Microsoft on Monday announced a new partnership with French start-up Mistral AI – Europe’s answer to ChatGPT maker OpenAI — as the U.S. tech giant seeks to expand its footprint in the fast-evolving artificial intelligence industry.
Microsoft said in a statement that it was investing in the 2 billion euro ($2.1 billion), 10-month-old business to help it unlock “new commercial opportunities” and expand to global markets, without providing further financial details.
Under the deal, Mistral’s large language models (LLM) — the technology behind generative AI products — will be available on Microsoft’s Azure cloud computing platform, becoming only the second company to host its LLM on the platform after OpenAI.
It will also see Microsoft bolster the start-up’s access to new customers as it rolls out its ChatGPT-style multilingual conversational assistant “Le Chat,” or “the cat.”
Mistral AI logo is displayed on a mobile phone screen.
Anadolu | Getty Images
Microsoft President Brad Smith said on Monday that the deal was an “important” signal of the company’s backing of European technology.
“I really think this day is one of the most important days in terms of Microsoft’s technology support for Europe,” Smith told CNBC’s Karen Tso at the Mobile World Congress tech conference in Barcelona, Spain.
“What we’re fundamentally agreeing to a long-term partnership with Mistral AI so that they can train and deploy their next generation models for AI on our AI data centres, our infrastructure, effective immediately,” he added.
Growing scrutiny
It comes as Microsoft is facing pressure from EU antitrust regulators over its reported $13 billion investment in San Francisco-based OpenAI. Asked whether the investment was an effort to appease competition concerns, Smith said the company was committed to having a diverse product offering.
“It’s important for us to show that this isn’t just about Microsoft technology, it’s not just about American products. This is going to be an engine for technology, innovation and growth in Europe as well,” he said.
Smith said the investment in Mistral AI would also see funds dedicated to research and development, including AI models for public sector services in Europe.
“I think Europe wants, needs and deserves a wide range of offerings,” Smith said.
Earlier on Monday, Spanish telecoms giant Telefónica announced that it had struck a deal to integrate Microsoft’s Azure AI Studio into its digital ecosystem, Kernel, allowing staff to interpret data using generative AI language models.
Rival tech giants such as Google and Amazon have also been ramping up their investments in AI amid a growing frenzy around the emerging technology. Global markets hit record highs last week on the back of bumper results from AI chipmaker Nvidia.
A Tesla robotaxi drives on the street along South Congress Avenue in Austin, Texas, U.S., June 22, 2025.
Joel Angel Juarez | Reuters
Tesla is recruiting test drivers in New York to operate cars with “automated driving systems,” but the company hasn’t applied for the permits it would need to test autonomous vehicles in the nation’s largest city.
A job opening on Tesla’s website says the company is looking to hire vehicle operators in the borough of Queens. The hires will be “responsible for driving an engineering vehicle for extended periods, conducting dynamic audio and camera data collection for testing and training purposes.”
A spokesperson for the New York City Department of Transportation told CNBC on Monday that Tesla has not applied for approvals to test AVs on city streets in New York. InsideEVs, an electric vehicles news site, previously reported that Tesla was hiring test drivers for its robotaxis in Brooklyn.
Any company that obtains a permit to test AVs in New York has to keep “a trained safety driver behind the wheel, ready to take control of an AV-enabled vehicle at all times,” according to the DOT spokesperson.
Tesla didn’t respond to a request for comment.
Alphabet’s Waymo, the robotaxi leader in North America, has applied to test its AVs in New York, but its application remains under review, the DOT said Monday.
Tesla CEO Elon Musk has been trying to sell investors on a future for his company that’s built around AI and robotics, rather than sales of its existing vehicles. But Tesla still earns almost all of its revenue from sales of EVs and battery energy storage systems.
Tesla’s EV sales have been on the decline this year, especially in Europe, partly due to Musk’s decision to focus on the Cybertruck, rather than producing a more affordable EV with mainstream appeal. Some of the company’s struggles are the result of a political backlash against Tesla because of Musk’s incendiary political rhetoric, work with President Donald Trump, and endorsements of Germany’s anti-immigrant AfD party.
Along with its recruiting efforts in Queens, Tesla is also seeking to hire test drivers for its Autopilot team to gather data from drives in cities and suburbs of Dallas, Houston, Tampa, Orlando and Miami, as well as Palo Alto, California, home to Tesla’s engineering headquarters.
