An interview with Arkady Volozh, CEO of Nebius, formerly Yandex NV, on selling its Russian assets and pivoting to become a full-stack AI infrastructure provider
Yandex plans to triple its Nvidia GPU deployments
Context & Ripple Effects
The pivot follows the finalized sale of Yandex’s Russian business and renaming of the new company as Nebius, separating the former Yandex NV from the Russian search-and-ad operation. The interview frames the newly named company around AI infrastructure rather than its prior internet-business identity.
Later coverage records both Nebius’s $700M financing from Nvidia, Accel and others and a larger cloud-capacity agreement with Microsoft, evidence that the infrastructure repositioning became tied to external capital and enterprise demand.
First-order effects
- Nebius’s planned tripling of Nvidia GPU deployments commits more of its near-term infrastructure buildout to AI compute capacity and supports its stated full-stack provider positioning.
- The post-sale company has a clearer operating focus: building AI cloud services under the Nebius name, while the Russian business continues separately as Yandex.
Second-order effects
- A larger GPU fleet raises the competitive requirement for AI-cloud rivals: selling raw capacity alone becomes less differentiated, increasing pressure to pair compute with software and managed infrastructure.
- The expansion deepens Nebius’s dependence on Nvidia GPU supply and makes financing and customer commitments more central to how quickly it can turn capacity into a cloud business.
Third-order effects
- If this model holds, corporate separations can turn legacy internet platforms into focused AI-infrastructure companies, with compute assets and cloud operations becoming the core investable business.
- AI infrastructure is moving toward platformization: providers that combine scarce compute with an integrated operating layer may hold a stronger position than operators that offer hardware access alone.
The trend: Nebius is one data point in the commercialization of AI compute, as companies reorganize around GPU-backed cloud platforms and pursue scale through capital, supply access, and enterprise customers.