Nvidia raids Groq expertise to consolidate its management in AI chips | Companies | EUROtoday

New main operation from Nvidia. Well-known startup Groq has introduced the signing of a non-exclusive licensing settlement with the chip big for its inference expertise, indicating that this alliance displays a shared strategy to increasing entry to high-performance, low-cost inference.
As a part of this settlement, Jonathan Ross, founding father of Groq, Sunny Madra, president of the start-up, and different members of the staff will be part of Nvidia to drive and scale the licensed expertise.
According to CNBC and different American media, the settlement consists of the acquisition of Groq property by Nvidia, valued at $20 billion, in what could be the most important company operation for Nvidia, the world’s largest firm by market capitalization.
Groq has indicated that it’ll proceed to function as an impartial firm, with Simon Edwards taking up as CEO. In addition, it has specified that its cloud enterprise, GroqCloud, will proceed to function with out interruptions.
Last September, Groq, a pioneer in AI inference, introduced a brand new funding spherical of $750 million, with a post-money valuation of $6.9 billion. The spherical was led by Disruptive with important funding from Blackrock, Neuberger Berman, DTCP and a significant West Coast funding fund supervisor. The increase additionally had ongoing help from Samsung, Cisco, D1, Altimeter, 1789 Capital and Infinitum.
The agency stated on the time that it was already offering greater than two million builders and Fortune 500 corporations with quick, inexpensive computing, and was increasing its world footprint, leveraging current knowledge facilities in North America, Europe and the Middle East.
With Groq on its aspect, Nvidia is poised to change into much more dominant in making synthetic intelligence (AI) chips.
Nvidia GPUs have change into the trade commonplace. However, Groq has been engaged on a special sort of chip referred to as an LPU (language processing unit), which the corporate claims can run LLM at 10 instances the pace whereas consuming a tenth of the facility. In reality, the aforementioned Ross is a specialist on this expertise.
Sources within the sector clarify that giant acquisitions within the US these days are made by way of asset purchases or acquisitions of lower than 50% to keep away from interference from the regulator or excessively delayed deadlines, such because the Meta and ScaleAI operation, wherein Facebook’s guardian firm purchased 49% however took over your entire staff. Groq has two fundamental companies: {hardware} (chips) and cloud. In this case, Nvidia doesn’t wish to enter cloud providers as a result of it might imply competing immediately with its fundamental purchasers similar to AWS, Azure or Google Cloud, however it’s fascinated about TPU expertise and the staff to develop evolutions.
In this fashion, it buys all of the property, takes the expertise staff that pursuits it, leaves the cloud service, Groq Cloud inside the firm, and doesn’t purchase it.
Thus, it acquires the property and expertise that pursuits it and avoids shopping for the corporate, additionally avoiding competitors and regulatory points, which might delay the transaction.
https://cincodias.elpais.com/companias/2025-12-25/nvidia-asalta-la-tecnologia-de-groq-para-consolidar-su-liderazgo-en-los-chips-de-ia.html