Nebius, the Amsterdam-headquartered AI cloud company listed on Nasdaq as NBIS, said on October 1, 2026 that it has acquired Inferize, an inference optimization company whose technology shortens the time needed to launch and scale large AI models. Inferize's technology and team have joined Nebius Token Factory, the company's managed inference platform for production AI.
What Nebius said it bought
Nebius described Inferize as a company whose technology makes inference workloads elastic by shortening the time needed to launch and scale large AI models. The terms of the transaction were not disclosed.
According to the announcement, Inferize was founded in January 2026 and the team had a working prototype within three months. Nebius said Inferize's engineers will work across Nebius Token Factory, starting with the integration of their technology.
The cold-start problem
Nebius said cold starts — the time models need to load before serving a single request — impose additional time and costs on teams running AI models in production at scale. The company said this leaves assigned GPUs idle at launch time, when demand spikes and new instances spin up, and when weights are updated mid-run, for example during reinforcement learning, and forces platforms to hold spare capacity just to hit service-level targets.
Inferize's technology cuts this "idle GPU tax", Nebius said, enabling capacity to scale much more closely with actual usage and driving higher capacity utilization and better token economics. These are the company's claims; no independent performance figures were provided.
What the executives said
Danila Shtan, chief technology officer of Nebius, said: "Running inference well takes more than fast GPUs and optimized models. The whole system needs to respond when demand changes, including how quickly additional capacity is ready to serve customers." He added that the company is bringing both the technology and the team into Token Factory to make it more responsive to customer demand, and that the team's contribution will extend well beyond this first integration.
Guy Bortnikov, co-founder and chief executive of Inferize, said: "Keeping spare GPUs running is the price of being ready for demand. Removing that cost is what we built Inferize to do, and Nebius is where it can go straight into the platform."
How it fits the existing stack
Nebius said Inferize adds another layer to how it runs production inference. The company said Eigen AI brought optimization at the model, kernel and system levels into Nebius Token Factory, while Clarifai's core team and licensed technology added system-level inference and compute orchestration.
The press release, filed with the US Securities and Exchange Commission as an exhibit, carries a forward-looking statements disclaimer noting risks inherent in the integration of acquired businesses, competitive pressures and the company's ability to secure and retain clients.
Key facts and where they come from
- Nebius announced the acquisition of Inferize on October 1, 2026.
Amsterdam, October 1, 2026 — Nebius (Nasdaq: NBIS), the AI cloud company, today announced it has acquired Inferize
- Inferize's technology and team joined Nebius Token Factory.
Inferize’s technology and team have joined Nebius Token Factory, Nebius’s managed inference platform for production AI.
- Deal terms were not disclosed.
The terms of the transaction were not disclosed.
- Inferize was founded in January 2026 and had a prototype within three months.
Founded in January 2026, the Inferize team had a working prototype within three months.
- Nebius says the technology cuts the idle GPU tax and improves utilization.
Inferize’s technology cuts this “idle GPU tax”, enabling capacity to scale much more closely with actual usage and driving higher capacity utilization and better token economics.
- Eigen AI and Clarifai technology are already part of the Token Factory inference stack.
Eigen AI brought optimization at the model, kernel and system levels into Nebius Token Factory, while Clarifai’s core team and licensed technology added system-level inference and compute orchestration.