NVIDIA finalized $20 billion acquisition of Groq to integrate specialized AI inference processors.

Nvidia announced its largest-ever acquisition, agreeing to buy assets from AI chip startup Groq for approximately $20 billion in cash, alongside a non-exclusive licensing agreement for Groq's technology and the recruitment of key executives and engineers. The deal strengthens Nvidia's position as demand for artificial intelligence hardware continues to surge worldwide, with Nvidia's market capitalization exceeding $3 trillion, driven largely by booming data-center revenues. The transaction underscores Nvidia's strategy of consolidating cutting-edge technologies to maintain its lead in the AI semiconductor market.

Founded in 2016 by former Google engineers, Groq specializes in high-speed, energy-efficient AI inference chips, designed to deliver rapid responses from trained AI models. These chips directly challenge Nvidia's dominance in inference workloads, an increasingly critical segment as generative AI applications scale across cloud services, finance, healthcare and government systems. Nvidia is expected to integrate Groq's designs and talent into its future product roadmap, complementing its leadership in AI training chips.

Groq had raised around $1 billion from investors and secured partnerships with major cloud providers, positioning itself as one of the most credible challengers to Nvidia. The acquisition comes amid intensifying competition and regulatory scrutiny in the AI semiconductor space, making this consolidation strategically significant for maintaining competitive advantage.

Importantly, the transaction involves Groq, the chipmaker, and not Grok, the AI chatbot developed by Elon Musk's xAI, a distinction that has caused some confusion. Separately, Musk's xAI has pursued a different front in the AI race. In October 2025, it launched Grokipedia, an AI-generated encyclopedia built using the Grok language model. Musk has described the project as an effort to create the "largest and most truthful" knowledge base, combining AI-generated content with crowdsourced inputs, while openly criticizing what he sees as bias in Wikipedia.