The current listings on Tesla’s website say Autopilot vehicle operators may need to travel to international and domestic destinations and must be familiar with “automated driving systems,” suggesting planned or ongoing testing of Tesla’s robotaxi and FSD or Full Self Driving system, currently marketed as FSD Unsupervised in the U.S.
Tesla notched a win in Texas last week, obtaining a permit to run a ride-hailing service in the state. The Tesla Robotaxi LLC permit and state regulations do not require Tesla to keep a human safety driver on board.
However, Tesla has been operating a fleet of robotaxis in Austin since late June, with employees riding in the front passenger seat, tasked with manually intervening during a trip if necessary. The service has only been accessible to invited users. Musk said in a post on X over the weekend that he intends for the Austin service to open to the general public next month.
In San Francisco, Tesla is also operating a limited, manned car service but promoting it as “autonomous ride-hailing.”
Musk posted last week that the company is “working as quickly as possible to get 100+ Teslas operating for autonomous ride-hailing (can’t use the word “taxi” or “cab” in California) in the Bay Area and allow anyone to request a ride.”
The company is not authorized to carry passengers on public roads in autonomous vehicles in California, the California Public Utilities Commission told CNBC in a recent email.
Tesla’s approach to AVs has drawn federal probes, product liability lawsuits and recalls following injurious or damaging collisions that occurred while drivers were using the company’s Autopilot or FSD systems.
The California DMV previously sued Tesla, accusing it of false advertising around its driver assistance systems.
While Tesla owners manuals say the Autopilot and FSD features in their cars are “hands on” systems that require a driver ready to steer or brake at any time, Tesla and Musk have shared statements through the years saying that a Tesla can “drive itself.”
Nikoli’s president, Yoshinao Anpuku, poses for a photo at Nikoli headquarters in Tokyo on March 19, 2025. LinkedIn worked with Nikoli and Sudoku champion Thomas Snyder to launch its Mini Sudoku game.
Nikoli
LinkedIn on Tuesday released a new game for the professional social networking app’s 1.2 billion users. It’s a miniature version of Sudoku, an old game with a rich history.
The new Mini Sudoku is LinkedIn’s sixth game. It’s scaled down from the traditional 9-by-9 grid and meant to be completed in two or three minutes.
“We don’t want to have a puzzle on LinkedIn that takes 20 minutes to solve, right?” said Lakshman Somasundaram, a senior director of product at the Microsoft subsidiary, in an interview with CNBC. “We’re not games for games’ sake.”
The introduction has the potential to strike a nostalgic chord and spark competition with colleagues, friends and family members for how fast the puzzle can be solved.
As with other puzzles in the app, Mini Sudoku gets harder as the days progress through the week.
LinkedIn added games last year to increase the fun and give users something new to talk about with one another.
Millions of people play LinkedIn’s games every day, a spokesperson said. The most popular time is 7 a.m. ET, and Gen Z is the top demographic. Of those who play today, 86% will return tomorrow, and 82% will be playing next week, the spokesperson said.
Launched in 2003 and acquired by Microsoft for $27 billion in 2016, LinkedIn remains in growth mode. Revenue increased about 9% to $4.6 billion in the latest quarter and membership reached 1.2 billion. Meta‘s social networks are more popular, with a combined 3.5 billion daily users and 22% revenue growth.
Unlike Meta, LinkedIn gives recruiters tools for finding candidates, and job seekers can apply for openings listed on the site. LinkedIn also now promotes a personalized feed of videos, similar to Google’s YouTube, TikTok and Meta’s own Facebook and Instagram.
Read more CNBC tech news
Making the game
LinkedIn’s development of the game resulted from an encounter with Japanese publisher Nikoli, which popularized Sudoku.
Somasundaram and a band of LinkedIn associate product managers visited Nikoli’s Tokyo headquarters late last year and spoke through a translator about puzzles with the publisher’s employees. That led to weeks of meetings involving LinkedIn, Nikoli and Thomas Snyder, a three-time World Sudoku Championship winner who has helped LinkedIn with its gaming strategy.
The group hoped to make Sudoku more accessible, building several prototypes before landing on the board with six rows and six columns of squares.
“It’s very easy to just make a Sudoku grid,” Snyder said. “It’s very hard to make art in the form of Sudoku. And that’s what both Nikoli and we do.”
Snyder is founder and CEO of Grandmaster Puzzles, a publisher of Sudoku books. With a Ph.D. in chemistry, he goes by the nickname Dr. Sudoku and has contributed to the hint feature in LinkedIn’s Mini Sudoku and constructed some of the puzzles. With each day’s puzzle, there will be a video showing how Snyder solves it.
