French AI Mistral raises $3.5 bn amidst strategy shift

LONDON — Mistral, a French artificial intelligence company, has been promoted as one of Europe’s best hopes of competing against American and Chinese firms in developing the rapidly advancing technology.
But in recent months, the Paris-based startup has found it harder to keep pace. Mistral has struggled to match the financial resources or the technical output of its international AI competitors.
The company is now shifting its strategy.
On Tuesday, Mistral said it had raised 3 billion euros, or about $3.5 billion, to help make the change. The company is moving away from solely developing AI models that compete so directly with companies such as Anthropic, OpenAI and DeepSeek. Instead, Mistral will also build data centers and other infrastructure that businesses can use in Europe and elsewhere to deploy AI.
Mistral is trying to take advantage of growing concern among many businesses and governments in Europe and elsewhere about overdependence on American technology. Companies such as Anthropic and OpenAI charge customers a premium to use their AI systems but provide little access to the underlying data or models. President Donald Trump’s decision in June to use an export ban to briefly limit international use of Anthropic’s most advanced model also caused alarm.
Founded by three French AI researchers with experience at Google and Meta, Mistral started by developing open-source models that businesses can freely download and customize.
Mistral has now raised about 6 billion euros since it was started in 2023. The latest round, which was led by Samsung Electronics, values the company at about 21 billion euros. The Scaleup Europe Fund, which has money from the European Union, also contributed.
Mistral’s funding levels are huge for a startup in Europe, which has far less venture capital and a smaller tech industry than the United States does, but are still small in comparison with its overseas rivals.
Anthropic has raised roughly $125 billion and is exploring an initial public offering that could value the company at $2 trillion. OpenAI has raised roughly $180 million and was valued in March at more than $850 billion.
Building the most advanced AI models requires huge amounts of money for computing power, data centers and other infrastructure.
Mistral first outlined its strategy shift last month. Johan Bergqvist, Mistral’s chief financial officer, said the new money would be used to continue developing frontier AI models, as well as to provide infrastructure and engineering help for companies and governments to integrate AI into their operations. Organizations will be able to use Mistral’s models or open-source models made by other companies, including Chinese firms such as Z.ai.
Mistral’s strategy centers on demand for AI “sovereignty,” in which companies and governments in Europe want more control over how the technology is built and used inside their operations. Mistral is building data centers in France and Sweden, as organizations seek ways to keep their data stored in Europe.
“We see companies increasingly reaching out to us because they want to be in control of their own AI transformation,” Bergqvist said.(The New York Times)