AI cloud operator Groq raises $350M more in funding
SiliconANGLE Maria Deutscher ● Covered by 2 sources
Groq just raised $350M more, less than three months after a $650M round. The bigger bet: its AI cloud is scaling fast, and Nvidia is tied into the story.
Based on reporting by SiliconANGLE, Maria Deutscher — 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 has added another $350 million to its war chest, with Disruptive leading the Series A round. Nvidia plans to join later, though Groq didn’t say how much the chip giant will put in.
The timing is striking. This new money lands less than three months after Groq closed a $650 million round, which tells you the company is still in heavy expansion mode rather than settling down to harvest what it already built.
Groq started in 2016 by working on AI accelerators. Then came last December’s Nvidia deal, where Nvidia licensed Groq’s chip technology and hired several people from Groq’s leadership team. After that, Groq shifted toward a public cloud built for AI workloads, using its earlier chip technology as the foundation.
That cloud, GroqCloud, now runs on hardware in 13 data centers around the world. The company says it offers bare-metal setups for customers who want to tune the hardware themselves, plus GroqStack for people who want the infrastructure handled for them. And Groq is broadening the pitch: it now says the platform can handle not just inference, but AI training too, even though the Groq 3 LPU itself is aimed at inference and is meant to work alongside Nvidia’s Rubin GPU.
The capacity plans are the real tell. Groq says it will use the new money to expand the cloud from 57 megawatts to more than 200 megawatts next year. That is not a modest tweak. It is a very loud bet that demand for AI compute will keep climbing, and that Groq wants to be one of the places people go when the GPUs are all spoken for.
My take — AI-written commentary, not fact-checked reporting
This is the old AI infrastructure playbook with shinier branding: raise a pile, lease a lot of power, and call it momentum. Groq is smart to hitch itself to Nvidia instead of pretending it can out-muscle Nvidia. The funny part is that everyone keeps talking about models, while the real money is still going into the pipes under them.
Read more about this at: SiliconANGLE