90+ Footnoted Sources

Nebius IaaS Strategy

Competitive Intelligence Report — From Yandex's Ashes to $20B AI Infrastructure
For: AI Infrastructure Strategy & Product Leaders Analyst: MinjAI Agents Date: February 16, 2026
01 Executive Summary
$530M
FY2025 Revenue[1]
478%
YoY Growth[1]
$20.4B
Backlog (MSFT + Meta)[2][3]
265
Open Roles[4]

Nebius Group (NASDAQ: NBIS) is the fastest-growing AI cloud, having transformed from the remnants of Yandex N.V. into a $20B+ market cap AI infrastructure company in under 18 months.[5] The company posted $529.8M in FY2025 revenue[1], signed $20.4B in hyperscaler contracts (Microsoft $17.4B[2] + Meta $3B[3]), and guides $3.0–$3.4B revenue for 2026 with ~40% adjusted EBITDA margins.[6]

What They Are Building A full-stack AI cloud platform spanning GPU clusters (H100/H200/Blackwell/Rubin), managed inference (Token Factory[7]), managed Kubernetes, and agentic AI infrastructure (Tavily acquisition[8]). Vertically integrated from in-house server design through managed AI services.
Competitive Context Nebius's Token Factory competes in the managed inference space where The platform is also building capabilities. Nebius prices inference up to 50% below hyperscalers[9] and leverages Yandex-heritage software. However, The platform's unique energy infrastructure, US-first positioning, and vertically integrated approach provide distinct competitive advantages that Nebius's colocation-dependent model cannot easily replicate.
02 Company Background: From Yandex to Nebius
1997
Arkady Volozh founds Yandex, Russia's dominant search engine. IPOs on NASDAQ in May 2011 at $1.3B valuation.[10]
Feb 2022
Russia invades Ukraine. NASDAQ suspends Yandex N.V. trading.[10]
Jul 2024
Yandex N.V. sells all Russian assets to Kremlin-aligned consortium for $5.4B.[10] Retains non-Russian cloud, autonomous driving (Avride), edtech (TripleTen), data labeling (Toloka).
Oct 2024
Renames to Nebius Group N.V., resumes NASDAQ trading under NBIS.[10] Volozh returns as CEO.[11]
Dec 2024
$700M oversubscribed private placement from Accel, NVIDIA, Orbis Investments.[12]
Sep 2025
$1B public offering at $92.50/share + $17.4B Microsoft deal announced.[13][2]
Nov 2025
$3B Meta deal + Token Factory launch + Q3 results ($146M revenue, 355% YoY).[3][7]
Dec 2025
AI Cloud 3.1 launch with NVIDIA Blackwell Ultra compute (HGX B300, GB300 NVL72).[14]
Jan 2026
Israel national supercomputer win ($140M, 4,000 B200 GPUs).[15] NVIDIA Vera Rubin NVL72 announced for H2 2026.[16]
Feb 2026
Tavily acquisition ($275M–$400M) for agentic search.[8] Q4/FY2025 results: $228M Q4 revenue, $530M full year.[1]

Portfolio Companies

CompanyFocusStatus
Nebius AI CloudFull-stack AI infrastructure (core business)Core Revenue Driver
AvrideAutonomous vehicles & delivery robotsR&D Stage
TripleTenEdTech for tech career trainingRevenue Generating
TolokaAI data labeling & model developmentPartner Ecosystem
ClickHouseOpen-source analytics databaseMinority Stake
03 Financial Performance & Capital Structure

Revenue Trajectory

MetricFY2024Q1 2025Q2 2025Q3 2025Q4 2025FY2025
Group Revenue $91.5M[1] ~$50M ~$106M $146M[17] $228M[1] $529.8M[1]
YoY Growth — — — 355%[17] 547%[1] 478%[1]
QoQ Growth — — — 39%[17] 56%[1] —
$1.2B
Core AI Cloud ARR[6]
830%
Core Cloud YoY Growth[6]
24%
Core Cloud Adj. EBITDA Margin[6]
$4.1B
FY2025 CapEx[1]

2026 Guidance

MetricTargetSource
Revenue$3.0–$3.4B[6]
ARR (End of 2026)$7–$9B[6]
Adj. EBITDA Margin~40%[6]
CapEx$16–$20B[6]
Contracted Power>3 GW[6]
Connected Power800 MW–1 GW[6]

