The 17 person Israeli startup provides optimization tools that minimize model cold starts to ensure graphics processing units do not sit idle during demand surges. Nebius has acquired the company for an estimated $100M to $150M to integrate its technology into the Token Factory inference platform, allowing the company to serve more requests and lower the cost per token using existing infrastructure.

Founded 10 months ago by former members of Intel-acquired Granulate, Inferize is part of a vertical integration strategy by Nebius to control the software stack between hardware and applications. This approach includes the previous acquisition of Eigen AI for model optimization, Clarifai's orchestration technology, and Tavily for agentic search to maximize the number of useful tokens produced per GPU.