The American electronic chip giant Nvidia will poach the management team of its competitor Groq, specializing in processors dedicated to generative artificial intelligence (AI), with which it has concluded a licensing agreement, according to a press release published Wednesday, December 24. Groq should not be confused with Grok, the AI interface from xAI, a company controlled by Elon Musk.
The announcement came a few minutes after CNBC quoted one of Groq’s investors assuring that the Californian start-up would be bought for $20 billion. Questioned by Agence France-Presse (AFP), the two groups declined to comment beyond Groq’s press release. “Nvidia will not acquire Groq”a source close to the matter told AFP, contradicting the boss of the private equity company Disruptive, Alex Davis, cited by CNBC.
Joining Nvidia are CEO and co-founder of Groq, Jonathan Ross, and President, Sunny Madra, “and other team members”revealed the company in its succinct press release. The operation is similar to a « acqui-hire » (mixture of hiring and acquisition), an old concept but brought up to date in recent years in the technology sector.
It involves one company poaching key executives from another and, in some cases, securing access to its technology. The maneuver is sometimes accompanied by a stake, but a minority one.
No rejection from regulators to fear
L’« acqui-hire » presents two major interests for a company seeking acquisitions. On the one hand, it does not take control of its target and therefore does not have to fear rejection from regulators for competition reasons. On the other hand, it avoids the substantial expenses linked to the repurchase of shares from existing investors in the event of an outright acquisition.
In June, Meta used a similar process with Scale AI, a specialist in data shaping for AI, taking only 49% of the capital and attracting boss Alexandr Wang.
Founded in 2016, Groq has developed chips called LPU (language processing unit) designed for inference, that is to say the use of large generative AI models, with greater efficiency, particularly energy. He thus chose specialization when Nvidia offers products adaptable to many different contexts of use, in particular the development of AI models before they are put online.
“Nvidia and [son patron] Jensen Huang is like Michael Jordan, the greatest basketball player of all timeGroq general manager Jonathan Ross explained to AFP in 2024. But inference is like baseball. And when Michael Jordan tried his hand at baseball, he didn’t do very well. »