Groq raises $650M, pivots to neocloud after Nvidia took its talent

Key Takeaways

• Groq raised $650M led by Disruptive and Infinitum, six months after Nvidia licensed its LPU technology and hired away founder Jonathan Ross
• The company has pivoted to its neocloud inference business, now operating 13 data centers serving 5 million developers
• Nvidia launched its own Groq 3 LPX inference hardware at GTC in March using the licensed IP

Groq, the AI chip company that watched Nvidia walk off with its founder and core IP last December, just raised $650 million. The funding round, led by Disruptive and hedge fund Infinitum, confirms that investors still see value in what's left. The question is whether a neocloud pivot can replace the hardware business that made Groq interesting in the first place.

The raise comes six months after one of AI's stranger transactions. Nvidia signed a non-exclusive licensing agreement for Groq's Language Processing Unit technology, then hired founder and CEO Jonathan Ross, president Sunny Madra, and other key employees. The deal reportedly valued the whole package around $20 billion. Groq's investors did well. The company itself was left to figure out what comes next.

What did Nvidia actually get from Groq?

Ross, who helped create Google's Tensor Processing Unit before founding Groq a decade ago, designed a chip architecture specifically for inference. Running trained AI models, not training them. The LPU approach delivered impressive speeds in public demos, generating hundreds of tokens per second on large language models.

Nvidia wasted no time putting the licensed IP to work. At its GTC conference in March, the company unveiled the Nvidia Groq 3 LPX inference hardware system. The naming alone signals how thoroughly Nvidia absorbed Groq's technology. The GPU giant now sells hardware built on the architecture that Groq's team spent years developing.

Who is running Groq now?

Doug Wightman, Ross's co-founder and fellow Google alum, stayed behind and became CEO. He's been rebuilding the executive team. Alan Rice joined as COO after stints at xAI and Meta, plus a career in the U.S. Navy. Sinclair Schuller took over as CTO, and Rakesh Malhotra as CPO. Schuller and Malhotra previously co-founded Nuvalence, a software engineering firm that EY acquired in 2024. Malhotra also spent about a decade on Microsoft's cloud products.

The new leadership team tilts heavily toward cloud and software experience. That's not accidental.

The neocloud pivot

Groq is betting its future on inference-as-a-service rather than chip sales. The neocloud business, originally run by Madra after Groq acquired his AI data analytics company Definitive Intelligence in 2024, has grown to 13 data centers across North America, Europe, the Middle East, and Asia-Pacific.

The company claims it now serves over five million developers and thousands of AI companies, processing trillions of tokens weekly. Those are substantial numbers if accurate. They suggest Groq built meaningful distribution before losing its hardware differentiation to Nvidia.

The catch: competing in inference cloud means competing against Nvidia's customers and partners, who now have access to the same underlying technology. AWS, Google Cloud, Microsoft Azure, and a growing list of specialized inference providers all want this market. Groq's speed advantage, once tied to proprietary silicon, becomes harder to defend when Nvidia sells equivalent hardware to everyone.

Can companies survive these deals?

The not-acqui-hire structure, where a big tech company pays investors handsomely while absorbing talent and IP without technically acquiring the company, has become common in AI. Groq isn't the only company navigating the aftermath.

Scale AI's CEO Jason Droege told Forbes that his company rebounded after Meta did a $14.3 billion not-acqui-hire about a year ago. Scale claims it's on track for $1 billion in revenue. That's an encouraging precedent, though Scale's data labeling business differs fundamentally from Groq's hardware-turned-cloud model.

Groq did not disclose its new valuation. The company was last valued at $6.9 billion after raising $750 million in September. Whether the $650 million came at a similar valuation, or reflects the reality of a company that just ceded its core technology, remains unclear.

Who funded the round?

Disruptive, a Dallas-based late-stage investment firm, led the round. The firm was founded by Alex Davis, who also serves as Groq's chairman. That dual role raises questions about whether this was truly arm's-length pricing. Infinitum, a Fort Lauderdale hedge fund, joined as co-lead.

The investor base appears smaller and more concentrated than Groq's previous rounds. When your chairman's firm leads your recovery financing, it's worth asking whether other investors passed.

Frequently Asked Questions

What happened between Groq and Nvidia?

In December, Nvidia signed a non-exclusive licensing agreement for Groq's LPU technology and hired founder Jonathan Ross, president Sunny Madra, and other key employees. The deal reportedly totaled around $20 billion.

What is Groq's LPU chip?

The Language Processing Unit is a chip architecture designed specifically for AI inference, meaning running trained models rather than training them. Groq claimed significant speed advantages over traditional GPUs for this workload.

How much did Groq raise in its new funding round?

Groq raised $650 million, led by Disruptive and Infinitum. The company did not disclose its new valuation.

Who runs Groq after the Nvidia deal?

Co-founder Doug Wightman became CEO. The company hired Alan Rice as COO, Sinclair Schuller as CTO, and Rakesh Malhotra as CPO.

What is Groq's neocloud business?

Groq's neocloud offers AI inference as a service through 13 data centers globally. The company says it serves over five million developers and processes trillions of tokens weekly.

Source: Startups | TechCrunch / Julie Bort