The company founded by a former Google TPU lead is scaling its cloud services and next-gen chip development to challenge Nvidia's dominance.
Groq, the AI inference chip startup that’s been turning heads with its Language Processing Unit technology, is raising $650 million to fund its next phase of growth.
For context, Groq closed a $640 million Series D in August 2024 at a $2.8 billion valuation, with BlackRock funds leading the charge. Less than a year later, the company is back at the fundraising table, this time at a valuation that could reach roughly $6 billion, more than doubling its worth in under twelve months.
What Groq actually does, and why investors keep writing checks #
Groq’s LPU architecture is purpose-built for inference workloads. Founded in 2016 by Jonathan Ross, the former lead of Google’s Tensor Processing Unit division, Groq has positioned itself as a direct alternative to Nvidia’s GPU-centric approach.
The new capital is earmarked for expanding GroqCloud inference capacity and developing the company’s next-generation LPU technology. Disruptive Capital is reportedly leading the round, with participation from BlackRock and Neuberger Berman among others.
The bigger picture: Groq’s meteoric rise #
Perhaps most telling is a reported deal with Nvidia in December 2025, valued at approximately $20 billion, involving technology licensing and key personnel.
What this means for investors and the AI hardware market #
Groq’s rapid valuation growth from $2.8 billion to a potential $6 billion in roughly a year puts it in the conversation alongside other well-funded AI chip companies.
The involvement of firms like BlackRock and Neuberger Berman across multiple rounds suggests durable institutional conviction. These are institutions that typically invest with multi-year horizons, and their repeated participation indicates they see Groq as a durable player rather than a hype-cycle beneficiary.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our