Groq just raised $650M after Nvidia stole its founder and its chip. Your inference bill is about to change.

Nvidia hired Groq's founder and president in December. Licensed Groq's chip technology for $20 billion. Then Groq's board did something no one expected.

They raised $650 million to become the thing Nvidia just helped kill.

Groq is no longer a chip company. It is an inference cloud. 13 data centers. Five million developers. Trillions of tokens per week. And a new CEO, CFO, COO, CTO, and CPO — none of whom were there six months ago.

Here is what your CTO needs to understand.

Inference is about to become 15 to 20 times larger than training as a compute market. Every model your company deploys needs to run somewhere. Today that somewhere is Nvidia GPU clusters at $40+ per million tokens.

Groq is betting that number drops to single digits. And they just got $650 million to prove it.

The implication for enterprise AI budgets is immediate. If inference costs drop 5x, the ROI math on every AI project in your pipeline changes. Projects that were marginal become obvious. Projects that were obvious become mandatory.

Audit your inference spend today. If you are locked into a single cloud provider for AI workloads, you are about to get a negotiating tool you did not have last quarter.

SOURCE: https://groq.com/newsroom/groq-raises-usd650m-to-scale-its-ai-inference-cloud-business
VERIFIED: Groq official newsroom (June 22, 2026), eWeek (June 23, 2026), The AI Insider (June 24, 2026)
SIGNAL: The inference cost curve is about to break. Enterprises locked into single-vendor AI infrastructure just got leverage.