Google DeepMind hires Contextual AI team in licensing deal

Google DeepMind has struck one of its most aggressive talent grabs yet, bringing in more than 20 researchers from startup Contextual AI while licensing the company’s technology in a transaction valued at roughly $80 million to $90 million. The move gives DeepMind a larger bench of scarce AI specialists and adds another example of how the industry is using hybrid deals to secure expertise without buying startups outright.

Contextual AI’s co-founder and chief executive, Douwe Kiela, is among those moving to DeepMind. The arrangement follows a pattern Google has used before, including a 2024 deal with Character.AI that gave Google a non-exclusive license to the startup’s model technology while co-founders Noam Shazeer and Daniel De Freitas joined Google. Alphabet also agreed last year to a deal involving Windsurf technology and staff, underscoring how the company has leaned on licensing and hiring to pull in talent and intellectual property.

The Contextual transaction lands at a moment when such acquihires are drawing closer scrutiny from antitrust officials. Acting Assistant Attorney General Omeed Assefi said in March 2026 that deals structured to sidestep merger review are especially concerning, describing acquihires as a warning sign for enforcement. The concern is not just whether a startup is sold, but whether a dominant tech company is effectively consolidating talent and technology through contracts that avoid the full scrutiny attached to a formal acquisition.

For Contextual AI, the deal closes a short but well-funded run. The company announced an $80 million Series A on August 1, 2024, led by Greycroft, with Bain Capital Ventures and Lightspeed Venture Partners also among its backers. Greycroft said it had backed founders Douwe Kiela and Amanpreet Singh since their seed financing in April 2023. Reuters previously reported that Contextual had raised $20 million in seed funding in 2023, after the startup was founded that year to build tools for production-grade enterprise large language models.

The broader message for the AI market is stark. Microsoft’s deal for Inflection and Amazon’s hiring of Adept founders and staff have already shown how common these structures have become. DeepMind’s latest move suggests the battle for frontier AI is increasingly being fought not only over model performance, but over who can lock up the researchers, ideas and engineering teams that make those models possible.