Figure is partnering with AI infrastructure company Nscale to deploy up to 100,000 GPUs on NVIDIA’s Vera Rubin platform, marking a major step toward scaling the AI systems behind its humanoid robots.
The initial deployment is planned to begin in the second half of 2027 in Barstow, Texas, with an initial compute commitment of $3.5 billion and plans to potentially expand the deal to more than $6 billion.
Scaling the AI Behind Humanoid Robots
For Figure, the partnership addresses one of the biggest challenges in advancing its Helix robotics model: the enormous amount of data and compute required to train increasingly capable physical AI systems.
Figure recently introduced Index, an initiative focused on building a highly diverse humanoid training dataset. The system is generating around 35 minutes of training data every second, creating a rapidly growing demand for computing power to process and learn from that data.
Through Nscale’s AI cloud infrastructure and NVIDIA’s Vera Rubin platform, Figure aims to gain the compute capacity needed to train the next generation of models capable of operating in the physical world.
The partnership will also connect different stages of Figure’s robotics development—from training models on NVIDIA GPUs, to testing them in NVIDIA Isaac Sim, and eventually deploying them on physical humanoid robots.
Building the Physical AI Flywheel
NVIDIA CEO Jensen Huang described the collaboration as a “physical AI flywheel,” where models are trained using large-scale compute, tested in simulation, and then deployed into real-world robots.
Nscale will also make a strategic investment in Figure, while the two companies plan to explore opportunities to expand Nscale’s supply chain using humanoid robots.
With access to massive computing resources, Figure is positioning Helix for continued scaling as it works toward its broader goal of bringing humanoid robots into everyday environments—and eventually, homes around the world.
