Groq Raises $350M at $3.5B to Expand Its Inference Cloud With Nvidia Systems
August 17, 2026
Groq, once known for building its own inference chips, has raised $350 million at a $3.5 billion valuation to fund an inference cloud business that is adding Nvidia’s computing platform.
TechCrunch reported a $6.9 billion valuation for Groq in September 2025. The lower number is real, and so is the major shift in strategic focus the company made in between.
Groq set out to beat Nvidia on inference silicon. It now runs an AI inference cloud that is adding Nvidia’s computing platform to its data centers, and this round funds that shift.
Inside the $350 Million Round
The Series A, announced on August 17, 2026, was led by the investment firm Disruptive, with planned participation from Nvidia. Together with $650 million raised in June 2026, recent funding reaches $1 billion. The round remains subject to customary closing conditions.
Groq says it operates 13 data centers across North America, Europe, the Middle East, and Asia Pacific, serving more than six million developers, Fortune 500 enterprises, and thousands of AI-native companies. It expects to scale from 54 megawatts to more than 200 megawatts in 2027.
Groq is a certified Nvidia Cloud Partner, qualified to design, deploy, and operate Nvidia accelerated computing to Nvidia’s reference architecture and operational standards for customers seeking medium and larger clusters.
How a Chip Challenger Moved Into Nvidia’s Ecosystem
The turn came on December 24, 2025, when Groq announced a non-exclusive inference technology licensing agreement with Nvidia. Founder and CEO Jonathan Ross, president Sunny Madra, and other leaders moved to Nvidia. No price appears in the announcement; TechCrunch reported the deal at about $20 billion.
Groq continued as an independent company, with Simon Edwards stepping in as CEO at the time. Its strategic weight shifted from competing on LPUs (language processing units), the custom inference chips it designed, to an inference cloud business that also incorporates Nvidia systems.
Whether to run on a proprietary compute stack or fold external platforms into a service business is now a capital question as much as a technical one. That shift stands in contrast to Tesla and SpaceX investing over $16.8 billion in the Terafab chip plant.
The Economics of Inference Clouds Are Still Opaque
Emerging cloud providers that specialize in AI compute infrastructure such as GPUs are known as neoclouds. Inference demand is growing, but whether the business returns cash is a separate question.
On CoreWeave, TechCrunch reported strong second-quarter revenue growth and major contracts including Meta and Anthropic, alongside investor concern about heavy capital expenditure, reliance on debt, and exposure to hardware that depreciates quickly.
Groq’s financials are private, so outsiders have little to test against. Nvidia supplies GPUs to CoreWeave, Lambda, and Nebius while also investing in some of them, and its planned participation here places Groq inside that circle.
What to Watch Next
No completion date has been given for the closing. The 2027 capacity target comes without a breakdown by site.
With financials undisclosed, there is no external way to judge whether the inference cloud model pays. The next signal is how much Groq reveals about contract scale and utilization.
Sources
Groq – Groq Closes $350 million Series A, Building the World’s Leading AI Inference Cloud
TechCrunch – Groq raises $350M to fuel its pivot from AI chips to neocloud
