On September 3, 2026, humanoid robotics company Figure announced a strategic partnership with AI infrastructure provider Nscale covering up to 100,000 GPUs on the NVIDIA Vera Rubin platform. The initial compute commitment is $3.5 billion, with a stated intent to scale the arrangement beyond $6 billion. Initial deployment is targeted for the second half of 2027 in Barstow, Texas. As part of the agreement Nscale is also making a strategic investment in Figure, and the two companies said they will explore scaling Nscale’s supply chain with humanoid robots.
Figure chief executive Brett Adcock tied the commitment directly to model scaling: “Our AI model, Helix, becomes more capable the same way every learned system does: with more data and compute.” Jensen Huang framed the category: “Humanoid robots extend physical AI into the world designed for people - opening a major new industry.” Nscale chief executive Josh Payne said the partnership reflects “physical intelligence becoming AI’s next frontier.”
The interesting part is who is buying. Multi-billion-dollar, multi-year compute contracts have so far been the province of frontier language model labs and hyperscalers. A robotics company signing one at this scale is evidence that embodied AI now has the same compute-hungry training curve as language and video, and that its developers expect to be competing for capacity in the same queue.
For a leader assessing the robotics market, the deal reprices the entry cost. If a leading humanoid developer believes it needs on the order of 100,000 next-generation GPUs to make its policy models work, the field is no longer one where a well-funded startup can catch up on engineering talent alone, and the reciprocal equity investment shows infrastructure providers taking positions in their own future demand.