A French Startup Wants to End Europe’s Reliance on American AI Tools
13:29 Share

As companies create AI-powered solutions, how can they ensure they're effective and trustworthy? Join IBM at the break to hear how companies can build trust in their AI with Hrithika Gunnar, IBM's General Manager for Data and AI.

Welcome to Tech News Briefing. It's Thursday, June 12th. I'm Victoria Craig for The Wall Street Journal. Silicon Valley is a dominant force in the global AI race, but European customers say they want to break dependence on American firms. Can a French startup be part of that solution? Then The World is Running Out of Clean Water will introduce you to a technology that promises to fix the problem.

But first, anxiety has been rising in Europe since February, when U.S. Vice President J.D. Vance took the stage at the Paris AI Summit and said this. AI, we believe, is going to make us more productive, more prosperous, and more free. The United States of America is the leader in AI, and our administration plans to keep it that way. Vance criticized European regulations that the Trump administration worries could constrain further development in artificial intelligence.

But he said the U.S. is open to working with European players. Now, French startup Mistral is stepping up to the plate. At the Viva Technology conference in Paris this week, WSJ tech reporter Sam Schechner spoke to Mistral's CEO, who said demand for his company's AI tools is booming because, quote, European leaders just don't want to be talked to that way.

Sam, let's just back up a second because I'm not sure Mistral is a household name for many of our listeners. So remind us what it is and what it does. Mistral is a French company. It's less than two years old, and it is one of Europe's biggest developers of large language models. It's a startup.

But it's in competition with folks like OpenAI and Google when it comes to the market of really developing this general purpose generative AI tools for things like computer coding, for

So it wants to compete with the largest of the large companies in this AI space. How competitive is it with its U.S. rivals? Because I think a lot of our audience tends to think of AI and think of either the behemoths in the U.S. or even those in China. But how competitive is it with its U.S. rivals?

The company's chief executive told you that European companies and governments are increasingly reaching for AI tools not dependent on U.S. tech giants.

Yeah, well, I mean, Mistral is notable because it is really one of the few companies in Europe that is keeping up in this space. It's founded by somebody who formerly was at DeepMind and two other people who were at Meta, but they're French and they wanted to start a company here in part to show that it was possible to do this in Europe. And their tools maybe aren't like

at the cutting cutting edge of the most recent tools from those big labs but perform almost as well and what they say is that it's a lot more efficient it's a cheaper thing to run for a business and they also release them open source like some of those chinese companies that you reference deep seek for instance open source can be a more efficient way to run some companies

And governments aren't comfortable with potential security risks of running software from China. And that makes perhaps Mistral a sweet spot for some businesses. At least that's certainly what the company is hoping. NVIDIA CEO Jensen Wang was actually giving the keynote speech at the conference that you're attending there. And he talked about partnering with Mistral on a new data center. Today, we're announcing that we're going to build an AI cloud together here.

to deliver their models as well as deliver AI applications for the ecosystem of other AI startups so that they can use the Mistral models or any model that they like.

Tell us more about what we know about this data center. This is part of Mistral's new expansion, not just into developing models and into then partnering with companies that are going to use tools based on those models, but actually to offer the underlying compute that will allow companies to basically contract with Mistral in what we call the full stack way. So where they're controlling the hardware and the AI and the software on top of it for

For some companies in Europe and governments as well, which are looking for strategic autonomy, they don't want to necessarily have a supply chain that is tied too closely to the U.S. This could be a solution for that. Obviously, NVIDIA is an American company, but if they're going to be sitting 20 miles south of Paris in a Mistral data center, if you're a company, you want to control your data, that could be very attractive.

In addition, NVIDIA's CEO said, and Arthur Mensch, the CEO of Mistral, told me that they are looking possibly to work with other companies that want to develop their own models and could use this compute. So in a way, Mistral is moving to a more base level of the AI business and they could become a provider to other AI developers to train their own models too. Yeah.

This isn't just an ambition that Mistral has to become a leader or a power player in AI. France has promoted itself as a destination for AI. And President Emmanuel Macron on stage this week with Mistral and NVIDIA's CEOs encouraged French companies to sign agreements to use Mistral's new data center. He called it our fight for sovereignty or strategic autonomy.

There's a race by countries to make themselves friendly for AI. I think everybody wants to have

AI that's based in their area. It's almost a sovereign imperative. You think if this technological revolution is going to upend businesses and industries globally, you want some of it where you're based. And so you see companies like OpenAI doing deals with places like the UAE. And you see Jensen Huang of NVIDIA going around the world selling sovereign AI. That helps him sell chips.

France's card to play in this is something that French President Emmanuel Macron spoke about in February when there was an AI summit here in Paris, which is its nuclear power. It has a surplus in energy and these chips use a lot of energy. So it's

So being in a country where there's energy to spare is something that's valuable. And you see investments from private actors, from governments, including the UAE, in building AI campuses here in France. And that's something where Mistral is hoping to capitalize on that. They actually have the models that run on those tools. And so when you have open AI and Mistral end up being major players doing this kind of thing outside of the U.S.

