From $850M Entry to a $20B Strategic Exit
Specialized Chips for AI Inference
Groq is a semiconductor and systems company that has built purpose-designed processors for AI inference workloads, particularly large language models. The company's Language Processing Unit (LPU) architecture achieves dramatically faster token generation speeds compared to GPUs, enabling real-time AI applications that weren't previously feasible. Groq's technology focuses on deterministic performance and low latency rather than training, making it ideal for production deployment of AI models in applications requiring instant responses. The company offers both hardware systems and cloud inference services to developers and enterprises.
.png)
From $850M Entry to a $20B Strategic Exit
We invested in Groq's Series C in 2021, recognizing the industry shift from model training to inference deployment. In early 2026, Groq’s technical dominance was validated by a landmark $20B strategic acquisition by Nvidia, one of the largest semiconductor exits in history, validating the LPU as foundational infrastructure for the real-time AI economy.
.jpg)