Nebius acquires Inferize to enhance Token Factory platform

Nebius: Nebius is an Amsterdam-headquartered AI cloud infrastructure company that designs and operates its own data centers and servers across Europe and the US, offering GPU compute, storage, networking, and managed services for AI workloads. It provides a full-stack platform for model training, fine-tuning, and inference targeted at developers and enterprises building AI applications. Nebius is acquiring Inferize to integrate its optimization technology and team directly into the Token Factory platform to enhance inference performance. Inferize: Inferize is an AI inference optimization startup focused on technologies that address inefficiencies in model serving. Its solutions target reductions in model cold starts to minimize GPU idle time during loading and new capacity activation. Nebius has acquired Inferize to incorporate its technology and team into Token Factory for improved GPU utilization and cost efficiency in inference operations. Token Factory: Token Factory is Nebius's managed inference platform that enables testing, deployment, and scaling of open-source AI models via an OpenAI-compatible API for generative tasks. It supports various model types including text-to-text, embeddings, and vision, with options for serverless and dedicated endpoints along with built-in optimizations. Nebius plans to integrate Inferize's technology into Token Factory to further reduce inference costs and increase capacity from existing infrastructure. Platform Expansion: Companies building AI infrastructure are integrating acquired optimization teams and technologies into their core inference offerings to better serve growing enterprise demand. Inference Optimization: AI cloud providers are incorporating specialized startups to refine model serving efficiency and maximize hardware performance.