Major Contracts

CustomerValueDurationScope
Microsoft$17.4B[2]5 yearsDedicated GPU capacity from Vineland, NJ data center
Meta$3.0B[3]5 yearsAI infrastructure, deployed within 3 months
Israel Innovation Authority$140M[15]Multi-yearNational AI supercomputer (4,000 B200 GPUs, 16,000 PFLOPS[18])

Capital Raises

DateTypeAmountKey Investors
Dec 2024Private Placement$700M[12]Accel, NVIDIA, Orbis Investments
Sep 2025Public Offering$1.0B[13]Public markets ($92.50/share)
Sep 2025Convertible NotesAdditionalConcurrent with equity offering[13]
Profitability Warning Despite exceptional top-line growth, Nebius posted a net loss of $249.6M in Q4 2025.[19] The company is heavily investing in capacity with $16–$20B CapEx planned for 2026[6], creating significant execution risk if AI demand cools.

CEO Shareholder Letter Insights (Q1–Q4 2025)

Arkady Volozh's quarterly shareholder letters reveal a consistent strategic narrative and escalating ambition:

QuarterKey CEO Quote / InsightStrategic Signal
Q1 2025 "On track to achieve $750M–$1B ARR by year-end." Revenue: $55.3M (+385% YoY). Expected EBITDA positive in H2 2025.[51] Early validation; conservative guidance setting
Q2 2025 Revenue grew 625% YoY and 106% QoQ. Achieved positive adj. EBITDA "sooner than expected." Raised ARR guidance to $900M–$1.1B.[52] Demand exceeding plans; securing >1 GW power
Q3 2025 "Every time we bring capacity online, we sell it." Running completely sold out, forcing rejection of enterprise customers. Meta $3B deal announced. Gross margins: 71%.[53] Demand-constrained, not supply-limited; hyperscaler validation
Q4 2025 "Everything we build, we sell. We are in the very early days of one of the biggest industrial and technological revolutions." Sold out Q3, Q4, and Q1 2026. Core cloud EBITDA margin: 24%.[54] Operating leverage emerging; 9 new DCs announced; Rubin commitment
Key Shareholder Letter Themes
  • Demand outstrips supply: Sold out for 3 consecutive quarters (Q3 2025 through Q1 2026)[54]
  • Customer quality escalation: From startups to Shopify, World Labs, Cursor, Black Forest Labs, then to Microsoft and Meta[53]
  • "Hyperscale AI cloud" ambition: Volozh consistently frames Nebius not as a AI cloud but as an "AI-native hyperscaler"[54]
  • Capital discipline narrative: Management describes itself as "dilution-sensitive" while planning $16–$20B CapEx, funded ~60% from operating cash flows[55]
  • Avride upside: Robotaxi launch with Uber in Dallas (Dec 2025); 215K+ deliveries YTD; potential value unlock[53]

Operational Metrics Detail (From Shareholder Communications)

MetricQ3 2025Q4 2025Notes
Cost of Revenue—31% of revenue[56]Implies ~69% gross margin
R&D Expense—33% of revenue ($177.3M FY)[56]Heavy investment in platform
SG&A Expense—71% of revenue ($378.5M FY)[56]Includes heavy GTM buildout
Operating Cash Flow—$834M (Q4)[55]$401.9M FY from continuing ops
Cash Position$4.9B[53]$3.7B[55]Spending down on CapEx
Convertible Notes—$4.16B issued[56]Significant debt instrument
Shares Outstanding—253M total[56]219.5M Class A + 33.6M Class B
04 Product Architecture & Technology Stack
Layer 4: Managed AI Services
Token Factory[7] • AI Studio (60+ models)[9] • Post-Training (Papyrax)[20] • Tavily Agentic Search[8]
Layer 3: Platform Services
Managed Kubernetes[21] • Managed Slurm (Soperator) • Container Registry • Capacity Blocks[14] • Terraform/CLI
Layer 2: Core IaaS
GPU Instances (H100/H200/B200/GB200)[22] • Storage (1 TB/s read, S3-compatible)[23] • VPC • PostgreSQL • MLflow
Layer 1: Physical Infrastructure
6+ Data Centers[24] • ~60,000 GPUs[25] • InfiniBand 800 Gbps[14] • In-house Server Design (ODM)[26]

