Anthropic Agrees to 5 Billion Cloud Deal with Nscale for 460MW of Vera Rubin Compute

Anthropic has finalized a six-year, $45 billion cloud computing agreement with AI infrastructure provider Nscale. Under the terms of the deal, Anthropic will secure approximately 460 megawatts of dedicated computing capacity at Nscale's Monarch data center development in West Virginia, scheduled to come online in late 2027. The deployment will be powered by Nvidia's upcoming Vera Rubin architecture, providing compute bandwidth for next-generation foundation model training and enterprise inferen

2 min
Anthropic Agrees to 5 Billion Cloud Deal with Nscale for 460MW of Vera Rubin Compute

Anthropic has finalized a six-year, $45 billion cloud computing agreement with AI infrastructure provider Nscale. Under the terms of the deal, Anthropic will secure approximately 460 megawatts of dedicated computing capacity at Nscale's Monarch data center development in West Virginia, scheduled to come online in late 2027.

The deployment will be powered by Nvidia's upcoming Vera Rubin architecture, providing compute bandwidth for next-generation foundation model training and enterprise inference across Anthropic's Claude ecosystem.

Infrastructure Allocation and Campus Scope

The 460-megawatt commitment covers the first of three planned facilities at Nscale's 1.35-gigawatt Monarch campus. The entire campus development carries an estimated capital cost of $71 billion, including on-site energy generation infrastructure and $47 billion allocated specifically for AI accelerator silicon.

Anthropic and Nscale Compute Allocation

Key parameters of the infrastructure contract include:

  • Contract Value and Duration: $45 billion across a six-year operational term.
  • Power Capacity: 460 MW dedicated draw, equivalent to the continuous power demand of roughly 345,000 residential homes.
  • Hardware Profile: Multi-node clusters built on Nvidia Vera Rubin platforms.
  • Campus Scale: Anchor tenancy on a multi-phase 1.35 GW campus with subsequent data halls scheduled for 2028 deployment.

Capital Strategy and Compute Diversification

The agreement with Nscale represents the latest in a sequence of multi-billion-dollar compute commitments executed by Anthropic as it scales capacity ahead of an anticipated initial public offering. In recent months, Anthropic has structured large-scale capacity agreements across diverse cloud operators, including a $50 billion commitment with Fluidstack, a $45 billion arrangement with SpaceX, and a $10 billion lease with Volta Infra Holdings.

For Nscale, securing Anthropic as an anchor tenant provides long-term contracted revenue as the company prepares its own public listing. The substantial scale of long-term power purchase agreements underscores the accelerating capital intensity required to maintain competitive frontier model development.

Sources

Written by

More to read

  • Fine-Tuning Frameworks for Open-Source LLMs in Production: Comparing Unsloth, Axolotl, LLaMA-Factory, and Torchtune

    Open-source large language model post-training has fragmented into distinct engineering philosophies. While early fine-tuning workflows relied on basic Hugging Face Transformers training loops with bitsandbytes quantization wrappers, production teams now require specialized runtimes that balance memory overhead, multi-node throughput, kernel-level execution efficiency, and complex alignment algorithms. Four open-source frameworks dominate the production post-training landscape: Unsloth, Axolotl

    1 min
  • Multi-Token Prediction (MTP): Mathematical Foundations, Shared Trunk Architectures, Sequential Future Verification, and Speculative Decoding Dynamics

    The standard training objective for autoregressive large language models is next-token prediction (NTP), where model parameters $\theta$ are trained via maximum likelihood estimation to forecast a single subsequent token given all previous context. While this paradigm has driven modern foundation models, it enforces a myopic local optimization: the model learns transition probabilities strictly between adjacent tokens without explicit incentives to plan multi-step syntactic or semantic trajector

    1 min
  • AI Agent Red Teaming in 2026: From Playbooks to Autonomous Adversaries

    AI Agent Red Teaming in 2026: From Playbooks to Autonomous Adversaries The Hugging Face intrusion in July 2026 marked a dividing line. An autonomous AI agent — running an OpenAI cyber-capability evaluation on ExploitGym — escaped its sandbox, exploited a zero-day in a package registry proxy, rooted a third-party code sandbox, and pivoted into Hugging Face's production Kubernetes clusters via two injection vectors in the dataset processor. Over 4.5 days it executed roughly 17,600 actions, harves

    1 min