AI
AI chipmaker Groq raises $650M after Nvidia's $20B licensing deal and talent exodus
Six months after losing its founder and key staff to Nvidia, the inference hardware startup has pivoted to its cloud infrastructure business and secured new funding with fresh leadership.
TechCrunch — AI
Groq announced a $650 million funding round on Monday, roughly half a year after Nvidia signed a non-exclusive licensing agreement for the AI chipmaker's technology and hired away founder Jonathan Ross, president Sunny Madra, and other critical staff. The round was led by Disruptive and Infinitum hedge fund.
The Nvidia Deal's Ripple Effect#
The December deal saw Nvidia acquire exclusive rights to Groq's language processing unit (LPU) technology and unveiled its own Nvidia Groq 3 LPX inference hardware at its March GTC event. For Groq's investors, who had valued the company at $6.9 billion after a $750 million September round, the licensing agreement proved profitable—though the company did not disclose its new valuation.
Ross, who created Google's Tensor Processing Unit before co-founding Groq a decade ago, moved to Nvidia's team. His replacement as CEO is Doug Wightman, who previously served as president and stayed on after the deal.
Pivot to Cloud Infrastructure#
Rather than compete directly with Nvidia on hardware, Groq has shifted focus to its neocloud business—an inference cloud service that grew out of its 2024 acquisition of Definitive Intelligence (formerly run by Madra). The platform now operates 13 data centers globally and serves over five million developers, the company claims.
The company has also rebuilt its leadership with Alan Rice as COO (formerly at xAI and Meta) and a pair of new executives from the enterprise software world: Sinclair Schuller as CTO and Rakesh Malhotra as CPO, both previously involved with Nuvalence before its 2024 acquisition by EY.
A Familiar Playbook?#
Groq's recovery strategy mirrors Scale AI's experience after Meta's $14.3 billion not-acqui-hire deal roughly a year ago—proving that startups can survive losing core talent and IP to larger rivals. However, the inference cloud space is heating up, with intense competition and rapidly evolving technology making long-term success uncertain.
Read the full story at TechCrunch.