Data Center Footprint

LocationTypeCapacityGPU HardwareStatus
Finland (Mantsala) Owned 25 MW active, expanding to 75 MW[27] H100/H200 supercomputer clusters Operational
New Jersey (Vineland) Partnership (DataOne) 300 MW[28] Dedicated Microsoft capacity Launching 2025
Missouri (Kansas City) Colocation (Patmos) 5 MW initial, 40 MW expandable[24] Blackwell & Hopper H200s Operational
UK (Longcross, Surrey) Colocation (Ark Data) Liquid-ready, high-density 4,000 Blackwell Ultra GPUs[29] Q4 2025
Iceland (Keflavik) Colocation (Verne) 10 MW[24] 100% renewable energy Operational
France (Paris) Colocation (Equinix) Urban campus Early H200 adopters[24] Operational

GPU Hardware Roadmap

GPU PlatformAvailabilityKey Spec
NVIDIA HGX H100Available nowFoundation compute, cost-effective[22]
NVIDIA HGX H200Available nowExtended GPU memory[22]
NVIDIA HGX B200Available nowBlackwell architecture for LLMs[22]
NVIDIA GB200 NVL72Available nowRack-scale liquid-cooled[22]
NVIDIA HGX B300Q4 2025[14]AI reasoning architecture
NVIDIA GB300 NVL72Q4 2025[14]800 Gbps InfiniBand (first globally)
NVIDIA Vera Rubin NVL72H2 2026[16]Next-gen; among first cloud providers
Key Technical Differentiator: Yandex Heritage Unlike most AI clouds building software from scratch, Nebius inherited decades of Yandex cloud engineering. Their managed Kubernetes, storage layer, networking, and database services represent years of production hardening that competitors like Crusoe, CoreWeave, and Lambda are still building.[30]
05 Token Factory: Managed Inference Deep Dive

Token Factory is Nebius's flagship managed inference product, launched November 2025.[7] It directly competes with The platform's Inference-as-a-Service strategy.

Architecture

Disaggregated Inference Splits prefill-specialized workers (long context processing) from decode-specialized workers (token generation), significantly improving Time-To-First-Token (TTFT) and throughput.[31]
Speculative Decoding Uses a smaller "draft" model to predict tokens, which are then verified by the target model, reducing latency for interactive applications.[31]
KV Cache Disaggregation Separates the prefill phase from the decode phase to optimize memory bandwidth and compute utilization independently.[31]
Autoscaling Dynamic worker scaling based on concurrent request volume ensures bursts don't impact latency. Dedicated endpoints with 99.9% SLA.[32]

Performance & SLA

MetricSpecification
Uptime SLA99.9% on dedicated endpoints[32]
Latency TargetSub-second[32]
Throughput Capacity100M+ tokens per minute[32]
APIOpenAI-compatible[32]
Data ResidencyEU (Finland, France) or US[32]
Zero-Retention ModeNo data stored or used for training[32]

Supported Models (60+)[32]

DeepSeek R1/V3, Llama 3.1/3.3, Qwen 2.5/3, GPT-OSS 120B/20B, Mistral, NVIDIA Nemotron, Gemma, GLM-4.5, Kimi-K2, Hermes, and more. Image generation via Flux and SDXL.

Token Factory Pricing (per 1M tokens)[33]

ModelTierInputOutput
DeepSeek-R1-0528fast$2.00$6.00
DeepSeek-R1-0528base$0.80$2.40
DeepSeek-V3-0324base$0.50$1.50
Llama 3.1 405Bbase$1.00$3.00
Llama 3.3 70Bbase$0.13$0.40
Qwen3-235B-Instructbase$0.20$0.60
Qwen3-30B-Instructbase$0.10$0.30
GPT-OSS 120Bbase$0.15$0.60
GPT-OSS 20Bbase$0.05$0.20
Kimi-K2-Instructbase$0.50$2.40
GLM-4.5base$0.60$2.20
Gemma-2-9b-itbase$0.03$0.09
Batch inference 50% of real-time price
Cost Advantage Claim Nebius claims pricing "up to 50% lower than other big-name providers"[9] and "up to 3x better cost-to-performance" independently verified by Artificial Analysis.[9]

Post-Training & Fine-Tuning[20]

Tavily Acquisition: Agentic Search[8]

