Dongfang Starlink Launches AI Hyperspectral Satellites, Plans 2M Lunar Robots for Data Center
新智元 · wechat · 2026-08-06
Building a Planetary Physical AI
Dongfang Starlink recently launched the Dongfang Huiyan hyperspectral AI satellites (05/06). Unlike standard remote sensing, its 22 spectral channels act like 'fingerprints' to identify material composition and health. Equipped with 400TOPS of onboard AI compute, the satellites perform in-orbit analysis to directly output actionable insights, such as crop disease risks or water anomalies.
From Earth Data to Lunar Compute
The founder argues that internet data is insufficient for models to understand the physical world, hence treating Earth as the ultimate AI training ground. The long-term roadmap includes landing autonomous micro-robots on the moon around 2030. The ultimate goal is to deploy 2 million AI robots networked via MESH to form a modular, mobile 'cosmic data center' on the lunar surface by 2035.
A New Window for Extraterrestrial Life
The team is currently simulating lunar environments in Earth's deserts to train the robots. The ultimate vision is to use spectral data of Earth's evolution to train a model explaining why Earth has life, and use it to search for extraterrestrial biology.
More from Embodied
- CoRL 2026 Opens Call for Demos with New Fast Track for Accepted Papers — GeorgiaChal · 2026-08-06
- 51WORLD to Unveil Physical AI Master Plan 2030 on August 18 — rsasaki0109 · 2026-08-06
- Developer Shows Sneak Peek of New Vibescape Environment for Apple Vision Pro — dreamwieber · 2026-08-06
- 100k Hours of Real-World Data Yields Only 57% Success in Robotics — eigenron · 2026-08-06
- USTC Study: AI Agents Achieve Only 3.3% Zero-Intervention Execution in Robot Labs — 新智元 · 2026-08-06
- Ego2Robot: Synthesizing 18,500+ Hours of Robot Training Data from Egocentric Videos — Ye Wang · 2026-08-06