๐Ÿ“ŠStalecollected in 36m

Nvidia $2B Investment in Nebius AI Centers

Nvidia $2B Investment in Nebius AI Centers
PostLinkedIn
๐Ÿ“ŠRead original on Bloomberg Technology
#funding#data-centers#ai-hardwarenebius-ai-data-centers

๐Ÿ’กNvidia's $2B data center bet fuels AI infra boom for practitioners

โšก 30-Second TL;DR

What Changed

Nvidia invests $2 billion in AI data center firm Nebius

Why It Matters

Bolsters global AI compute capacity amid surging demand. Reinforces Nvidia's dominance in AI hardware supply chain. Signals more infrastructure investments to support LLM training and inference.

What To Do Next

Assess Nebius platforms for scaling your Nvidia GPU-based AI workloads.

Who should care:Founders & Product Leaders

๐Ÿง  Deep Insight

Web-grounded analysis with 7 cited sources.

๐Ÿ”‘ Enhanced Key Takeaways

  • โ€ขNebius, valued at $23.38 billion, reported 106.23% revenue growth to $363.3 million over the last twelve months and plans to deploy NVIDIA Vera Rubin NVL72 across US and Europe data centers starting H2 2026[1][2]
  • โ€ขNebius signed a 5-year, $3 billion AI infrastructure agreement with Meta and has current capacity sold out, targeting 25GW by end of 2026 with 800MW-1GW connected power[4][7]
  • โ€ขNebius leased 80MW across two new Israeli data centers from Mega Or for $880 million construction cost, with facilities in Masmiyya (Q3 2026) and Beit Shemesh[5]
  • โ€ขNebius operates efficient facilities like the Mรคntsรคlรค data center in Finland with low PUE via heat-recovery and launched a robotaxi service in Dallas with Uber[1][6]

๐Ÿ› ๏ธ Technical Deep Dive

  • โ€ขNebius deploys NVIDIA Vera Rubin NVL72, launched at CES 2026, for complex AI workloads including agentic and advanced reasoning, complementing existing GB200 NVL72 and Grace Blackwell Ultra NVL72[1][2]
  • โ€ขFirst in Europe to operate NVIDIA GB300 NVL72 and HGX B300 systems in production; validated as NVIDIA Exemplar Cloud Partner against reference architectures and benchmarks[1][4]
  • โ€ขProprietary server racks and InfiniBand-based networking software reduce latency tax for large-scale model training; Nebius AI Cloud 3.1 adds Blackwell Ultra with real-time GPU dashboards[4][6]

๐Ÿ”ฎ Future ImplicationsAI analysis grounded in cited sources

Nebius achieves positive EBITDA by H2 2026
Analyst projections and infrastructure scaling with major contracts like Meta's $3B deal support consistent profitability as capacity connects 800MW-1GW by year-end[6][7]
Nebius expands to 25GW total power by end-2026
Company targets include sold-out current capacity, new Israeli 80MW leases, and US/Europe Rubin deployments amid accelerating AI demand[5][7]
NVIDIA Rubin boosts Nebius customer AI apps from H2 2026
As first-mover NVIDIA Cloud Partner, Nebius integrates Rubin NVL72 for reasoning/agentic workloads with benchmark-validated bare-metal performance[1][2]

โณ Timeline

2025-12
Launched Nebius AI Cloud 3.1 with NVIDIA Blackwell Ultra, HGX B300, and GB300 NVL72 as first in Europe
2025-12
Entered 5-year $3B AI infrastructure agreement with Meta
2026-01
Announced deployment of NVIDIA Vera Rubin NVL72 for H2 2026 across US and Europe data centers
2026-03
NVIDIA collaborates with Nebius and Nscale for UK AI infrastructure with 14,000 Blackwell GPUs

๐Ÿ“ฐ Event Coverage

๐Ÿ“ฐ

Weekly AI Recap

Read this week's curated digest of top AI events โ†’

๐Ÿ‘‰Related Updates

AI-curated news aggregator. All content rights belong to original publishers.
Original source: Bloomberg Technology โ†—