Groq Raises $350M at $3.5B Valuation, Nvidia Joins Round

Groq, the company that spent years building specialized AI inference chips, has raised $350 million in a new funding round at a $3.5 billion valuation, with planned participation from Nvidia, according to PYMNTS' reporting on the announcement. The round was led by Disruptive, the Dallas-based growth investment firm run by Groq's own executive chairman, Alex Davis.

The Nvidia connection here is genuinely unusual and traces back to a major restructuring that reshaped Groq entirely. In December 2025, Nvidia signed a $20 billion non-exclusive licensing deal for Groq's chip technology, and much of Groq's senior engineering team, including founder Jonathan Ross and president Sunny Madra, left to join Nvidia directly as part of that arrangement, according to PYMNTS' earlier coverage of the deal. Nvidia was explicit at the time that it hadn't acquired Groq outright, saying only that it had "taken a non-exclusive license to Groq's IP and have hired engineering talent from Groq's team."

Rather than folding entirely, Groq's remaining leadership, with CFO Simon Edwards stepping into the CEO role, pivoted the company's entire strategy: from building and selling AI chips to operating what's known as an inference "neocloud," a cloud service specialized specifically for running AI models rather than training them.

This $350 million round is Groq's third major capital raise in less than a year, and tracking that sequence reveals how quickly the company's business model has shifted. Groq raised $750 million at a $6.9 billion valuation in September 2025, while it was still primarily a chip company. Following the Nvidia licensing deal and team departure, Groq raised $650 million in June 2026 specifically to fund its pivot to an AI inference cloud, targeting scale toward 200 megawatts of capacity by 2027, according to Groq's own announcement of that round. Notably, this newest $350 million raise landed at a $3.5 billion valuation, roughly half of what the company commanded less than a year earlier, reflecting the genuine reset that followed losing its founding technical team and shifting business models entirely.

Alex Davis, Groq's executive chairman and CEO of lead investor Disruptive, framed the strategic bet behind the raise directly: "Inference will without a doubt become the largest and most critical layer of AI infrastructure." That distinction between training and inference matters enormously for understanding where AI infrastructure spending is actually headed. Training a large language model happens once or occasionally, but inference, the actual process of running a model to answer a user's request, happens every single time anyone interacts with an AI system, according to PYMNTS' broader analysis of the inference market. A single popular model might handle millions of inference requests monthly, each consuming compute, adding latency, and generating real operating costs.

Groq isn't alone chasing this specific opportunity. Cerebras, which also builds inference-specific chips, raised $5.5 billion in its IPO with shares doubling on debut, while Modal Labs raised $355 million at a $4.65 billion valuation after crossing $300 million in annualized revenue, according to WOWTALE's analysis of the broader inference cloud category. That competitive intensity connects directly to the broader multi-model, cost-conscious AI infrastructure shift we've tracked closely, including Stripe's $7 billion acquisition of AI model router OpenRouter announced the same week.

For businesses running AI-powered, customer-facing applications, inference performance directly affects user experience, system reliability, and operating costs, exactly the value proposition Groq and its inference-focused competitors are racing to own. Groq currently operates 13 data centers across North America, Europe, the Middle East, and Asia-Pacific, serving more than five million developers processing trillions of tokens weekly, according to Groq's own reporting on its June raise.

For companies evaluating AI infrastructure vendors, the rapid emergence of dedicated inference-cloud providers like Groq, Cerebras, and Modal Labs signals that businesses no longer need to rely solely on the same providers hosting their AI model training, opening genuine new options specifically optimized for the ongoing operational cost of running AI at scale.

What is AI inference, and why does it matter?
AI inference is the process of running an already-trained AI model to generate a response to a user's request, happening every time someone interacts with an AI system, unlike training, which typically happens only once or occasionally.

Why did Nvidia invest in Groq after taking its technology?
Nvidia signed a $20 billion non-exclusive licensing deal for Groq's chip technology in December 2025, absorbing much of its engineering team, but is now investing in the restructured company's new inference cloud business as a strategic partner.

How much funding has Groq raised in total this year?
Groq raised $750 million in September 2025 at a $6.9 billion valuation, $650 million in June 2026 to fund its inference cloud pivot, and this newest $350 million round at a $3.5 billion valuation.

Groq raised $350 million at a $3.5 billion valuation, with Nvidia participating in the round just months after signing a $20 billion licensing deal for Groq's chip technology and absorbing much of its engineering team. The raise funds Groq's continued pivot from building AI chips to operating an AI inference cloud, a business model shift executive chairman Alex Davis says positions the company in what will become "the largest and most critical layer of AI infrastructure." Groq now competes with Cerebras, Modal Labs, and other dedicated inference providers in a rapidly growing category separate from AI model training infrastructure.