SpaceX has revealed plans to launch its Vera Rubin NVL72 rack‑scale system into orbit next year, with a larger deployment slated for 2028. The company said the new platform will enable rapid, scalable data processing for space‑based observatories and other scientific missions. The announcement comes as the company seeks to expand its commercial payload services and support a growing demand for on‑orbit computing.
The Vera Rubin NVL72 is designed to fit within a standard 19‑inch rack, making it compatible with existing ground‑based data centers. SpaceX claims the system will provide significant computational power while consuming only a fraction of the power and mass of traditional on‑orbit supercomputers. Engineers say the platform will be integrated into the Starship launch vehicle’s payload bay and will use the spacecraft’s existing power and cooling systems.
SpaceX’s engineers explain that the Vera Rubin NVL72 will house multiple high‑performance GPUs and a custom FPGA architecture. This combination allows for real‑time image processing, machine‑learning inference, and data compression directly in space. By reducing the need to transmit raw data back to Earth, the system can lower bandwidth costs and shorten the time between observation and analysis. The company estimates the platform will process up to 10 terabytes of data per day, a capability that could accelerate research in astronomy, Earth observation, and planetary science.
The system is named after astronomer Vera Rubin, who pioneered the study of galaxy rotation curves. SpaceX’s spokesperson said the name reflects the company’s commitment to supporting fundamental science. „We want to give scientists the tools they need to answer the biggest questions about the universe,” the spokesperson said. The launch is scheduled for early 2027, with a follow‑up deployment of a larger, „significant scale” version in 2028.
SpaceX’s timeline has drawn skepticism from some experts. Critics point out that integrating a complex computing platform into a launch vehicle requires extensive testing of power, thermal, and radiation tolerance. They also note that the Starship’s payload bay is already crowded with other commercial and government payloads. „It’s a bold claim, but the engineering hurdles are non‑trivial,” said Dr. Elena Martinez, a space systems engineer at the University of Colorado.
SpaceX has responded by highlighting its rapid prototyping cycle and extensive use of simulation tools. The company has already built a ground‑based testbed that mimics the Starship’s environment. „We’re confident in our ability to meet the 2027 launch date,” the spokesperson said. The company also announced a partnership with a leading semiconductor manufacturer to produce radiation‑hardened components for the NVL72.
If successful, the Vera Rubin NVL72 could set a new standard for on‑orbit computing. Smaller agencies and private companies might be able to deploy their own data‑processing systems without building large, expensive supercomputers on Earth. The platform could also open up new business models, such as data‑as‑a‑service, where customers pay for real‑time analysis rather than raw data downloads.
SpaceX’s move also signals a shift toward more integrated space‑science solutions. By combining launch, power, and computing in a single package, the company could streamline mission planning and reduce costs. However, the industry will watch closely to see whether the promised performance and reliability can be delivered within the announced timeline.
The Vera Rubin NVL72 is a rack‑scale computing system designed for space missions. It fits into a standard 19‑inch rack and uses GPUs and FPGAs to process data in orbit.
SpaceX plans to launch the first NVL72 unit in early 2027, with a larger version following in 2028.
The system can process large volumes of data on‑orbit, reducing the need for high‑bandwidth data transmission and speeding up scientific analysis.