Groq lines up $650m for ‘neocloud’ spin-out after $20b Nvidia deal

Groq is raising up to $650 million from existing investors to fund a new “Groq2” entity, just months after inking a $20 billion licensing and asset deal with Nvidia that effectively rewired the AI chipmaker’s future.

Groq Inc. is raising as much as $650 million from its current backers to capitalize a new company that will focus on building AI “neoclouds,” following a landmark $20 billion licensing agreement under which Nvidia agreed to acquire Groq’s inference technology assets and hire much of its senior leadership. According to Axios, existing investors, including Disruptive and Infinitum, are prepared to cover the entire $650 million round if necessary, with participation offered pro rata to current shareholders in the successor Groq2 vehicle.

The fundraising comes on the heels of Nvidia’s largest-ever transaction, a roughly $20 billion cash deal for Groq’s AI inference stack that was structured as a non-exclusive technology license coupled with an asset sale.

As Groq explained in its own announcement, “Today Groq entered into a non-exclusive licensing agreement with Nvidia for Groq’s inference technology,” while emphasizing that it would “continue operating separately” and retain ownership of its intellectual property and cloud business.

Nvidia takes the silicon, Groq keeps the cloud

Nvidia’s agreement effectively lifts out Groq’s high-performance accelerator architecture and much of its leadership team, while leaving behind Groq’s neocloud platform and related customer contracts.
Groq founder and CEO Jonathan Ross, company president Sunny Madra and “almost everyone of note from Groq” are joining Nvidia to help integrate and scale the licensed inference technology, according to analyst Ian Cutress’s breakdown of the deal and Groq’s own blog language.

Groq’s statement stressed that the company will continue to run its cloud business independently, writing that GroqCloud “is not part of the transaction and will continue operating without disruption.”
Alex Davis, CEO of Austin-based investment firm Disruptive, which led Groq’s prior funding round, told CNBC that Nvidia is “acquiring all of Groq’s assets, with one major exception: Groq’s cloud computing business,” a carve-out that now appears to be the core of the Groq2 strategy.

The new Groq2 entity, as described by Axios, will shift definitively away from chip manufacturing and toward building “AI neoclouds,” a term Groq has used for tightly optimized inference clouds targeting real-time workloads rather than general-purpose infrastructure. That pivot puts Groq2 directly into competition with hyperscalers and emergent AI hosting providers that sit on top of Nvidia hardware rather than designing their own silicon, effectively turning Groq from a challenger chip designer into a vertically focused AI services company riding on Nvidia’s platform.

Second act bets on ‘neoclouds’

For Nvidia, the $20 billion Groq package is part of a broader push to entrench its dominance in AI inference and data center acceleration, after reporting quarterly revenue of $57 billion in 2025 and becoming the central supplier to the current AI build-out. For Groq’s remaining business, the $650 million raise is a survival-and-scale bet that its neocloud model can attract enough demand to justify a standalone platform once its chip design talent and physical assets migrate into Nvidia’s orbit.

Davis, whose firm also helped engineer the Nvidia–Groq transaction, has warned that a “build it and they will come” approach to AI data centers could lead to “a significant financing crisis” for speculative operators as early as 2027 or 2028, highlighting the risk profile for Groq2 and similar ventures. Against that backdrop, locking in up to $650 million of committed capital from existing investors gives Groq2 a substantial runway to prove out its neocloud thesis in a market already dominated by hyperscalers and Nvidia-powered incumbents.