Groq raises $350 million, drops in-house chip plans to become an NVIDIA customer

A company that once built chips now leases someone else's

Groq was originally built to go head-to-head with NVIDIA. Its founding premise was to shake up the AI inference market — running already-trained models in real time — using its own custom LPU (Language Processing Unit) chips. That same Groq has now raised $350 million in new funding. The round was led by investment firm Disruptive, with NVIDIA reportedly set to participate as well. The valuation assigned was $3.5 billion, half of the $6.9 billion figure from last September.

The company says it doesn't view this as a down round. A Groq spokesperson told TechCrunch that the new valuation reflects "the version of Groq after the NVIDIA licensing deal" — signaling that the company's identity itself has fundamentally changed over the past few months.

The seat NVIDIA took from the founder

A few months after that $6.9 billion valuation in September, NVIDIA brought over Groq's founder and CEO Jonathan Ross along with key team members through a $20 billion licensing deal — effectively absorbing the brain trust of a company that had once positioned itself as a direct rival. Having lost its star team, Groq abandoned its custom chip development path and pivoted to becoming a cloud and data center operator running NVIDIA systems. What remains of Groq today is no longer NVIDIA's competitor, but its customer.

This transition actually began back in June, when Groq raised $650 million to take its first step into the pivot. The new $350 million round builds on that momentum, with the company saying the funds will support "demand to use large and mid-sized NVIDIA-accelerated computing clusters for training and inference."

Toward 13 data centers and 200 megawatts

Groq currently operates 13 data centers across North America, Europe, the Middle East, and Asia-Pacific, serving more than 6 million developers, enterprises, and AI-native startups. The company aims to expand its power capacity from the current 54 megawatts to over 200 megawatts by 2027. Megawatts serve as a measure of how many GPU servers a data center can run simultaneously, meaning Groq's infrastructure footprint would grow nearly fourfold within three years under this target.

Groq Chairman Alex Davies, who also serves as head of Disruptive, said in a statement, "We are building Groq into the world's leading AI inference cloud." His view is that inference will become the largest and most critical layer of AI infrastructure.

Neoclouds: everyone's doing it, everyone's worried about it

The business of leasing NVIDIA GPUs to sell AI infrastructure is commonly referred to in the industry as a "neocloud." Groq is far from alone in making this choice. NVIDIA supplies GPUs to companies like CoreWeave, Lambda, and Nebius while also investing billions of dollars into them — playing the dual role of supplier and investor, helping companies that use its GPUs grow in scale.

The question is whether this business can turn a profit in the long run. CoreWeave recently reported strong revenue growth in the second quarter and landed major contracts with Meta and Anthropic, but investors continue to flag concerns about the company's heavy capital expenditure burden, reliance on debt, and the risk of rapidly depreciating hardware. Whether growth can be converted into actual cash flow remains a challenge for the neocloud model as a whole. Groq's financial details remain undisclosed, but through this pivot, the company has now fully embedded itself within NVIDIA's ecosystem.

Editor's take

What's more interesting about this Groq round than the halved valuation is the story behind it. A company that set out to beat NVIDIA with its own chips lost its founder to NVIDIA — and was reborn as a business that survives by selling NVIDIA's GPUs. It's a rare, almost literal enactment of Silicon Valley's well-worn line: "If you can't beat them, join them." The company's insistence that this isn't a down round reads like a defense conscious of exactly this narrative.

Looking at the trajectory of inference chip startups over the past year or two, the outcomes converge on a similar pattern. Whether it's Cerebras or Groq, attempts to challenge NVIDIA's CUDA ecosystem with proprietary architectures have lost momentum against the wall of software compatibility and developer inertia. Leasing out NVIDIA GPUs directly, by contrast, has a lower barrier to entry but thinner margins. Groq's pivot illustrates what the most realistic alternative looks like once that wall proves too high to clear.

For teams in Korea preparing AI semiconductor or infrastructure ventures, the lesson here is clear. Competing with a proprietary chip means budgeting far more time than expected for building out the software stack and developer community alongside it. If that time can't be bought, pivoting to a business that leases someone else's chips isn't a shameful fallback — it's a survival strategy. But as CoreWeave's case shows, even that business only remains sustainable if cash flow can keep pace with capital expenditure.

In the coming weeks, Groq's next moves are likely to include announcements of data center expansions and new client disclosures. At the same time, competition on price and capacity will intensify with other NVIDIA-backed neoclouds, and whoever reports profitability first will become the test case that determines how much credibility the neocloud model holds overall.