That was Sam Schechner, who covers tech for The Wall Street Journal. Coming up, can technology from the 60s solve the modern problem of severe water scarcity? We look into that after the break. Enterprise AI is an unstructured data problem at scale. How does generative AI address it? Rithika Gunnar, General Manager for Data and AI at IBM, explains. Think of this as emails, PDF, PowerPoint decks that sit in an organization. Generative AI has allowed us to unlock information

opportunity to be able to take the 90% of data that is buried in unstructured formats, which really unlocks a new level of driving data and insights of that data into your workflows, into your applications, which is essential for organizations as we go forward.

More extreme weather patterns, decimation of the world's aquifers, and growing urban populations. All of those things threaten to make water scarcity a much more acute problem in the coming decades. But a technology first floated, or rather sunk deep into the ocean, in the early 1960s is offering a possible solution. WSJ tech columnist Christopher Mims has been looking into this.

Christopher, what is this not-so-new technology for getting clean water, and why is it getting fresh attention now? So this desalination technology was invented in the 1960s, but it wasn't viable then because in order to make it work, you have to put the desalination plants at the bottom of the ocean, 400 meters down. And the reason you want it down that deep is just that the water's pretty clean, and

And the water pressure is very high. And the way that modern desalination works is you're pushing water across a membrane with tiny holes that only admit water molecules. And the holes are so tiny that they don't let salt through. And that's how you desalinate water. So the whole idea is, why don't we use the enormous pressure at the bottom of the ocean to push seawater through this membrane and

And yeah, you still got to pump some fresh water up to the surface, but it just takes a lot less energy. And that is the real bugaboo of desalination. It's why it's not ubiquitous. It just requires so much energy. It's the most expensive way to produce water. Yeah, I was going to say it's a difficult and a costly process. So what about this technology is making it viable now? Ironically, no.

It's the oil and gas industry that's making this technology viable now. There's been so much drilling and exploitation in the Gulf and in the North Sea and in other places like that, that you now have affordable, available, capable undersea robots and other types of technology, pumps, power cables, whatever, that have been tested and

and work at great depths. And a lot of these new deep sea desalination companies are coming out of a place in Norway around Oslo. Tell us some of the ways that these companies are testing this technology in this way. There are a handful of companies doing this

They all have pilot plants somewhere either under the ocean or in a reservoir. One of them has this cute little 40 ton pilot plant that they are supplying water to a boutique cocktail ice company with ice.

There are more serious applications. One is about to go into production at an offshore drilling facility in Norway called Mongstad, where, you know, if you have a platform out in the ocean, you still need fresh water there for your workers. There are islands in Greece that are talking to these companies. One of these companies, which is based in the U.S. called OceanWell, is in early conversations with the state of California about

to put these offshore and help with water-stressed cities in California like Los Angeles or Huntington Beach. So what will it take to make this technology applicable on a much larger scale? And how big is the need for desalinated water globally? Globally, about half of all humans on Earth

have at least one month out of the year where they experience severe water stress, according to the United Nations. And something like two to four billion people are permanently water stressed. And this is just getting worse. But their promise of this is going to be significantly cheaper than conventional onshore desalination, that has yet to be tested. And when you put things at the bottom of the ocean where we don't really know what the conditions are like long term,

There could be big temperature fluctuations. There could be fluctuations in salinity. There could be

seasonal changes in deep water currents that we don't know about. And all of this could affect the performance of these plants. And so if they can't produce water with significantly less energy and at a lower cost than conventional desalination, then this whole technology is a dead end. That was WSJ Tech columnist Christopher Mims. And that's it for Tech News Briefing. Today's show was produced by Julie Chang with Deputy Editor Chris Dinsley. I'm Victoria Craig for The Wall Street Journal. We'll be back this afternoon with TNB Tech Minute. Thanks for listening.

How can companies build AI they can trust? Here again is Hrithika Gunnar, General Manager for Data and AI at IBM. A lot of organizations have thousands of flowers of generative AI projects blooming. Understanding what is being used and how is the first step. Then it is about really understanding what kind of policy enforcement do you want to have on the right guardrails on privacy enforcement.

The third piece is continually modifying and updating so that you have robust guardrails for safety and security. So as organizations have not only a process, but the technology to be able to handle AI governance, we end up seeing a flywheel effect of

more AI that is actually built and infused into applications, which then yields a better, more engaging, innovative set of capabilities within these companies. Visit IBM.com to learn how to define your AI data strategy. Custom content from WSJ is a unit of the Wall Street Journal Advertising Department. The Wall Street Journal News Organization was not involved in the creation of this content.