Groq raises $350M at a $3.5B valuation to scale its Nvidia-powered cloud

Groq has raised $350 million in a round led by the investment firm Disruptive, with planned participation from Nvidia, at a $3.5 billion valuation, according to TechCrunch. The number sits below the $6.9 billion Groq carried last September, a few months before Nvidia hired founder and CEO Jonathan Ross and other top talent as part of a $20 billion licensing deal it paid out to the startup's investors.

A company spokesperson told TechCrunch that the lower figure is not seen as a down round. The company treats it as a new valuation for what it calls the "post-Nvidia-licensing-deal version of Groq."

How a chipmaker became a neocloud

Before the talent exit, Groq built its own silicon. The company designed chips it called LPUs, language processing units, aimed at inference, the compute needed to run AI workloads in real time, in direct competition with Nvidia. After Ross and the rest of its star team moved to Nvidia, Groq shifted from a pure AI chipmaker to a cloud and data center provider that operates Nvidia hardware. That is the neocloud model: buy accelerators at scale and sell access to them.

The pivot started in June with a $650 million round. Groq now operates 13 data centers across North America, Europe, the Middle East, and Asia Pacific, serving more than 6 million developers, enterprises, and AI-native companies, according to the company. Its stated plan is to grow capacity from 54 megawatts to more than 200 megawatts in 2027, roughly a fourfold increase in power footprint.

The fresh capital goes in the same direction. Groq says the funds will support "those seeking usage of medium and larger sized clusters of Nvidia accelerated computing for training and inference."

What is still unverified

Alex Davis, Groq's chairman and CEO of Disruptive, laid out the goal in a statement: "We are building Groq into the world's leading AI inference cloud. Inference will without a doubt become the largest and most critical layer of AI infrastructure." That is the ambition the round is financing.

Whether neoclouds can make the economics work is an open question. CoreWeave, the most visible example of the model, reported strong second-quarter revenue growth and recently landed major contracts, including deals with Meta and Anthropic. Investors remain concerned about its high capital expenditures, its reliance on debt, and its exposure to rapidly depreciating hardware, and about whether growth converts into free cash flow. Groq's financials are still private, so the comparison has to stay at the level of business models.

What the pivot makes clear is the direction of the dependency. Groq now sits inside Nvidia's AI infrastructure ecosystem, and it is not alone there: Nvidia supplies the GPUs that power clouds from CoreWeave, Lambda, and Nebius, while investing billions into some of those companies as they build capacity. A neocloud's revenue depends on hardware that its largest investor and main supplier sells.

For a team buying inference capacity today, the round changes little in the near term beyond what Groq has already stated: the number to watch is the 200 megawatt target for 2027, which is where the added serving capacity actually lands.