AI chip startup Groq has acquired Definitive Intelligence for an undisclosed sum.

The acquisition will support the company’s newly formed GroqCloud unit, which will be led by Definitive Intelligence co-founder and CEO Sunny Madra. GroqCloud is a cloud platform that provides Groq hardware documentation, code samples, and on-demand access to the company’s AI chips.

groq chip
– Groq

Based in Palo Alto, Definitive Intelligence was founded in 2022 by Madra and former EMC director of engineering Gavin Sherry. The company provides AI-powered applications that enables users to query datasets with natural language instructions, and had raised $25.5 million in venture capital prior to its acquisition.

In a statement, Groq said that Madra and the GroqCloud team will initially focus on expanding capacity, improving efficiency, forming partnerships, and building out the company’s developer platform. The company is also formalizing a Groq Systems business unit, which will focus on providing its hardware to public sector organizations.

“At Groq, we’re committed to creating an AI economy that’s accessible and affordable for anyone with a brilliant idea,” said Groq founder and CEO Jonathan Ross.

“The Definitive team has expertise in AI solutions and go-to-market strategies, as well as a proven dedication to sharing knowledge with the community. They are the perfect fit to lead our GroqCloud business unit. Together, we will empower developers to leverage GroqCloud services to help accelerate the development of AI.”

Groq was co-founded in 2016 by Ross, who previously helped lead Google's Tensor Processing Unit development. The company provides dedicated chips and accelerators within its own servers and racks designed for artificial intelligence, machine learning, and high-performance computing.

The company’s flagship LPU (language processing unit) inference engine can reportedly run existing LLMs such as ChatGPT at 10x the speed of other available inference engines. In November, Groq’s LPU set an inference speed record while running Meta’s Llama 2 70B LLM during a benchmark test.