06 GPU Cloud Pricing
GPUOn-Demand ($/hr)Reserved ($/hr)Discount
NVIDIA HGX H100$2.95[35]$2.00[35]~32%
NVIDIA HGX H200$3.50[35]$2.30[35]~34%
NVIDIA HGX B200$5.50[35]Contact—
NVIDIA GB200 NVL72Contact[35]Contact—
NVIDIA L40S (AMD)From $1.82[35]——
NVIDIA L40S (Intel)From $1.55[35]——
Explorer Tier (H100) $1.50 (first 1K hrs/mo)[36] — ~49% off on-demand
SemiAnalysis ClusterMAX Rating: Gold Tier Nebius earned a Gold medal in SemiAnalysis's GPU Cloud ClusterMAX rating system.[26] Among Gold-tier providers, Nebius offers the lowest absolute price while maintaining robust technical capabilities. Their ODM hardware strategy significantly lowers total cost of ownership.[26] GPU provisioning takes 45–60 seconds; 128-GPU clusters can be provisioned within 2 days.[26]

Reserved clusters: up to 35% discount on on-demand rates for multi-month commitments.[35]

07 Leadership & Organizational Structure

Organizational Chart

Arkady Volozh
CEO & Founder[11]
Ophir Nave
COO[37]
Roman Chernin
CBO[38]
Andrey Korolenko
Chief Product & Infra Officer[39]
Sarah Boulogne
Chief People Officer[40]
Dmitry Romashev
Chief Legal Officer[40]
Neil Doshi
VP, Head of IR[40]
John Boynton
Board Chairman[37]
Kira Radinsky
Independent Director[37]
Charles Ryan
Independent Director[37]
Elena Bunina
Non-exec Director[37]
Key Product/Infra Hiring (Open Roles)
PM, Kubernetes
Hiring[41]
PM, Security
Hiring[42]
PM, Public API & Tools
Hiring[4]
PM, Marketplace
Hiring[4]
TPM, Cluster Experience
Hiring[43]
DevRel, AI Cloud
Hiring[4]
DevRel, Token Factory
Hiring[4]
GPU Cluster Architect
Hiring[4]

Headcount

PeriodHeadcountNote
Dec 2023 (pre-restructuring)26,361[44]Included all Yandex Russia employees
Dec 2024 (post-restructuring)~1,371[44]Non-Russian operations only
Feb 2026 (current)~2,000+ (estimated)265 open roles suggest rapid growth[4]
08 Hiring Analysis: 265 Open Roles
DepartmentEst. Open RolesKey Signal
Sales & Business Development~40+Global GTM (APJ, DACH, France, Middle East, US)[4]
Customer Experience / Solutions Architecture~30+Enterprise land-and-expand, HPC specialists[4]
Hardware Infrastructure & Data Centers~35+Massive physical buildout across 6+ sites[4]
Engineering (Backend, SRE, ML, HPC)~50+Platform maturity, C++, storage virtualization, YDB[4]
Product Management~8–10K8s, Security, Marketplace, API, Cluster Experience[4]
Security & Compliance~10+SOC, incident response, privacy, AI governance[4]
Recruiting & HR~15+Scaling the scaling team (exec recruiter, ML recruiter)[4]
Marketing & Developer Relations~20+Brand, community, 3 DevRel hires[4]
Finance & Legal~15+Public company infrastructure (audit, payroll, treasury)[4]
Operations & Strategy~15+GTM ops, pipeline management, strategic programs[4]

Key Product Manager Roles (Detail)

RoleFocus AreaLocationKey Requirements
PM, Kubernetes[41] Managed K8s service, IaaS stack integration Amsterdam, Remote EU K8s expertise as user and DevOps engineer; IaaS knowledge (cloud networking, virtualization)
PM, Security[42] Cloud security features, compliance Amsterdam, Berlin, Finland, London, Remote EU Security product experience, enterprise compliance
PM, Public API & Tools[4] Developer experience, CLI/API Amsterdam, Belgrade, Berlin, Prague, Remote EU API design, developer tooling
PM, Marketplace[4] ISV/partner ecosystem Multiple EU locations Platform marketplace, partner ecosystem
TPM, Cluster Experience[43] GPU cluster UX for training/inference Netherlands, Remote HPC/GPU cluster management experience

