Figure Commits $3.5 Billion to Nscale Compute on NVIDIA Vera Rubin

Figure announced on September 3, 2026, that it has signed a strategic partnership with AI infrastructure provider Nscale to deploy the NVIDIA Vera Rubin platform for up to 100,000 GPUs, with initial deployment targeted for the second half of 2027 in Barstow, Texas. According to the company’s announcement, the deal represents an initial commitment of $3.5 billion of compute, with intent to scale to over $6 billion.

As part of the agreement, Nscale is also making a strategic investment in Figure, and the two companies will explore the potential to scale Nscale’s supply chain with humanoids. Figure did not disclose the size, structure, or valuation terms of Nscale’s investment in the announcement.

Figure develops humanoid robots and the AI models that run them. The company said it is entering a phase where it is largely bound by the data and compute needed to train Helix, its AI model, and that the Nscale partnership provides the compute runway needed to train the next generation of AI models for general robotics.

“Our AI model, Helix, becomes more capable the same way every learned system does: with more data and compute. Today we’re excited to partner with Nscale to bring the compute required to make this vision a reality,” said Brett Adcock, founder and CEO of Figure.

Nscale founder and CEO Josh Payne said Figure is pushing the boundaries of AI further and that Nscale is proud to be enabling the future of AI robotics together.

The Vera Rubin Platform

The deployment centers on NVIDIA’s Rubin platform, which NVIDIA (NVDA ) launched on January 5, 2026, as a six-chip design spanning the Vera CPU, Rubin GPU, NVLink 6 Switch, ConnectX-9 SuperNIC, BlueField-4 DPU, and Spectrum-6 Ethernet Switch. NVIDIA stated at launch that the platform trains mixture-of-experts models with four times fewer GPUs and generates inference tokens at up to ten times lower cost per token than its Blackwell platform. Each Rubin GPU delivers 50 petaflops of NVFP4 compute for AI inference, and the platform’s Vera Rubin NVL72 rack-scale system combines 72 Rubin GPUs with 36 Vera CPUs.

NVIDIA said in the January announcement that Rubin is in full production, with Rubin-based products available from partners in the second half of 2026. It named Nscale among the first NVIDIA Cloud Partners to deploy Vera Rubin-based instances in 2026, alongside CoreWeave (CRWV ), Lambda, and Nebius, and listed Nscale among the companies expected to adopt the Rubin platform.

NVIDIA founder and CEO Jensen Huang addressed the Figure arrangement directly in Figure’s announcement. “Humanoid robots extend physical AI into the world designed for people—opening a major new industry,” Huang said. “Nscale and Figure have activated the robotics flywheel: training Figure’s models on NVIDIA Vera Rubin through Nscale’s AI cloud, validating them in NVIDIA Isaac Sim, and deploying them on NVIDIA GPUs in Figure’s robots. This is the physical AI flywheel that will accelerate the path from models to robots in the world.”

Nscale’s Vera Rubin Buildout

The Figure commitment lands on top of Nscale’s previously announced Vera Rubin deployment plans. Nscale said in an earlier announcement that it will be among the first providers globally to deploy the Vera Rubin platform in 2027, bringing more than 100,000 GPUs to Europe across its sites in the United Kingdom, Norway, and beyond. That deployment of Vera Rubin NVL72 systems supports Microsoft (MSFT ), making Nscale the first provider outside of Microsoft itself to deploy the platform.

In that announcement, Nscale said it delivers a full-stack AI cloud for Vera Rubin NVL72, including fleet management and observability, quota and identity controls, automated health checks, and unified policy enforcement, with bare metal, virtual machines, Slurm, and Kubernetes environments plus inference and fine-tuning services. Nscale also said its inference service has grown 148-fold and now serves more than 5,000 users.

Ian Buck, NVIDIA’s vice president and general manager of hyperscale and HPC, said in the Nscale announcement that deploying agentic and physical AI requires a new class of supercomputing infrastructure and that Nscale’s deployment of the Vera Rubin NVL72 platform will provide critical AI infrastructure for European developers.

Figure’s Data and Compute Spending

The partnership follows Figure’s August 25, 2026, launch of Index, the company’s effort to build what it describes as the most diverse humanoid training dataset ever assembled. In the Index announcement, Figure said the system is processing 30 minutes of video uploads every second — the partnership announcement cited a rate of 35 minutes of data every second — and that it is committed to spending over $1 billion over the next 12 months on data and compute.

Figure said Index has crossed 264,000 app downloads across more than 100 countries, with over 44,000 weekly active users contributing data, and that contributors, which it calls Creators, have uploaded over 16 million videos and earned $15 million to date. Per 1,000 hours collected, the dataset contains 373 unique tasks, 1,146 unique manipulated objects, and 116 unique environments, according to the company.

Figure said the Nscale partnership secures access to reliable compute as it works to bring humanoid robots into homes. Initial deployment of the Vera Rubin systems is targeted for the second half of 2027.