Nebius Acquires Israeli Startup Inferize to Cut AI Inference Costs

Nebius has bought the Israeli AI startup Inferize in a deal valued between 100 million and 150 million dollars.

The acquisition adds technology that speeds up the launch of large AI models and improves how GPUs are used during inference.

Why This Deal Matters for AI Infrastructure

Inferize was founded only ten months ago yet built a working prototype in three months with a team of 17 people in Tel Aviv.

The startup raised 10 million dollars in a seed round led by TLV Partners while operating in stealth mode.

Its founders previously helped build Granulate which Intel acquired in 2022 showing a clear pattern of Israeli optimization talent moving into bigger platforms.

Nebius will fold the team and tech into its Token Factory platform to make production inference more responsive to demand changes.

This move follows Nebius earlier purchase of Tavily and signals ongoing investment in software layers around raw GPU capacity.

Longer-Term Effects on Founders and the Market

Founders and early investors gain a fast exit that rewards rapid execution in the inference optimization space.

Over the next twelve months similar small teams may attract attention from cloud providers seeking to lower idle GPU costs and improve token economics.

Broader adoption of such tools could shift industry focus from training hardware to efficient runtime scaling benefiting operators who run models at scale.

Israel continues to supply specialized engineering talent as Nebius deepens its presence in the local ecosystem.