The integration of a new AI inference optimization tool will allow Nebius to serve a larger volume of users with its existing graphics processing unit cluster. The company is absorbing the team and technology of Inferize to minimize model "cold starts," which reduces the amount of time GPUs remain idle while loading and spinning up new capacity on its Token Factory platform.

While the purchase terms remain officially undisclosed, the deal is estimated at $100M to $150M. The acquisition aims to lower the cost per token for users and increase overall capacity utilization to handle growing demand for AI inference from the same underlying infrastructure.