Groq seeks $650M to scale inference cloud after Nvidia's $20B talent-and-license deal

AI chip startup Groq is back in the funding game, reportedly asking existing investors for $650 million to expand its inference-as-a-service cloud business. The news, first reported by Axios, comes just months after the company struck an unusual $20 billion arrangement with Nvidia that was part talent acquisition, part technology license.

That December deal saw several top Groq engineers move to Nvidia and the licensing of Groq's proprietary hardware designs—effectively a payout to investors without a full buyout. Now those same backers are being asked to double down on Groq's inference cloud, which lets developers and enterprises run AI models after training, a segment currently generating far more demand than model building itself.

The company is currently led by interim CEO Adam Winter and interim CFO Matt Eng. According to Axios, the round may be all but guaranteed: existing investors Disruptive and Infinitium have agreed to cover the full amount if other backers decline their pro-rata allocations. Groq's pivot reflects a broader industry shift as companies race to serve the compute-hungry inference workloads that power real-world AI applications.