didn't build power plants and had to rely on neighboring countries for electricity. He pointed out that while AI shares some similarities with electricity, as a content-generating technology capable of transmitting national values and culture, AI requires more participation from governments.
Mensch appeared alongside Nvidia's CEO, Jensen Huang, in this program focused on sovereign AI. Both executives agreed that, given the extensive application of AI technology in sectors like public services, agriculture, and defense, each country needs to develop a national AI strategy and build its own AI capabilities. Huang has previously expressed similar views to government officials.
Mistral, founded by former DeepMind and Meta researchers, is working on constructing large language models and has launched a generative AI chatbot called "Le Chat." This product directly competes with rivals such as OpenAI, Anthropic, and DeepSeek, claiming that its models operate much faster than similar offerings.
As one of Europe’s most valuable startups, Mistral secured backing from several investment firms, including General Capital, Lightspeed, and Andreessen Horowitz, in a funding round in June. The company reached a valuation of $6.2 billion. Mensch earlier disclosed plans to take the company public rather than pursue an acquisition.
In the podcast, Mensch reiterated his strong support for open-sourcing large language models. He believes that open-source initiatives accelerate the democratization of technology since anyone can modify and redistribute the code. In contrast, companies like OpenAI, which favor closed-source models, argue that keeping code private enhances security.
"From 2010 to 2020, the acceleration of technological progress was due to various labs building on each other's work. However, this momentum has somewhat diminished with the advent of OpenAI's first large language model," Mensch noted. "Our goal in founding Mistral is to restart the open-source flywheel where 'I contribute something, another lab contributes something else, and we iterate from there.'"
