Nvidia raises the stakes on AI: billion-dollar deal with Groq between chips, licenses and talent | 3 components of computer hardware | Cpu hardware list and their functions | Hardware examples | Turtles AI

Nvidia has signed a deal worth about $20 billion with AI chip startup Groq, acquiring its assets and licensing its inference technology. The founder and senior executives will join Nvidia, while Groq remains independent.

Key Points:

• Nvidia enters huge deal worth around $20 billion for Groq technology.
• The agreement is not a traditional merger, but a non-exclusive license with talent transfer.
• Groq will continue to operate as an independent company, with new CEO Simon Edwards.
• Nvidia aims to integrate low-latency processors into its AI inference designs.

In a move that many are calling the largest AI-related deal in Nvidia’s history, the chip giant has reached an agreement with the young startup Groq, known for its accelerator chips dedicated to AI model inference, with a value in circulation of around 20 billion dollars: it is not an acquisition tout court but a mix of non-exclusive licensing of the technology and "acqui-hire" of key talents.In other words, Nvidia will pay large sums to obtain the right to use and develop Groq’s technology in its products and, at the same time, will bring founder Jonathan Ross and other executives into the workforce, including president Sunny Madra, bringing skills considered strategic into its ecosystem.

Groq, founded in 2016 by former engineers who designed Google TPUs, had emerged as one of the hottest startups in the AI ​​inference chip sector, with a valuation of around $6.9 billion after a $750 million funding round led by prominent investors including BlackRock, Neuberger Berman, Samsung and Cisco.The deal with Nvidia, which would far surpass the Santa Clara company’s previous record, the purchase of Mellanox in 2019 for almost 7 billion, is structured to keep Groq as a separate legal entity: its cloud business, GroqCloud, is not included in the transaction and will continue to operate under the leadership of new CEO Simon Edwards.

Internally, according to an internal email from Nvidia CEO Jensen Huang, the idea is to integrate Groq’s low-latency processors into Nvidia’s AI architecture, thus expanding the ability to handle real-time inferences and workloads, a segment of growing importance as AI applications move beyond simple training to immediate, scalable responses.Groq itself, while confirming the technological license, underlined that it will maintain operational independence, creating a curious hybrid between partnership, technological alliance and talent transfer.

The deal fits into a broader context in which large tech players seek to secure expertise and technologies without facing the regulatory hurdles that a full merger would entail, as demonstrated by other similar cases of targeted licensing and acquisitions in the AI ​​sector.Nvidia, for its part, comes to this operation with a very solid financial position, having accumulated over 60 billion dollars in liquidity and short-term investments, a figure that allows it to engage in aggressive investment activity in the AI ​​chip market.

Thus, between astronomical figures and the intertwining of technology and people, the semiconductor sector is experiencing a chapter full of meaning, where every market move can also be read as a sign of the times in the evolution of AI.