Groq's $650 Million Comeback and the Infrastructure Behind Your AI Inference
Groq confirmed a $650 million funding round on June 22, 2026. The raise comes roughly nine months after the company last closed funding at a $6.9 billion valuation following a $750 million round in September 2025.
What Actually Happened to Groq First
In December 2025, Nvidia signed a non-exclusive licensing agreement for Groq's LPU (language processing unit) technology. Then Nvidia hired Groq's founder and CEO Jonathan Ross, president Sunny Madra, and other key employees.
Ross had previously co-created Google's Tensor Processing Unit before founding Groq. Nvidia got the IP license and the people who built it. In March 2026, Nvidia announced the Nvidia Groq 3 LPX inference hardware system at its GTC event.
That's the company now raising $650 million.
The New Groq
After the Nvidia deal, Groq pivoted to a neocloud business model. Doug Wightman, a Groq co-founder, became CEO. The company brought in Alan Rice from xAI and Meta as COO, Sinclair Schuller as CTO, and Rakesh Malhotra as CPO.
That's a complete executive team rebuild following the departure of the founding leadership to Nvidia.
What They're Actually Running
Groq operates 13 data centers across North America, Europe, the Middle East, and APAC. The company serves over 5 million developers and thousands of AI companies. It processes trillions of tokens per week across its inference cloud.
In 2024, Groq acquired AI data analytics company Definitive Intelligence. The LPU chip was originally designed specifically for AI inference workloads, sold as a cloud service or on-premises hardware cluster.
A Familiar Pattern
The Groq situation resembles a structure that has appeared elsewhere in AI. Meta completed a $14.3 billion not-acqui-hire deal with Scale AI approximately one year ago. Scale AI is on track to reach $1 billion in revenue following that deal, and it continues operating independently.
In both cases, a major incumbent acquired technology and talent from a smaller AI infrastructure company. The smaller company continued. Investors continued funding it.
The Actual Tension
Groq's LPU was purpose-built for inference. Nvidia licensing that technology and hiring the people who designed it suggests the approach had real merit. The $650 million raise suggests investors believe the neocloud business built around that technology has a viable path forward.
The question Groq now has to answer: can it compete in the inference cloud market while Nvidia sells hardware based on Groq-licensed technology? The new executive team has the next phase to prove it.
Source: Techcrunch