🐯虎嗅•Stalecollected in 6m
Arkady Volozh Rebuilds AI Empire as Nebius in Amsterdam
💡Learn how a former tech giant's international arm is pivoting to become a dedicated AI cloud infrastructure provider.
⚡ 30-Second TL;DR
What Changed
Yandex N.V. sold its Russian business for $5.2 billion, leaving the international assets to form Nebius.
Why It Matters
Nebius emerges as a new player in the European AI cloud market, potentially offering a neutral, Western-aligned alternative for AI infrastructure needs.
What To Do Next
Monitor Nebius's upcoming cloud service benchmarks to see if they offer a competitive alternative to AWS or GCP for large-scale model training.
Who should care:Founders & Product Leaders
Key Points
- •Yandex N.V. sold its Russian business for $5.2 billion, leaving the international assets to form Nebius.
- •Nebius is pivoting to focus exclusively on AI cloud infrastructure and high-performance computing.
- •The company has fully relocated its operations to Amsterdam to ensure independence from Russian regulatory control.
- •Non-core assets like autonomous driving (Avride) and ed-tech (TripleTen) have been spun off to streamline the AI cloud focus.
🧠 Deep Insight
Web-grounded analysis with 21 cited sources.
🔑 Enhanced Key Takeaways
- •Nebius is a Nasdaq-listed public company (NBIS) headquartered in Amsterdam, operating globally with additional R&D and commercial hubs in Europe, North America, and Israel.
- •The company has secured substantial funding, including a $4.34 billion convertible debt funding round in March 2026, and aims to achieve two gigawatts of contracted capacity and $7-9 billion in annual recurring revenue by the end of 2026.
- •Nebius has established significant strategic partnerships, including a $27 billion infrastructure agreement with Meta and a $17.3 billion supply deal with Microsoft, which Arkady Volozh describes as financing tools to build Nebius's own higher-margin AI cloud.
- •Beyond core infrastructure, Nebius Group retains stakes in Toloka AI (data solutions), Avride (autonomous driving), and TripleTen (EdTech), and has acquired agentic search company Tavily and AI model optimization startup Eigen AI.
- •Nebius is building a vertically integrated, full-stack AI platform, controlling aspects from substation design and proprietary server/rack design to inference optimization software, aiming to capture margin across the entire AI stack.
📊 Competitor Analysis▸ Show
| Feature/Category | Nebius AI Cloud | CoreWeave | Runpod | Hyperstack |
|---|---|---|---|---|
| Primary Focus | Full-stack AI cloud infrastructure, model training, inference, HPC | High-scale GPU cloud for AI/ML | On-demand GPU compute, developer-friendly | GPU cloud with European data residency, GDPR compliance |
| GPU Offerings | NVIDIA Blackwell, Hopper systems, InfiniBand fabric | Cutting-edge GPUs (unspecified models in search results, but generally high-end) | Wide selection including RTX 5090, H200, B200, consumer RTX cards | NVIDIA H100 PCIe, H100 SXM, H200 SXM, HGX B200, HGX B300 |
| Managed Services | Managed Kubernetes, Slurm, MLflow, Serverless AI, observability | One-click clusters, enterprise backing | Instant Clusters, Docker control, pre-built templates, serverless GPU endpoints | Managed Kubernetes (implied by cloud-native focus) |
| Pricing Model | On-demand, contracts for competitive rates | Enterprise contracts, high-scale | Per-second billing, community cloud for cheaper rates | On-demand, competitive rates for H100 PCIe |
| H100 PCIe On-demand | ~$2.95/hr (HGX H100 rate) | (Not specified in search results) | ~$1.99/hr (community cloud) | $1.90/hr |
| H200 SXM On-demand | $3.50/hr | (Not specified in search results) | $4.23/hr (on-demand) / $1.43/hr (spot) | $3.50/hr |
| Geographic Focus | Global, with data centers in Europe (Finland, Paris), US (Kansas City, Vineland) | Global (implied) | Global (implied) | European data residency, GDPR compliance |
| Key Differentiators | Purpose-built AI infrastructure, bare-metal performance, deep NVIDIA collaboration, vertical integration, large anchor contracts | "AI Developer Cloud" with enterprise backing, high-scale | Cost efficiency, ease of use, broad GPU selection, developer-friendly | Focus on EU data residency, competitive H100 pricing |
🛠️ Technical Deep Dive
- Nebius AI Cloud provides access to NVIDIA accelerated computing clusters, featuring the latest Blackwell and Hopper systems.
