From "Connected" to "Understanding": Space-Based Communication and Computing Startup Bets on Unmanned Systems, Raises Millions in First Round | 36Kr Exclusive 从“连得上”到“算得懂”，天基通算融合初创公司押注无人系统，首轮融资数千万元｜36氪首发

TL;DR

• Xinglian Tianshu secured tens of millions in angel funding to develop integrated satellite communication and computing solutions specifically for unmanned systems.
• The company employs a "ground data, space computing" model, processing AI decisions on satellites to reduce latency and bandwidth costs compared to transmitting raw data.
• Core products include a lightweight 2kg Ka-band satellite terminal and heterogeneous computing payloads designed for specific sensor data rather than general-purpose GPU clusters.
• The strategy targets immediate commercial viability in sectors like emergency management and autonomous vehicles, contrasting with long-term "space data center" approaches.
• Market explosion is projected in 3-5 years, pending the maturation of low-orbit constellations and the establishment of pricing models for satellite data traffic.

Why It Matters

This development highlights a pragmatic shift in the space economy from infrastructure deployment to application-layer monetization, demonstrating how satellite internet can solve critical connectivity gaps for autonomous systems where terrestrial networks fail. For AI practitioners, it underscores the growing importance of edge computing in space and the engineering challenges of optimizing heterogeneous hardware for specific data types under strict power and thermal constraints.

Technical Details

• Architecture: Utilizes a heterogeneous computing architecture on satellites, avoiding pure GPU reliance to optimize for power, weight, and specific sensor data types (optical, infrared, SAR, LiDAR).
• Hardware Specifications: Features a 2kg Ka-band broadband satellite communication terminal, noted as one of the lightest domestically available for unmanned systems.
• Data Flow: Implements a "ground data, space computing" pipeline where raw data is transmitted to satellites for AI processing and decision-making, with only results returned to ground stations.
• Development Roadmap: Plans to launch terminal prototypes in September, complete ground-based verification systems within the year, and conduct space-ground joint experiments next year.

Industry Insight

• Commercial Viability: Focusing on high-value, rigid-demand scenarios like unmanned systems allows for faster business loop closure compared to building full-scale space data centers, which face significant technical and cost barriers.
• Infrastructure Dependency: The success of such services is tightly coupled with the commercialization of low-orbit constellations; investors should monitor the rollout timelines and pricing structures of major satellite internet projects.
• Competitive Differentiation: Companies that tailor hardware and software to specific data characteristics (heterogeneous computing) may outperform those attempting generic high-performance computing solutions in the resource-constrained space environment.