Nvidia has entered into a landmark $20 billion cash agreement to acquire key assets from Groq, a rapidly growing artificial intelligence chip startup focused on high-speed inference computing. The transaction, the largest in Nvidia’s history, signals a major escalation in the company’s strategy to strengthen its position across every layer of the AI hardware stack as global demand for AI computing continues to accelerate.
The deal was disclosed by Alex Davis, chief executive of Disruptive, the venture firm that led Groq’s most recent funding round. According to Davis, Nvidia moved decisively to secure the agreement shortly after Groq completed a major capital raise, despite the startup not actively seeking a buyer at the time.
A Startup Fresh Off a Major Fundraising
Groq raised $750 million in September at a valuation of roughly $6.9 billion, drawing interest from some of the world’s largest investment firms and technology companies. Participants in the round included BlackRock, Neuberger Berman, Samsung, Cisco, Altimeter, and 1789 Capital. Disruptive has backed Groq since its founding in 2016 and has invested more than $500 million in the company over that period.
The timing of Nvidia’s move underscores how quickly competitive pressures are reshaping the AI semiconductor industry. Rather than allowing Groq to continue scaling independently, Nvidia opted to fold the startup’s core technology and talent into its own ecosystem through an asset-based transaction.
Licensing Structure Instead of a Full Buyout
Rather than acquiring Groq outright, Nvidia structured the deal as an asset purchase combined with a non-exclusive licensing agreement covering Groq’s inference technology. The companies have not disclosed financial details of the licensing arrangement, but the structure allows Nvidia to integrate Groq’s designs into its own platforms while leaving Groq legally independent.
As part of the transition, Groq founder Jonathan Ross, company president Sunny Madra, and other senior leaders will move to Nvidia, where they are expected to focus on scaling and commercializing the licensed technology. Groq will continue operating independently under new leadership, with finance chief Simon Edwards stepping into the role of chief executive.
The company has emphasized that the leadership change does not mark an end to Groq’s operations, but rather a reorganization following the transfer of its core assets.
GroqCloud Remains Independent
While Nvidia is acquiring most of Groq’s assets, the startup’s cloud-based offering is excluded from the deal. GroqCloud, which provides developers with access to Groq’s hardware through a managed platform, will continue operating separately and without interruption.
This carve-out highlights Nvidia’s focus on hardware architecture, intellectual property, and engineering expertise rather than customer-facing cloud services. By leaving GroqCloud intact, the transaction avoids potential conflicts with Nvidia’s own partners and customers in the cloud computing market.
Nvidia’s Financial Firepower on Display
The $20 billion deal represents a dramatic increase in the scale of Nvidia’s transactions. The company’s largest previous acquisition was the $7 billion purchase of networking firm Mellanox in 2019. Since then, Nvidia’s financial position has transformed, largely driven by explosive demand for its AI accelerators.
By the end of October, Nvidia reported more than $60 billion in cash and short-term investments, a sharp rise from just over $13 billion at the start of 2023. That expanding war chest has enabled Nvidia to pursue aggressive investments, partnerships, and acquisitions across the AI ecosystem.
A Pattern of Talent-Driven Expansion
The Groq agreement follows a growing pattern in Nvidia’s strategy: acquiring teams and technology through licensing deals rather than traditional takeovers. Earlier this year, Nvidia spent more than $900 million to bring in leadership and staff from AI hardware startup Enfabrica, while also licensing its technology.
This approach mirrors similar moves by other technology giants, including Meta, Google, and Microsoft, all of which have increasingly used licensing and talent acquisitions to accelerate AI development while avoiding lengthy regulatory scrutiny.
Broad Investments Across the AI Landscape
Beyond Groq, Nvidia has steadily expanded its influence across the broader AI infrastructure market. The company has invested in energy and data center operator Crusoe, AI model developer Cohere, and cloud provider CoreWeave, which has been preparing for a public listing.
In September, Nvidia also announced plans to make a massive investment in OpenAI, potentially committing up to $100 billion, contingent on the deployment of large-scale Nvidia-powered infrastructure. That same month, Nvidia revealed a $5 billion investment in Intel as part of a broader strategic partnership, signaling a willingness to collaborate even with long-standing competitors.
Groq’s Role in the AI Inference Race
Groq carved out a niche by focusing on inference, the process of running trained AI models in real-world applications. Unlike training, which requires massive computational resources upfront, inference emphasizes speed, efficiency, and low latency—qualities increasingly critical as AI tools move into consumer products, enterprise software, and real-time systems.
The company had been targeting annual revenue of $500 million, reflecting strong demand for specialized inference accelerators. Groq’s technical roots trace back to Google’s early custom AI hardware efforts. Founder Jonathan Ross previously helped design Google’s tensor processing unit, one of the first major alternatives to Nvidia’s GPUs.
