Home / Technology / Crusoe AI Factories Power NVIDIA Vera Rubin Platform
Crusoe AI Factories Power NVIDIA Vera Rubin Platform
22 Oct
Summary
- Crusoe's modular AI factories have six years of proven operational experience.
- The company manufactures its own data centers and integrates them with its cloud.
- Crusoe Spark supports NVIDIA DSX POD for Vera Rubin deployments.
Crusoe, an AI factory company, is set to enhance NVIDIA's Vera Rubin Platform with its Crusoe Spark modular AI data center platform. This development supports the newly announced NVIDIA DSX POD reference design, a blueprint for modular infrastructure.
Crusoe brings six years of field-proven experience in building and operating modular AI factories, dating back to 2020. The company distinguishes itself through vertical integration, manufacturing its own data centers, including the Crusoe Spark Factory in Brighton, Colorado, and integrating them with its Crusoe Cloud platform.
The partnership leverages Crusoe's expertise to provide a foundation for high-performance AI and the safer deployment of AI agents. This modular approach offers a fast and flexible way to deploy compute capacity, complementing larger-scale builds.
Crusoe has championed modular infrastructure since 2020, expanding its capabilities by acquiring electrical manufacturing specialist Easter-Owens Electric in 2022. The Crusoe Spark platform, formalized in 2025, is now mass-produced at their 352,000-square-foot factory. The latest offering, Crusoe Spark Liquid, is engineered for ultra-high-density systems like NVIDIA Vera Rubin NVL72.
Bryan Schramm, Head of Spark for Crusoe, emphasized the company's long-standing commitment to modularity, stating, "Modular isn't a new strategy or pivot for us. It's been in our DNA since our founding." He highlighted Crusoe's nearly a decade of operational field experience and its domestic factory capacity, enabling productive AI capacity delivery in as little as three months.
Crusoe's vertical integration ensures seamless capacity transfer from manufacturing to live AI workloads. The company has demonstrated performance parity with NVIDIA Exemplar Cloud within its modular AI factories. Customers can access NVIDIA Vera Rubin capacity via Crusoe Cloud or deploy dedicated units through Crusoe Edge Zones.
Charlie Boyle, vice president of DGX Systems at NVIDIA, acknowledged the critical role of modular AI factories in meeting global demand for accelerated computing. He noted that Crusoe's application of its expertise to the NVIDIA DSX POD reference design helps customers deploy production inference faster with Vera Rubin infrastructure designed for the lowest token cost at scale.
