Urgent.News

What's breaking now, across thousands of outlets.

AI

Nvidia-backed AI company reveals staggering $103 billion number

Nvidia-backed AI company reveals staggering $103 billion number

Artificial intelligence firms are competing fiercely for a valuable resource: computing power. This race has led to significant investments in Nvidia processors, data centers, and power, as well as the emergence of specialized cloud providers to meet the infrastructure needs of AI models. UK-based AI infrastructure firm Nscale claims it has $103 billion in total contracted revenue, according to documents reviewed by The Information and cited by Reuters.

Nscale's contracts typically last about 5.7 years, which translates to roughly $18 billion in annual contract income. The company is considering an initial public offering as early as September. However, the $103 billion figure is based on committed revenue, not income already earned. A notable contract with Anthropic, valued at $45 billion, makes up nearly 44% of Nscale's contracted income, based on the figures provided by Reuters.

This commitment is for Nscale's data center in West Virginia, powered by Nvidia's new Vera Rubin processors. The $103 billion figure is illustrative, not an official revenue projection. Nscale will need to install the processors, networking, power, and data-center capacity to fulfill these contracts. The company's collaboration with Nvidia puts it in a larger ecosystem growing around Nvidia's technology, which could signal future AI demand for Nvidia investors.

Written by urgent.news from Yahoo Finance's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at finance.yahoo.com →

More in AI

Give Your Coding Agent Eyes

Every coding-agent session starts blind: your agent rediscovers how your features work, and sometimes guesses wrong. AGENTS.md is the usual fix, but it loads every fact into every session — too thin…

  • Coding agents begin sessions blind, needing feature rediscovery.
  • AGENTS.md file loads relevant info, but can be sparse or heavy.
  • Catalog system improves efficiency by storing feature notes.

Backtest 63.7%, production AUC 0.51 - a postmortem

Originally published at nbaproplab.com/learn/model-postmortem . A model that looks excellent offline and random in production is usually blamed on the serving code.

  • Model achieved 63.7% hit rate and 0.56 AUC in offline testing
  • Performance degraded to coin flip (AUC of 0.51) in production
  • Missing calibration snapshot caused train-serve mismatch

More from Saturday 5 September →