Unlock the Editor’s Digest for free
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
Nvidia is scooping up the founder and other top talent from Groq, one of the most prominent start-ups aiming to challenge the chipmaker’s dominance in artificial intelligence processors.
Jonathan Ross, a former Google chip engineer who founded Groq in 2016, and the start-up’s president Sunny Madra, are among those who will join Nvidia as part of a technology licensing deal, the two companies announced on Christmas Eve.
Jensen Huang, Nvidia’s chief executive, said in an email to staff that the Groq deal would “expand the capabilities” of the data centres that were built around its chips, which he calls “AI factories”.
“We plan to integrate Groq’s low-latency processors into the Nvidia AI factory architecture, extending the platform to serve an even broader range of AI inference and real-time workloads,” he wrote in the email, which was seen by the Financial Times.
Despite the loss of much of its leadership team, Groq said it “will continue to operate as an independent company”.
Groq has focused on developing chips that can accelerate AI “inference”, the process of returning responses to users’ queries via chatbots such as OpenAI’s ChatGPT or Google’s Gemini.
The start-up was valued at $6.9bn as recently as September, when it raised $750mn in funding. Groq claims its language processing units are up to 10 times more energy efficient than the kinds of graphics processing units produced by Nvidia and its main rival AMD.
Ross helped kick-start Google’s AI Tensor Processing Unit chips programme before leaving to found Groq. TPUs are widely seen as a significant asset in Google’s AI arsenal, helping its Gemini chatbot catch up with ChatGPT by OpenAI, which has largely relied on Nvidia chips.
Several Big Tech companies — including Microsoft, Meta and Google — have struck licensing agreements with start-ups that brought in top talent and assets without making an outright acquisition, after antitrust scrutiny over the sector’s dealmaking escalated in recent years.
“Antitrust would seem to be the primary risk here, though structuring the deal as a non-exclusive licence may keep the fiction of competition alive,” said Stacy Rasgon, an analyst at Bernstein Research, in a note to clients regarding Nvidia’s deal with Groq.
Groq said its agreement with Nvidia “reflects a shared focus on expanding access to high-performance, low cost inference”.
The tie-up comes as many of Nvidia’s biggest customers are developing their own AI processors or exploring alternatives to its GPUs, including adopting Google’s TPUs.
The FT reported last week that Amazon was in talks to invest more than $10bn in OpenAI as part of a deal in which the ChatGPT developer would use more of the ecommerce giant’s Trainium series of AI chips.
Nvidia has struck a series of high-value investment deals this year, including agreeing to invest up to $100bn in OpenAI.
The company hit a record $5tn valuation in late October but has seen its shares decline as concern has grown about the sustainability of the AI boom. Nvidia’s stock is up 36 per cent year to date, while Google parent Alphabet’s has risen 66 per cent over the same period, driven by growing investor enthusiasm about its latest Gemini models.
Additional reporting by Hannah Murphy and Michael Acton