Salary Benchmarks

Role25th PercentileMedian75th Percentile90th Percentile
Senior Product Manager[45] $181K $222K $277K $336K
What Hiring Tells Us
  • Global GTM machine: 40+ sales/BD roles across APJ, EMEA, Americas signals aggressive enterprise pursuit outside Europe
  • Data center buildout is real: 35+ hardware/DC roles for technicians, engineers, site managers backing the $16–$20B CapEx plan
  • Platform maturity: PM roles for K8s, Security, Marketplace, API suggest moving beyond raw GPU rental toward PaaS
  • Sovereign play: EU-based roles, compliance/privacy officers, AI governance specialist, data residency features
  • Developer ecosystem investment: 3 DevRel hires (AI Cloud, Physical AI & Robotics, Token Factory) plus Head of Developer Community
09 Compliance, Security & Certifications
CertificationStatusScope
SOC 2 Type IIAchieved[46]Full platform
HIPAAAchieved[46]Object storage audit logs, data-plane[14]
ISO 27001Achieved[46]Information security management
ISO 27799Achieved[32]Healthcare data protection
GDPRCompliant[46]EU data centers, zero-retention
Microsoft Entra IDIntegrated[14]Enterprise SSO
FedRAMPNot AnnouncedLikely roadmap item given Israel/US gov deals

Security features include VPC Security Groups[14], per-object access controls, granular service roles, and FOCUS-compliant billing exports for FinOps.[14]

10 Competitive Positioning

AI Cloud Landscape Comparison

MetricNebiusCoreWeaveCrusoeLambda
FY2025 Revenue $530M[1] ~$4.2B[47] ~$600M* ~$500M[47]
Connected Power 220 MW[25] 590 MW[47] 370 MW ~200 MW
Deployed GPUs ~60,000[25] ~600,000[47] ~45,000 ~40,000
Backlog $20.4B[2][3] $55B+[48] N/A N/A
ClusterMAX Rating Gold[26] Platinum[49] Gold Silver
Managed Inference Token Factory[7] Inference Workloads MemoryAlloy Lambda Inference
Software Stack Full (ex-Yandex)[30] Growing Growing Basic
Key Advantage Price + full stack Scale + NVIDIA priority[47] Energy + inference tech Simplicity
H100 On-Demand $2.95/hr[35] $2.62/hr $2.49/hr $2.49/hr

Nebius vs. Crusoe: Head-to-Head

DimensionNebiusCrusoe
OriginYandex cloud spinoff (decades of R&D)[10]Clean energy Bitcoin miner pivot
Software MaturityHyperscaler-grade (K8s, storage, networking, DB)[30]Building from scratch
Inference ProductToken Factory (disaggregated, 60+ models)[7]MemoryAlloy (KV-cache fabric, Rust/CUDA)
Enterprise ContractsMicrosoft $17.4B + Meta $3B[2][3]Growing enterprise base
Data Centers6+ locations (EU + US)[24]4+ US locations
Sovereign AIIsrael supercomputer, EU residency[15]US-focused
GPU Provisioning45–60 sec container pull, <2 day cluster[26]Fast provisioning (ClusterMAX Gold)
NVIDIA RelationshipCloud Partner, early Rubin access[16]Cloud Partner
CapEx (2026)$16–$20B[6]Not disclosed (smaller scale)
Customer Concentration Risk Nebius's $20.4B backlog is concentrated in two customers (Microsoft and Meta). If either relationship weakens, revenue projections face significant downside risk. This creates an opportunity for the platform to target the mid-market enterprises that hyperscaler-focused AI clouds underserve.[47]
11 Strategic Strategic Implications

The platform's Competitive Advantages

1. Energy Infrastructure Moat The platform owns and operates energy infrastructure, enabling lower total cost of ownership than Nebius's colocation-heavy model. Nebius depends on partners (DataOne, Ark, Equinix, Verne) for most data center capacity[24], creating margin pressure that The platform can avoid.
2. US-First Positioning Nebius is Europe-first with only recent US expansion.[50] The platform's deep US presence and relationships provide a natural enterprise beachhead that Nebius is still building.
3. Diversified Customer Base Nebius depends heavily on two customers (Microsoft and Meta) for revenue[47], creating concentration risk. The platform can build a more resilient customer portfolio targeting mid-market enterprises underserved by hyperscaler-focused AI clouds.
4. Vertical Integration The platform's integrated approach from energy to compute positions it to deliver differentiated inference performance and cost structure that Nebius's multi-vendor colocation model cannot easily replicate.

