Skip to header Skip to main navigation Skip to main content Skip to footer

Nvidia May Co-Sign the Data Center: Chip Supplier, Investor, Now Debt Backstop

Breadcrumb

  • Home
  • Nvidia May Co-Sign the Data Center: Chip Supplier, Investor, Now Debt Backstop

Nvidia is in talks to guarantee billions of dollars in debt financing for Nscale's planned Anthropic data center in West Virginia, The Information reported on Thursday, in a story TechCrunch published late on October 30 and Bloomberg corroborated. Anthropic has signed a 44.6 billion dollar contract for the project's first building alone, according to Nscale's IPO filing. In plain terms, the chipmaker would be co-signing a customer's loans, a sign of how hard the industry is working to keep its buildout afloat as investors grow jittery about the sums involved.

Under the proposed residual value guarantee, Nvidia would step in if Nscale defaulted on construction payments, temporarily take over the facility, and help find a buyer and a new tenant. That is a step beyond Nvidia's earlier support for cloud providers, which mostly covered leases and chip purchases, not construction itself. The guarantee would help Nscale finance the electricity and construction costs of the campus, whose first 460-megawatt building is expected to use Nvidia's next-generation Vera Rubin chips. The two companies are also discussing a second deal worth roughly 45 billion dollars for another building.

The stakes for Nscale are unusually exposed. The company is seeking to go public at a valuation of up to 35 billion dollars, yet it had only about 25,000 advanced AI chips running in August, a fraction of the nearly 500,000 it plans to deploy. Its West Virginia site has flooded nearby homes with construction runoff more than once before winning state certification in September, and this month Microsoft pulled out of a planned Nscale site in England after grid power there was pushed into the 2030s. Nvidia has already extended roughly 170 billion dollars in investments and credit support to neoclouds and large customers, including 1 billion dollars in convertible financing committed to Nscale in September, while shares of CoreWeave, Nebius and IREN had each fallen more than 30 percent in October.

For builders, the lesson is not really about one campus in West Virginia. It is that the frontier now runs on a web of guarantees in which the same company can be chip supplier, shareholder and debt backstop at once. That concentration helps projects get financed when a tenant like Anthropic may itself face a below-investment-grade credit rating after its IPO, but it also means a stumble at one neocloud does not stay contained. If you rent capacity, price that counterparty risk the way you would price a model: ask who actually stands behind the building, the power and the lease when markets tighten.

The practical deadline this week is smaller and closer to home. Google is winding down Jules, its asynchronous coding agent launched in December 2024, folding what it learned into Antigravity, its agent-first development platform. Developers can finish sessions until November 11, after which Jules turns read-only until it shuts down on December 31, with an export for settings, memories, setup scripts and session history. Code and pull requests already in GitHub stay there, but the agent's own context does not move itself. Export it now, before the read-only date turns a tidy migration into archaeology.

Rocky Mountain Madman LLC
AI Funding
Enterprise AI
Anthropic
Frontier Models
Google DeepMind

Ai Industry News

Mistral Goes Big: Large 4 "Le Chonk" Puts Europe on the Frontier Map

Cartoon illustration of a colossal friendly robot shaped like a mountain peak with glowing circuit patterns, towering over a stylized European cityscape at dusk.

Mistral has finally shipped a frontier-class model, and it is a statement piece. Large 4, immediately nicknamed "Le Chonk" by an internet that could not resist, is a natively multimodal mixture-of-experts model with 1.05 trillion total parameters and 49 billion active parameters, wrapped around a 1-million-token context window. It was trained from scratch on 3,800 Grace Blackwell GPUs in Mistral's own European datacenters — no rented American cloud, no borrowed infrastructure — and it is available right now as a public preview through Mistral's API.

  • Read more about Mistral Goes Big: Large 4 "Le Chonk" Puts Europe on the Frontier Map

Anthropic Slashes Haiku Pricing 90% as the Small-Model Price War Hits the Enterprise

Cartoon illustration of a cheerful robot mechanic tuning a row of small friendly robots at a workbench, with blank price tags floating in the air.

Anthropic released Claude Haiku 5.5 on October 7 with a price cut that rearranges the math of production AI work: about 90 percent cheaper than Haiku 4.5 on the prompts that matter. Short-context requests up to 100,000 tokens now cost 10 cents per million input tokens and 50 cents per million output tokens, with cache reads at a penny. Above the 100k line the discount drops to 50 percent.

  • Read more about Anthropic Slashes Haiku Pricing 90% as the Small-Model Price War Hits the Enterprise

Reflection AI's Beam: A 501B Open-Weight Bet on Efficient Reasoning

Cartoon illustration of a lighthouse beam made of glowing neural-network nodes rising from a small engine into a night sky.

Reflection AI has unveiled Beam, a 501-billion-parameter text-only mixture-of-experts model with just 23 billion active parameters, trained on 23.8 trillion tokens with a 1-million-token context window. The company's headline claim: Beam matches GLM-5.2 on reasoning while using three to four times less inference compute. The weights are expected later in October under Apache 2.0, once red-teaming wraps up. In a market obsessed with parameter counts, Beam is a bet that the number on the box matters less than the number actually doing work.

  • Read more about Reflection AI's Beam: A 501B Open-Weight Bet on Efficient Reasoning

Pagination

  • First page « First
  • Previous page ‹ Previous
  • …
  • Page 23
  • Page 24
  • Page 25
  • Page 26
  • Page 27
  • Page 28
  • Page 29
  • Current page 30
  • Page 31
  • Next page Next ›
  • Last page Last »

Copyright © 2026 Rocky Mountain Madman LLC - All rights reserved

Developed and Designed by Rocky Mountain Madman LLC