NVIDIA-backed Groq raises $350M at $3.5B as AI inference race accelerates
Tech Funding News Abhinaya Prabhu
Groq just raised $350M at $3.5B – half what it was worth last year. Nvidia hired away its founder and team, then decided to invest anyway.
Based on reporting by Tech Funding News, Abhinaya Prabhu — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Groq closed a $350 million Series A at a $3.5 billion valuation this week, led by Disruptive with Nvidia expected to join in. That number is exactly half the $6.9 billion the company was worth back in September 2025. The round pushes Groq's total fundraising since June 2026 to roughly $1 billion, following a $650 million raise just two months earlier. Demand for AI inference is clearly real. But the price Groq is paying to keep chasing it tells its own story.
Jonathan Ross, a former Google engineer who worked on the company's Tensor Processing Unit, founded Groq back in 2016. For nearly a decade the company built its case around Language Processing Units, pitched as a faster, cheaper way to run AI models than Nvidia's GPUs. That bet seemed to be paying off in September 2025, when Groq raised $750 million at that $6.9 billion valuation, with BlackRock, Neuberger Berman, and Deutsche Telekom Capital Partners joining Disruptive on the cap table.
Then came December 2025. Nvidia agreed to pay a reported $20 billion to license Groq's inference technology and hire away Ross, president Sunny Madra, and around 90% of the engineering team that had actually built the product. Groq kept the lights on, minus the people who invented what made it worth funding in the first place.
What's left is still substantial on paper — 13 data centres, more than six million developers, and a stated plan to scale capacity from 54MW to over 200MW by 2027. Alex Davis, Groq's executive chairman and the founder of Disruptive, framed the company's ambitions plainly: inference, he said, will become the largest and most critical layer of AI infrastructure. He's not wrong about the trend. He's just no longer the only one making that argument with serious money behind it.
Together AI raised $800 million at an $8.3 billion valuation in July 2026 to build its own inference cloud around open models. Fireworks AI went bigger still, closing a $1.5 billion Series D at a $17.5 billion valuation the same month, with Nvidia among its backers too. Groq now competes less on technology, since its core edge sits inside Nvidia already, and more on scale and its existing developer footprint. It remains a certified Nvidia Cloud Partner running Nvidia's own reference architecture — a relationship that looks a lot more like a supply agreement than a rivalry.
Whether Groq can rebuild something defensible after licensing away the tech that made it distinctive is the open question. The answer decides whether Nvidia's money turns Groq into a real platform, or just a well-funded, dependent supplier of the same compute Nvidia already sells everyone else.
My take — AI-written commentary, not fact-checked reporting
Nvidia didn't just outcompete Groq — it bought the team that made Groq worth funding, then turned around and reinvested in whatever was left standing. Calling this an independent inference challenger is generous; it reads more like a supply agreement wearing a funding round as a costume. Chip startups that set out to break Nvidia's grip keep ending up back inside it, and if that pattern holds, the inference race everyone's excited about might just be Nvidia financing its own supply chain under different logos.
Read more about this at: Tech Funding News