Groq

What it does

Groq is a cloud platform for AI inference. It pioneered the LPU and now uses LPX alongside NVIDIA GPUs to serve inference at scale. The company says it is building hundreds of megawatts of capacity.

Groq focuses on inference rather than training. The homepage states that training creates the possibility while inference creates the value, and that inference is becoming the bottleneck. Groq says it pioneered the LPU and that LPX now works alongside NVIDIA's next-generation GPUs to deliver inference reliably, affordably, and at scale. The company also says it is building hundreds of megawatts of capacity. The page does not list a pricing page, plan details, model catalog, rate limits, or platform support, so those facts remain unknown. The homepage mentions a $350 million Series A and a neocloud positioning, but gives no product-level specifics beyond inference hardware and scale.