Groq Secures $650M to Expand AI Inference Cloud Platform

Seven months after a $20 billion licensing deal with Nvidia, Groq Inc. announced it has raised $650 million in funding, led by Disruptive and Infinitum.

Groq's LPU chip design optimizes AI inference workloads and is part of Nvidia's Grok LPU 3, launched in March. The LPQ system features 32 trays housing Groq LPU 3 processors, ensuring efficient data traffic management with synchronized clocks.

The LPU 3 delivers 2.5 terabits per second and includes 500 MB of onboard SRAM, vastly improving inference speeds. Groq's cloud platform processes trillions of tokens weekly for 5 million developers across 13 data centers worldwide. The new funding aims to enhance processing capacity to 200 megawatts by 2027, utilizing Nvidia's LPX liquid-cooled appliance.

Groq plans to differentiate itself in the competitive landscape by expanding its offerings to include managed database services, positioning itself against rivals like CoreWeave Holdings.