Market Dynamics to Monitor

1. Inference Pricing Benchmarks Nebius's ODM hardware strategy enables aggressive pricing ($2.95/hr H100, 50% below hyperscalers on inference[9]). All inference market entrants must be pricing-aware.
2. Token Factory Product Maturity Token Factory offers 60+ models[32], OpenAI-compatible API, 99.9% SLA, and disaggregated architecture[31]. This sets the product feature baseline for managed inference platforms.
3. Hyperscaler Validation Effect $20.4B in committed contracts from Microsoft[2] and Meta[3] validates the AI cloud business model broadly, which benefits all entrants including the platform.
4. Sovereign AI Demand Israel supercomputer win[15] (Google and Amazon withdrew[18]) signals growing sovereign AI demand that The platform can also pursue in the US market.

Recommended Actions

#ActionPriorityRationale
1 Benchmark against Token Factory pricing Critical Ensure The platform's inference costs compete with $0.13–$3.00/M token range[33]
2 Accelerate managed K8s and storage Critical Table-stakes for enterprise AI cloud buyers; Nebius has years of head start[23]
3 Pursue US government/defense AI deals High Follow Nebius's sovereign playbook; The platform has US energy + infrastructure advantage
4 Publish independent inference benchmarks High Nebius uses Artificial Analysis verification[9]; The platform needs similar third-party validation
5 Build developer ecosystem Medium Nebius has 3 DevRel hires + Head of Developer Community[4]; The platform needs equivalent investment
6 Target Nebius's customer concentration gap Medium Mid-market enterprises need alternatives to hyperscaler-dependent AI clouds

