Nvidia’s (NVDA) recent licensing agreement with chip startup Groq (GROQ.PVT) underscores how the AI chipmaker is leveraging its massive cash reserves to sustain leadership in the AI market.
The company announced a non-exclusive deal to license Groq’s technology and hired Groq founder and CEO Jonathan Ross, its president, and other key employees. CNBC reported the deal could be worth $20 billion, marking Nvidia’s largest transaction to date, though the company declined to comment on the figure.
Bernstein analyst Stacy Rasgon said in a note to clients that the Nvidia-Groq agreement “appears strategic in nature for NVDA as they leverage their increasingly powerful balance sheet to maintain dominance in key areas.” Nvidia’s cash inflows rose over 30% year-over-year to $22 billion in its most recent quarter.
“This transaction is essentially an acquisition of Groq without being labeled one, likely to avoid regulatory scrutiny,” noted Hedgeye Risk Management analysts.
Part of a broader AI strategy
The Groq deal is the latest in Nvidia’s string of AI investments, which span the full AI ecosystem. These include large language model developers such as OpenAI (OPAI.PVT) and xAI (XAAI.PVT), as well as AI-focused cloud providers like Lambda (LAMD.PVT) and CoreWeave (CRWV), which compete with Nvidia’s Big Tech clients.
Nvidia has also invested in other chipmakers, including Intel (INTC) and Enfabrica, and previously attempted to acquire British chip designer Arm (ARM). While some critics have raised concerns about circular financing reminiscent of the dot-com era, Nvidia has denied any wrongdoing.
Groq, founded in 2016, had positioned itself as a potential rival, producing LPUs (language processing units) aimed at AI inferencing, marketed as alternatives to Nvidia’s GPUs (graphics processing units).
Inferencing vs. training: where competition is emerging
While Nvidia dominates AI model training, analysts note that inference workloads may see more competition. Chips like Google’s (GOOG) TPUs and Groq’s LPUs can outperform GPUs on certain tasks due to efficiency advantages. LPUs use on-chip SRAM memory, making them faster and more energy-efficient for specific models, while Nvidia GPUs rely on off-chip HBM from suppliers such as Micron (MU) and Samsung (005930.KS).
Groq CEO Jonathan Ross said the company aims to provide chips for half of global AI inference workloads at lower costs. Ross previously helped develop Google’s first-generation TPUs, making him a key figure in Nvidia’s competitive landscape.
Strategic “acqui-hire”
Cantor Fitzgerald analyst CJ Muse described Nvidia’s move as both offensive and defensive, allowing the company to license Groq’s intellectual property while securing top talent. Muse added that the deal positions Nvidia to capture “even greater share of the inference market.”
Nvidia shares rose about 1% following the announcement. However, some analysts questioned the $20 billion valuation, pointing out that Groq’s chips remain unproven for large AI models due to limited memory capacity. “Groq’s current technology is greatly limited to only a small subset of inference workloads,” noted DA Davidson analyst Alex Platt.
Overall, the Groq deal highlights Nvidia’s strategy of using its financial strength to consolidate AI leadership while preemptively neutralizing potential competitors.
Ready to start trading Forex? Join iXBroker today and kick-start your trading journey now!
