AI chipmaker Groq is looking to raise as much as $650 million in fresh capital from existing investors as the company pivots from selling its homegrown AI hardware to running an inference cloud business, according to Axios and TechCrunch reports published May 28 and 29.
A New Direction After the Nvidia Deal
The fundraise comes roughly six months after Groq struck a $20 billion licensing-and-talent agreement with Nvidia, often described as a "not-acqui-hire," that sent senior Groq engineers to the GPU giant and put the startup's hardware technology in Nvidia's hands. The original investors were paid out in cash through that deal, the largest such arrangement Nvidia has ever announced.
The new round, instead, is meant to bankroll the next chapter: a Groq 2.0 led by interim CEO Adam Winter and CFO Matt Eng. The leaders are doubling down on inference, the work of running already-trained AI models in production rather than training new ones, which has become the dominant compute workload across enterprise AI deployments.
Why Inference Is the New Battleground
Inference demand has exploded as developers ship copilots, agents and generative video apps that hit models thousands of times per session. Groq plans to operate what investors call a "neocloud," hosting customers' AI workloads on inference-optimized infrastructure that it builds itself. The company is competing with the likes of CoreWeave, Together AI and Lambda, as well as the hyperscalers, in a market where capacity is the gating factor.
Other AI inference players have also drawn fresh checks this month, including OpenRouter's $113M Series B and Tensormesh's $20M raise to cut inference costs through KV caching.
A Guaranteed Round
According to Axios, Groq's longtime backers Disruptive and Infinitium have agreed to backstop the round if other existing investors decline to take their pro-rata shares, essentially guaranteeing that the $650 million will close. That structure reduces execution risk for management as it rebuilds the engineering bench and reorients the product roadmap toward inference services.
Pivots in a Crowded AI Funding Market
The Groq round arrives during one of the busiest weeks ever for AI fundraising. Anthropic announced a $65 billion Series H at a $965 billion valuation, and Cognition closed roughly $1 billion at a $26 billion valuation, both centered on enterprise AI demand. Investors are increasingly distinguishing between speculative model labs and infrastructure businesses with clearly billable usage, a category Groq is now positioning itself within.
What's Next
Groq has not publicly confirmed the terms of the new round, and Axios noted that final pricing and structure could still change. If completed, the $650 million would give Groq the runway to expand data center capacity, refine its custom chip-and-systems stack, and chase enterprise inference contracts as customers shift workloads off general-purpose GPUs.
Reporting based on coverage from TechCrunch and Axios.