The Nebius October digest: AI Cloud 3.0 “Aether,” UK data center opening & Token Factory

This autumn has been one of acceleration for Nebius — from the rollout of the latest AI Cloud platform release to our brand new next-gen UK data center and the launch of Nebius Token Factory, an enterprise-grade inference platform.

Nebius AI Cloud 3.0 “Aether” release: Enterprise-grade security, compliance and control at scale

We announced version 3.0 of our AI Cloud platform, delivering all the trust and governance that organizations need without slowing AI developers down. In a bit more detail, Aether brings significant enhancements:

• Security and compliance: SOC 2 Type II (incl. HIPAA), ISO 27001 and alingment with NIS2, DORA and ISO 27032/27799. Built-in secrets management.
• Governance: Granular IAM controls and full observability with searchable logs, metrics and auditing.
• Developer productivity: Refined UI, simpler resource allocation and deeper integration with SkyPilot API Server and Anyscale.
• Performance and reliability: Automated health checks, selfhealing nodes, and faster object and file storage.

Our AI Cloud arrives in the UK

We are proud to announce our first deployment of NVIDIA Blackwell Ultra AI infrastructure in the UK, bringing the supercomputer power and hyperscaler simplicity of Nebius AI Cloud to UK customers from locally deployed specialized AI compute.

Accelerated by NVIDIA Blackwell Ultra GPUs and NVIDIA Quantum-X800 InfiniBand networking, the deployment delivers unprecedented performance for generative AI and future foundation model development.

Nebius Token Factory: Run open-source AI, at scale

We also launched Nebius Token Factory, the next evolution of our AI Studio platform. Token Factory brings together inference, posttraining, and governance in one place, enabling enterprises to turn open-source checkpoints into fully managed, production-ready AI systems.

It supports a growing library of models, including the including the NVIDIA Nemotron Nano 2 VL — a compact, production-ready multimodal reasoning model for real-world document intelligence and video understanding. Developers can now deploy and scale multimodal apps with greater flexibility and control than ever before.

Bringing industry-ready AI to life with Accenture

At NVIDIA GTC Washington, D.C., our own VP of Global Partnerships Laurelle Roseman joined Bryan Rich , Accenture ’s Global AI Lead for Health, Public Sector and Defense, to discuss how organizations can accelerate the move from proof of concept to production — fast, governed and sovereign.

Nebius delivers supercomputer-class performance with the flexibility of modern cloud. Accenture brings deep industry expertise and delivery at scale. Together, we’re enabling a secure, full-stack sovereign AI cloud — from infrastructure to applications — that helps enterprises and public institutions deploy AI with speed and confidence. You can watch the talk here.

Latest in our tech blog

• Discover how teams can run SkyPilot in our AI Cloud platform at scale with our new, deeper integration: no setup, no maintenance.
• Learn about one of the key features that makes NVIDIA GB200 NVL72 stand out: the fifth generation NVIDIA NVLink™ scale-up fabric.

Tech docs and tutorials

• Compute strengthens its security — learn how to enable automatic security updates for disks and keep your environments patched. We also added a guide on accessing a deployed container over a VM and a section on security updates for boot images.
• IAM grows with multi-tenant support — you can now create additional tenants to separate resources and projects within your organization.
• Third-party integrations expand — Anyscale joins the lineup, with a guide on connecting the platform for large-scale AI workloads to Nebius infrastructure.
• Billing moves to unified rates — starting this October, GPU, vCPU, and RAM billing for GPU VMs and standalone applications follow a unified model. See updated Compute pricing and Standalone applications pricing for details.
• Platform overview gets regional clarity — you can now check which services are available in each region.
• Virtual Networks gain routing control — explore three new articles in the Routing section that explain how traffic flows within your networks.