Nscale and Figure Sign $3.5B AI Compute Deal Backed by 100,000 NVIDIA GPUs

Nscale, a provider of AI infrastructure and high-performance GPU cloud services, has entered into a multi-year partnership with Figure, a humanoid robotics company, in a deal that underscores the escalating demand for large-scale computing power in the physical AI era. The agreement carries an initial commitment of US$3.5 billion in computing resources, with provisions to expand the total value beyond US$6 billion as the collaboration matures.

Under the terms of the partnership, the infrastructure is expected to support the deployment of up to 100,000 NVIDIA GPUs, which will underpin Figure's expanding computing backbone for its AI models and humanoid robots. The planned infrastructure may leverage NVIDIA's Vera Rubin platform, the company's next-generation AI computing architecture. Initial deployment is scheduled for the second half of 2027 in Barstow, Texas, marking a significant build-out in the region's data center landscape.

The collaboration positions Nscale as Figure's preferred compute provider and includes Nscale taking a strategic equity stake in the robotics firm. The computing infrastructure is specifically designed to support Figure's Helix AI models, which power its humanoid robots, and will integrate computing, AI infrastructure, and software across the full model development lifecycle—from training through deployment. Additionally, the companies plan to explore the use of humanoid robots within Nscale's own supply chain, creating a feedback loop between AI-driven automation and infrastructure operations.

Josh Payne, Founder and CEO of Nscale, pointed to the rapid growth of inference workloads and agentic AI as key drivers behind the deal, noting that Figure is pushing the boundaries of what is possible with AI. Brett Adcock, Founder and CEO of Figure, emphasized that the expansion of humanoid robots is largely constrained by access to data and compute, and that Helix becomes increasingly capable as both scale up. Jensen Huang, Founder and CEO of NVIDIA, framed the partnership as a milestone in the emergence of physical AI, describing it as a mechanism to activate the robotics flywheel—training Figure's models on NVIDIA Vera Rubin via Nscale's AI cloud, validating them in NVIDIA Isaac Sim, and deploying them on NVIDIA GPUs inside Figure's robots.

The deal reflects a broader industry shift in which robotics companies are securing dedicated, hyperscale-grade compute capacity to fuel advanced model training and real-time inference. By tying compute supply to equity participation and long-term infrastructure commitments, the agreement also signals a new template for strategic alignment between AI infrastructure providers and frontier technology developers.