- These systems are interconnected by non-blocking NVIDIA InfiniBand fabric, enabling high-speed distributed computing.
- The platform offers fully managed Kubernetes and Slurm orchestrators for efficient management and scaling of containerized ML/AI applications and distributed training.
- It includes granular observability and topology-aware job scheduling to optimize workload placement and monitoring.
- High-performance storage solutions deliver up to 1 TB/s read throughput for shared filesystems and 2 GB/s per GPU for object storage, with support for optimized in-house solutions or partners like WEKA and VAST Data.
- Nebius infrastructure is purpose-built for AI workloads, incorporating high-pressure liquid cooling and substation-level power engineering.
- The company emphasizes bare-metal performance by minimizing virtualization overhead to maximize Model FLOPS Utilization (MFU).
- Services include Compute (virtual machines, containers over VMs), Storage (Object Storage buckets, PostgreSQL clusters, Container Registry), and AI services (Serverless AI, Managed Service for MLflow, Standalone Applications).
- Nebius has acquired Eigen AI for model compression and graph optimization, and Tavily for fast, high-quality web search through an API, supporting agentic AI development.
- The company's networking fabric is designed to reach 3.2 TB/s per host, supporting demanding NVIDIA platforms like the GB300 NVL72 at full performance.
🔮 Future ImplicationsAI analysis grounded in cited sources
Nebius will emerge as a significant challenger to established hyperscale cloud providers in the specialized AI infrastructure market.
Its vertically integrated approach, massive funding, and strategic partnerships with tech giants like Meta and Microsoft position it to rapidly scale its dedicated AI compute capacity and offer competitive, purpose-built solutions.
The company's focus on a 'Token Factory' and acquisitions in AI optimization and agentic search will drive innovation in AI inference and higher-level AI services.
By moving beyond raw GPU hours to offering managed inference layers and tools for agentic AI, Nebius aims to capture higher-margin value and cater to the evolving needs of AI developers.
Nebius's European headquarters and adherence to data privacy regulations will attract a distinct segment of global enterprises and AI developers.
Its Amsterdam base and compliance with standards like GDPR, HIPAA, and SOC 2 make it a preferred choice for organizations prioritizing data residency and stringent security in their AI deployments.
⏳ Timeline
2022-06
Arkady Volozh resigned as CEO of Yandex after being sanctioned by the EU.
2022-12
Arkady Volozh announced his departure from Yandex, calling it 'the project of my life'.
2024-02
Yandex N.V. announced a definitive agreement to sell its Russian businesses for approximately $5.2 billion.
2024-03
Arkady Volozh was removed from the EU sanctions list.
2024-07
Yandex N.V. finalized the sale of its Russian assets and relaunched its international businesses as Nebius Group, with Arkady Volozh returning as CEO.
2024-10
Nebius Group went public, listed on Nasdaq (NBIS).
2025-09
Nebius secured a $17.3 billion supply deal with Microsoft.
2026-03
Nebius sold $2 billion in share warrants to Nvidia and closed a $4.34 billion convertible debt funding round, and announced a $27 billion infrastructure agreement with Meta.
2026-05
Nebius acquired AI model optimization startup Eigen AI.
📎 Sources (21)
Factual claims are grounded in the sources below. Forward-looking analysis is AI-generated interpretation.
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
📰
Weekly AI Recap
Read this week's curated digest of top AI events →
👉Related Updates
AI-curated news aggregator. All content rights belong to original publishers.
Original source: 虎嗅 ↗