Groq · 2026-05-28 · major
Groq Raising $650M for 'Neocloud' Second Act After Selling Hardware Tech to Nvidia for $20B
Groq is raising $650M to pivot from chip-making to running an AI inference cloud, six months after Nvidia paid $20B to license its hardware and hire its top engineers.
Groq pivots from chipmaker to inference cloud and lines up a guaranteed $650M from the same backers it just cashed out.
Key specs
| Round size | $650M |
|---|---|
| Nvidia deal | $20B |
| Backstop investors | Disruptive, Infinitium |
What is it?
Groq is a Silicon Valley AI chip startup best known for fast LLM inference on its custom LPU silicon. After Nvidia paid an estimated $20B in December 2025 to license Groq's hardware tech and hire most of its senior team, the rump company is reincorporating around its inference cloud business and raising $650M to fund it.
How does it work?
Existing investors Disruptive and Infinitium have agreed to backstop the full $650M round, meaning the raise is effectively guaranteed. Pro-rata rights are being offered to the same backers who took cash distributions from the Nvidia deal. Interim CEO Adam Winter and CFO Matt Eng are leading Groq 2.0 toward operating GroqCloud as a token-as-a-service neocloud, hosting third-party inference rather than designing new silicon.
Why does it matter?
The story is a clean read on where the AI infrastructure money is going after the chip wars: even the company that built credible Nvidia-alternative silicon is now competing on inference economics, not transistor counts. It also marks the official end of Groq-the-chipmaker as an independent threat, and the beginning of a Groq-the-cloud era that has to win on price and latency against AWS, Azure, and dozens of GPU-rich startups.
Who is it for?
infrastructure investors, inference customers, AI hardware watchers