Nvidia's (NVDA) licensing deal with chip startup Glock (GROQ.PVT) shows how the tech giant is leveraging its massive capital to maintain its prominence in the AI market.
Nvidia announced this week that it entered into a non-exclusive agreement with Groq to license its technology and hired Groq's founder and CEO Jonathan Ross, its president and other employees. CNBC reported that the deal is worth $20 billion, making it NVIDIA's largest contract in history. (The company declined a request for comment on this figure.)
Bernstein analyst Stacy Rasgon said in a note to clients Thursday that the NVDA-Groq partnership “appears to be strategic in nature as NVDA leverages its increasingly strong balance sheet to maintain its leadership in key areas.” Nvidia's cash flow increased more than 30% year over year to $22 billion in its most recent quarter.
“This transaction…is essentially an unlabeled acquisition of Groq (to avoid regulatory scrutiny),” analysts at Hedgeye Risk Management added in a note Friday.
The move is just the latest in a series of AI deals by Nvidia, the world's first $5 trillion company. Chipmakers' investments in AI companies span the entire market, from large language model developers like OpenAI (OPAI.PVT) and xAI (XAAI.PVT) to “neoclouds” like Lambda (LAMD.PVT) and CoreWeave (CRWV) that specialize in AI services and compete with Big Tech customers.
Nvidia also has investments in chipmakers Intel (INTC) and Enfabrica. The company unsuccessfully tried to acquire British chip architecture design firm ARM around 2020.
Nvidia's extensive investments, many of them in its own customers, have led to accusations that it is involved in a circular lending scheme reminiscent of the dot-com bubble. The company strongly denied these claims.
Meanwhile, Groq was trying to become one of Nvidia's competitors.
Founded in 2016, Groq makes LPUs (language processing units) focused on AI inference and sold as a replacement for Nvidia's GPUs (graphics processing units).
Training an AI model involves teaching the model to learn patterns from large amounts of data, while “inference” refers to using that trained model to generate output. Both processes require massive computing power from AI chips.
While Nvidia easily dominates the AI training chip market, some analysts argue that Nvidia could soon face increased competition in the inference space. That's because custom chips like Google's (GOOG) TPU (tensor processing unit), and perhaps Groq's chip called an LPU (language processing unit), may be better suited for certain tasks. For example, LPUs utilize a type of memory technology called SRAM within the chip that makes them faster and more energy efficient for certain models. Nvidia GPUs, on the other hand, rely on off-chip HBM manufactured by companies like Micron (MU) and Samsung (005930.KS).