“I think it’s got the potential to be the largest of the games, just because it’s going to have a lot of brand awareness from moment one,” he said.
Sudoku’s history
Howard Garns, an architect from Indiana, came up with a game called “Number Place” that required people to fill in a grid with numbers from one to nine. No number can be repeated in a row or column, and all nine numbers must appear just once in each of the nine 3-by-3 grids that make up the puzzle.
Number Place debuted in the magazine Dell Pencil Puzzles & Word Games in 1979. It only took off after Nikoli included a spin on the puzzle in the October 1984 issue of Puzzle Communication Nikoli under the name “Suji wa dokushin ni kagiru,” which means “The numbers must be single,” a Nikoli spokesperson told CNBC in an email.
Readers abbreviated the puzzle’s name, calling it Sudoku.
At first, the publisher employed both the long name and the shorter Sudoku title in Puzzle Communication Nikoli. In 1992 Nikoli started using only the Sudoku name, the spokesperson said.
U.S. and European newspapers began publishing Sudoku puzzles in the mid-2000s. Sudoku joined The New York Times’ NYT Games app, which boasts 10 million daily users, in 2023. More than 100 media companies have licensed Nikoli’s Sudoku puzzles, the spokesperson said.
“The daily puzzles will only be available on LinkedIn each day, but we are looking forward to republishing selected puzzles from those in our magazine,” the spokesperson wrote.
As of July 25, porn sites are required to implement effective age verification methods for U.K. users.
Jack Taylor | Getty Images
It was well intentioned but a U.K. law mandating age verification on adult sites and a number of other platforms has sparked a backlash from both internet users in the country, and U.S. politicians and tech giants.
Last month, new provisions in the Online Safety Act requiring large online platforms to implement age checks to prevent children from accessing pornographic and appropriate material came into force.
The measures have led PornHub, RedTube and other porn sites to force U.K. visitors to sign up and verify their age to gain access to their services.
What is the Online Safety Act?
Broadly, the Online Safety Act is a law that imposes a duty of care on social media firms and other user-generated content sites to ensure they take responsibility for harmful content uploaded and spread on their platforms.
In particular, the legislation aims to prevent children from being exposed to pornographic content and material that promotes suicide, self-harm, eating disorders or abusive and hateful behaviour.
The regulation has been years in the making and faced numerous delays in its development — not least due to concerns that it may infringe internet users’ right to privacy and result in censorship.
Why has it led to backlash?
The latest measures have been imposed with the aim of ensuring children aren’t able to view harmful and inappropriate content.
However, they have led to complaints from internet users due to the requirement of having to share personal information such as their ID, credit card details and selfies — in some cases for platforms that don’t even qualify as porn sites.
Spotify, Reddit, X and a number of other platforms have introduced their own respective age verification systems to stop users under the age of 18 from consuming explicit content.
These moves have subsequently led to providers of virtual private networks (VPNs) to report that their services, which allow users to mask their location, are surging in the U.K.
Meanwhile, on Monday, Wikipedia was dealt a legal blow in the U.K. as a High Court judge ruled the platform should be treated as a “category one” service, which would subject to certain user verification requirements.
The Online Safety Act requires category one platforms to offer users the ability to verify their identity and access tools that reduce their exposure to content from non-verified users.
Wikimedia, the parent company of Wikipedia, has said previously that it could limit visitor numbers from the U.K. in order to exempt it from category one status.
U.S. politicians weigh in
A number of U.S. politicians have blasted the new rules in recent days. Last week, Vice President JD Vance — who has previously criticized the U.K.’s internet safety rules — again raised concerns with the law, fearing it could unfairly restrict American tech companies.
“I just don’t want other countries to follow us down what I think was a very dark path under the Biden administration,” Vance told reporters during a trip to the country last week.
House Judiciary Chairman Jim Jordan, R-Ohio, who also visited the U.K. recently, said in a statement after his return that sweeping online safety laws in Europe are having “a serious chilling effect on free expression and threaten the First Amendment rights of American citizens and companies.”
There has been speculation over whether the U.S. may press Britain to relax the regulations during trade talks — however, U.K. officials say the issue is not open to debate.
Could other countries follow suit?
Other countries are already adopting their own respective internet age verification laws.
Australia and Ireland have both passed similar age verification measures, while Denmark, Greece, Spain, France and Italy have started testing a common age verification app to protect users online.
In the U.S., Louisiana passed a law in 2022 requiring age verification on websites where at least a third of the content is of an adult nature, while several other states are seeking to pass similar legislation.