Sources & Footnotes

  1. [1] Nebius Q4 & FY2025 Financial Results (Feb 12, 2026). nebius.com/newsroom/nebius-reports-fourth-quarter-and-full-year-2025-financial-results
  2. [2] Nebius-Microsoft Multi-Billion Dollar Agreement. nebius.com/newsroom/nebius-announces-multi-billion-dollar-agreement-with-microsoft-for-ai-infrastructure
  3. [3] Nebius $3B Meta Deal & Q3 Results. invezz.com/news/2025/11/11/nebius-lands-3b-meta-ai-contract
  4. [4] Nebius Careers: 265 Open Positions. nebius.com/careers/roles
  5. [5] TechCrunch: The Curious Case of Nebius. techcrunch.com/2024/11/24/the-curious-case-of-nebius
  6. [6] Nebius Q4 2025 Earnings Call / Guidance. finance.yahoo.com/news/nebius-group-q4-earnings-call
  7. [7] Nebius Token Factory Launch (Nov 2025). businesswire.com/news/nebius-launches-token-factory
  8. [8] Nebius Tavily Acquisition Announcement. nebius.com/newsroom/nebius-announces-agreement-to-acquire-tavily
  9. [9] Nebius AI Studio: Inference-as-a-Service Cost Efficiency. nebius.com/newsroom/nebius-ai-studio-inference-cost-efficiency
  10. [10] Nebius Group / Yandex History (Wikipedia). en.wikipedia.org/wiki/Nebius_Group
  11. [11] Arkady Volozh CEO Profile. nebius.com/board-of-directors/arkady-volozh
  12. [12] Nebius $700M Strategic Equity Financing (Dec 2024). nebius.com/newsroom/nebius-700m-equity-financing
  13. [13] Nebius $1B Public Offering (Sep 2025). nebius.com/newsroom/nebius-1b-public-offering
  14. [14] Nebius AI Cloud 3.1 Launch (Dec 2025). nebius.com/newsroom/nebius-ai-cloud-3-1
  15. [15] Nebius Israel National Supercomputer ($140M). globes.co.il/nebius-to-build-israels-national-supercomputer
  16. [16] Nebius NVIDIA Vera Rubin NVL72 Announcement. nebius.com/newsroom/nebius-vera-rubin-nvl72
  17. [17] Nebius Q3 2025 Financial Results. nebius.com/newsroom/nebius-q3-2025-results
  18. [18] Israel Innovation Authority Supercomputer Selection. innovationisrael.org.il/nebius-to-establish-supercomputer
  19. [19] Nebius Q4 2025 Net Loss Analysis. simplywall.st/nebius-group-q4-loss
  20. [20] Token Factory Post-Training. nebius.com/blog/posts/post-training-in-token-factory
  21. [21] Nebius AI Cloud Platform. nebius.com/ai-cloud
  22. [22] Nebius GPU Infrastructure & AI Cloud. nebius.com/ai-cloud
  23. [23] Nebius Storage: 1 TB/s Read Throughput. nebius.com/ai-cloud
  24. [24] Nebius Hardware & Data Centers. nebius.com/hardware
  25. [25] Nebius Data Center Capacity & GPU Deployment. datacenterdynamics.com/nebius-meta-deal-capacity
  26. [26] SemiAnalysis ClusterMAX GPU Cloud Rating: Nebius Gold. semianalysis.com/clustermax-rating-system
  27. [27] Nebius Finland Data Center Expansion to 75 MW. nebius.com/newsroom/finland-data-center-75mw
  28. [28] Nebius New Jersey 300 MW Data Center. nebius.com/newsroom/us-expansion-300mw-new-jersey
  29. [29] UK 4,000 Blackwell Ultra GPU Deployment. nebius.com/newsroom/nebius-ai-cloud-3-1
  30. [30] Nebius Full-Stack Software Heritage. seekingalpha.com/nebius-vs-coreweave
  31. [31] Token Factory Disaggregated Inference Architecture. nebius.com/blog/posts/invisible-architecture-behind-chat-apps
  32. [32] Nebius Token Factory Product Page. nebius.com/services/token-factory
  33. [33] Nebius Token Factory Pricing. nebius.com/token-factory/prices
  34. [34] Bloomberg: Tavily Acquisition $275M. bloomberg.com/nebius-tavily-275m
  35. [35] Nebius GPU Pricing Page. nebius.com/prices
  36. [36] Nebius Explorer Tier: H100 at $1.50/hr. nebius.com/blog/posts/h100-explorer-tier
  37. [37] Nebius Board of Directors. nebius.com/board-of-directors
  38. [38] Roman Chernin, CBO & Co-founder. linkedin.com/in/roman-chernin
  39. [39] Andrey Korolenko, Chief Product & Infrastructure Officer. SEC Filing: Nebius Group Overview
  40. [40] Nebius About Page: Executive Team. nebius.com/about
  41. [41] PM, Kubernetes Job Listing. careers.nebius.com/jobs/product-manager-kubernetes
  42. [42] PM, Security Job Listing. himalayas.app/nebius/product-manager-security
  43. [43] TPM, Cluster Experience Job Listing. remoterocketship.com/nebius/tpm-cluster-experience
  44. [44] Nebius Group Employee Count (MacroTrends). macrotrends.net/NBIS/number-of-employees
  45. [45] Nebius PM Salary Data (Glassdoor). glassdoor.com/nebius-senior-product-manager-salaries
  46. [46] Nebius Security & Compliance Certifications. nebius.com/ai-cloud
  47. [47] AI Cloud Comparison: Nebius vs CoreWeave. seekingalpha.com/nebius-vs-coreweave
  48. [48] CoreWeave Backlog & AI Cloud Revenue Trajectory. stansberryresearch.com/coreweave-backlog
  49. [49] CoreWeave Platinum ClusterMAX Rating. coreweave.com/blog/clustermax-platinum
  50. [50] Nebius US First Availability Zone. nebius.com/blog/first-availability-zone-us
  51. [51] Nebius Q1 2025 Results & Shareholder Letter. finance.yahoo.com/nebius-q1-2025-results
  52. [52] Nebius Q2 2025 Shareholder Letter. nebius.com/newsroom/nebius-q2-2025-results
  53. [53] Nebius Q3 2025 Earnings Review (MVC Investing). mvcinvesting.substack.com/nebius-q3-2025-earnings
  54. [54] Nebius Q4 2025 Earnings Call Highlights. defenseworld.net/nebius-q4-earnings-highlights
  55. [55] Nebius Q4 2025 Operating Cash Flow & Financing. defenseworld.net/nebius-q4-earnings-highlights
  56. [56] Nebius FY2025 Financial Detail (StockTitan). stocktitan.net/nebius-fy2025-financials
February 2026
Analyst: MinjAI Agents
56 Unique Sources • 90+ Inline Footnotes