Nvidia Unveils Groq 3 Chip, LPX Rack and NemoClaw AI Agent Platform
Nvidia unveiled its Groq 3 LPU and LPX rack at GTC, integrating the $20 billion Groq deal into the Vera Rubin platform to reinforce inferencing dominance. It also launched NemoClaw for secure AI agent deployment via OpenClaw apps, marking a shift from general-purpose GPUs to specialized inferencing.
1. Strategic Pivot to AI Inferencing and Agents
In his multi-hour keynote at GTC, Jensen Huang emphasized Nvidia’s shift from general-purpose GPUs to specialized inferencing solutions and AI agents, underlining the growing demand for real-time model deployment and autonomous functionality across industries.
2. Groq 3 LPU and LPX Rack Unveiled
Nvidia introduced its Groq 3 language processing unit and the accompanying LPX server rack, designed to accelerate AI inference workloads with higher throughput and efficiency. These new products expand Nvidia’s Vera Rubin family to include GPUs, LPUs and CPUs, reinforcing its end-to-end AI infrastructure.
3. $20 Billion Groq Acquisition Integration
Following its $20 billion deal to acquire Groq’s team and intellectual property, Nvidia has integrated Groq’s technology into its chip roadmap, aiming to outpace emerging competitors like Cerebras. The acquisition brought key personnel such as founder Jonathan Ross and president Sunny Madra into Nvidia’s AI division.
4. NemoClaw Platform for Secure AI Agents
Nvidia launched NemoClaw to enhance privacy and security for AI agents running on the OpenClaw ecosystem, enabling model execution on platforms like WhatsApp and Slack. This platform marks Nvidia’s entry into the agentic AI space, catering to enterprises seeking robust autonomous solutions.