Groq, the AI chip company, is raising up to $650 million from existing investors as of May 28, 2026, despite having completed a $20 billion technology licensing deal with Nvidia in December 2025. The apparently paradoxical fundraising reveals the complexity of AI infrastructure business models and highlights the strategic value of operational datacenter assets separate from chip technology.
The Reality Behind the Nvidia Deal
Nvidia did not fully acquire Groq despite widespread reporting of an "exit." Instead, the December 2025 deal involved a technology license and hiring of key technical staff, including Groq Founder Jonathan Ross and other senior executives. As analyst Zach Stein-Perlman clarified, "Nvidia did not acquire Groq. They licensed Groq's technology and hired Groq's key technical executives, but did not acquire the Groq corporate entity."
The remaining Groq company continued operating as an independent entity, maintaining datacenter operations and inference services through GroqCloud. This structure allowed Nvidia to gain access to Groq's chip technology while leaving the operational infrastructure business intact.
Datacenter Assets Drive Valuation
Groq operates four functional datacenters designed specifically for AI inference workloads. Building new datacenters faces extreme challenges due to power availability, regulatory hurdles, and expertise requirements, making existing operational facilities strategically valuable. The company is transitioning from AI chipmaker to AI inference neocloud business, leveraging these physical assets.
Market comparisons suggest substantial valuation potential for datacenter operators:
- CoreWeave operates 43 datacenters at a $50 billion valuation
- Nebius maintains 11 larger facilities at similar valuation levels
Groq's previous Series D round in August 2024 raised $640 million led by BlackRock Private Equity Partners, valuing the company at $2.8 billion. Investors included Cisco Investments, Samsung Catalyst Fund, Type One Ventures, KDDI, and Valour Capital.
Strategic Challenges and Market Position
Groq faces significant headwinds despite its datacenter infrastructure:
- Aging Hardware: Groq's datacenters use seven-year-old LPUv1 chips, potentially limiting performance competitiveness
- Lost Differentiation: Nvidia now sells newer LPUv3 chips commercially, eliminating Groq's technological exclusivity
- Brand Risk: Association with high-speed, high-cost inference may limit appeal as major organizations like Microsoft and Uber increasingly prioritize cost efficiency
The Hacker News discussion of the article "How is Groq raising more money?" generated 137 points and 59 comments on June 2, 2026, indicating significant community interest in understanding the company's unusual fundraising situation.
Key Takeaways
- Groq is raising up to $650 million as of May 28, 2026, despite a $20 billion Nvidia deal in December 2025 that licensed technology and hired key executives but did not acquire the corporate entity
- The company operates four functional AI inference datacenters and is transitioning from chipmaker to AI inference neocloud business
- Groq previously raised $640 million in Series D funding at a $2.8 billion valuation in August 2024, led by BlackRock Private Equity Partners
- Strategic challenges include aging seven-year-old LPUv1 chips and lost technological differentiation after Nvidia began selling newer LPUv3 chips commercially
- The situation illustrates how hardware, software, and operational infrastructure assets can be unbundled and valued separately in AI markets