Groq, a company specializing in artificial intelligence inference chips, has secured $650 million in new funding. The round was led by Disruptive and Infinitum, with participation from other investors. This capital infusion is intended to support the expansion of Groq's AI inference cloud services.
The company's focus on inference, the process of generating real-time responses from trained AI models, positions it within a critical segment of the AI infrastructure market. Groq's proprietary Language Processing Unit (LPU) is designed to offer high-speed, low-latency performance for AI workloads, differentiating it from traditional graphics processing units (GPUs). The LPU architecture emphasizes on-chip SRAM memory and a deterministic processing design to minimize delays in AI computations.
This latest funding follows Groq's substantial non-exclusive licensing agreement with NVIDIA, signed in late 2025. This deal for Groq's inference technology is reportedly valued at approximately $20 billion. Prior to this new funding, Groq had raised $750 million in September 2025, which valued the company at $6.9 billion. The company's Series D funding in August 2024 secured $640 million at a $2.8 billion valuation, led by BlackRock Private Equity Partners.
Groq was founded in 2016 by former Google engineers, including Jonathan Ross, who was involved in the development of Google's Tensor Processing Unit (TPU). The company's mission is to reduce the latency in AI interactions, making them feel instantaneous. Groq has built a cloud platform, GroqCloud, which hosts over 360,000 developers building applications on various open models.
The company plans to scale its capacity towards 200 megawatts by the end of 2027. Groq has also focused on integrating its hardware with software, aiming to provide a more complete solution for enterprises. This approach has seen Groq transition from being solely a hardware provider to offering a comprehensive solution vendor, with its GroqCloud tokens-as-a-service model generating recurring revenue.
Groq's technology is designed to accelerate AI inference for various applications, including large language models (LLMs), image classification, and predictive analysis. The company's LPU architecture is built on a software-first principle, with a custom compiler designed to optimize hardware utilization. This design allows for deterministic execution, providing predictable performance without run-to-run variation.
The company's strategy includes scaling manufacturing partnerships, expanding its engineering team, and building strategic alliances with cloud providers and enterprise customers. Groq has also entered into a licensing agreement with Nvidia Corp. for its inference technology, reportedly valued at approximately $20 billion. This agreement involves significant cash payments and a transfer of Groq executives to Nvidia.