Particle.news
Download on the App Store

Nscale Commits $3.5 Billion to Figure and Plans Up to 100,000 GPUs

The deal makes Nscale Figure’s preferred compute provider to speed training and deployment of Figure’s Helix models for humanoid robots.

Overview

  • The multi‑year agreement, announced Thursday, guarantees at least $3.5 billion of AI cloud capacity for Figure with an intent to scale the commitment beyond $6 billion.
  • Nscale plans to deploy up to 100,000 NVIDIA Vera Rubin GPUs with initial hardware installations targeted to begin in the second half of 2027 at a site in Barstow, Texas.
  • As part of the arrangement Nscale will take an undisclosed equity stake in Figure and serve as the startup’s preferred compute provider.
  • Figure will use the compute to train its Helix models that power humanoid robots, which ingest continuous video, lidar and sensor streams and are fed by Figure’s Index dataset that is producing large volumes of human activity video.
  • The partnership reflects a wider industry pattern of infrastructure firms investing in customers, a move that boosts scaling plans and Nvidia’s ecosystem but has raised analyst concerns that such ties could blur true end‑user demand and complicate investor scrutiny.