AI News AI资讯 3d ago Updated 3d ago 更新于 3天前 46

Beidou Veterans Startups to Build Space Data Center, 'Interstellar Origin' Completes Two Rounds of Financing Worth Tens of Millions 北斗人创业,要建太空数据中心,「星际原点」连续完成两轮数千万元融资|36氪首发

Interstellar Origin completed seed and angel rounds totaling tens of millions of RMB, backed by investors including Jiuhe Capital and Meihua Venture Capital. The company focuses on building space-based computing satellites, leveraging core team experience from the Beidou-3 navigation constellation project. Their flagship product, "Xuanji-1," is a 500kg low-Earth orbit satellite designed to host 9-16 GPU payloads with 7-10kW power capacity. Key technical breakthroughs include large-caliber comput 太空算力初创公司“星际原点”完成数千万元种子轮及天使轮融资,投资方包括九合创投、梅花创投及上海国投旗下基金。 核心团队源自北斗三号及哈工大,具备国家级卫星工程研制经验,创始人孙小雷拥有15年以上航天工程背景。 旗舰产品“璇玑1号”为500公斤级低轨卫星,设计功率7-10千瓦,可搭载9-16台算力载荷,目标实现10P算力并在轨运行轻量大模型。 公司聚焦解决太空算力关键技术瓶颈,如大算力载荷、低成本柔性太阳翼及主动热控系统,并规划通过中轨通信星实现用户接入。 商业化路径初期争取年内落地订单以验证技术,长期计划自建数百至上千颗低轨算力星座,参考地面数据中心模式运营。

65
Hot 热度
70
Quality 质量
60
Impact 影响力

Analysis 深度分析

TL;DR

  • Interstellar Origin completed seed and angel rounds totaling tens of millions of RMB, backed by investors including Jiuhe Capital and Meihua Venture Capital.
  • The company focuses on building space-based computing satellites, leveraging core team experience from the Beidou-3 navigation constellation project.
  • Their flagship product, "Xuanji-1," is a 500kg low-Earth orbit satellite designed to host 9-16 GPU payloads with 7-10kW power capacity.
  • Key technical breakthroughs include large-caliber computing payloads, low-cost flexible solar wings, and active thermal control systems using fluid loops and deployable radiators.
  • The business strategy involves initial B2B orders for real-time remote sensing data processing, with a long-term vision of operating a massive constellation of hundreds to thousands of satellites.

Why It Matters

This development marks a significant step in the convergence of aerospace engineering and artificial intelligence, validating the emerging "space computing" sector beyond theoretical concepts. For AI practitioners and investors, it highlights a potential solution to future terrestrial energy and cooling constraints for data centers, while offering unique advantages in latency and security for global applications.

Technical Details

  • Satellite Specifications: The "Xuanji-1" satellite weighs approximately 500kg, operates in low Earth orbit, and supports a power range of 7-10kW, enabling the deployment of 9-16 high-performance computing units.
  • Core Technologies: The team has successfully developed critical subsystems including large-caliber computing payloads, low-cost large-area flexible solar panels, and an active thermal management system combining fluid circuits with deployable radiators.
  • Computing Capability: The system aims to run lightweight large language models directly in space (approx. 10P computing power initially), eliminating the need to transmit raw data back to Earth for processing.
  • Architectural Vision: The long-term plan involves a multi-layered architecture comprising hundreds to thousands of low-orbit computing satellites and medium-orbit communication satellites to facilitate user access and data relay.

Industry Insight

  • Cost Reduction as a Prerequisite: The viability of space data centers hinges on drastic cost reductions in both launch (targeting $10k/kg via reusable rockets) and satellite manufacturing (targeting ~10 million RMB per unit). Early players must prioritize engineering efficiency over immediate profitability.
  • Strategic Security Value: Beyond computational economics, space-based infrastructure offers critical resilience against terrestrial physical attacks, making it a strategic asset for national security and disaster recovery scenarios, similar to the role of Starlink in conflict zones.
  • First-Mover Advantage in Ecosystem: While the technology is nascent, establishing early partnerships with remote sensing agencies and defining operational standards for space-to-ground AI services will create significant barriers to entry for latecomers.

TL;DR

  • 太空算力初创公司“星际原点”完成数千万元种子轮及天使轮融资,投资方包括九合创投、梅花创投及上海国投旗下基金。
  • 核心团队源自北斗三号及哈工大,具备国家级卫星工程研制经验,创始人孙小雷拥有15年以上航天工程背景。
  • 旗舰产品“璇玑1号”为500公斤级低轨卫星,设计功率7-10千瓦,可搭载9-16台算力载荷,目标实现10P算力并在轨运行轻量大模型。
  • 公司聚焦解决太空算力关键技术瓶颈,如大算力载荷、低成本柔性太阳翼及主动热控系统,并规划通过中轨通信星实现用户接入。
  • 商业化路径初期争取年内落地订单以验证技术,长期计划自建数百至上千颗低轨算力星座,参考地面数据中心模式运营。

为什么值得看

本文揭示了“太空算力”这一新兴交叉领域的早期商业化探索,展示了如何将地面AI算力需求延伸至太空基础设施。对于关注商业航天、AI算力边界拓展及卫星互联网应用的从业者而言,提供了关于技术可行性、成本控制逻辑及长期运营模式的直观案例。

技术解析

  • 核心技术与架构:旨在将地面数据中心的GPU算力板卡迁移至卫星,实现大语言模型在太空的直接部署与运行,而非简单的边缘计算。通过星上处理遥感数据,减少下行传输带宽压力,实现实时响应。
  • 硬件规格与性能:旗舰产品“璇玑1号”重约500公斤,位于低轨,设计功率7-10千瓦。单星算力目标为10P,计划搭载9-16台算力载荷。后续迭代需攻克100P算力所需的柔性辐射器、大惯量卫星控制及星地激光通信等技术。
  • 关键子系统攻关:重点突破低成本大型柔性太阳翼以保障供电,以及流体回路+展开辐射器主动热控系统以解决太空高算力带来的散热难题。同时规划建设中轨通信星,构建“算力+通信”的完整接入链路。
  • 成本与经济模型:当前卫星成本约3000万元/颗,发射成本6-7万元/公斤;目标通过技术迭代将卫星成本降至1000万元左右,发射成本随火箭回收技术降至1万元/公斤。通过建立测算模型,结合未来5年AI算力增长预测,反推太空数据中心的经济可行性。

行业启示

  • 算力基础设施的空间延伸:随着AI算力需求呈指数级增长,地面数据中心面临能源、土地及网络延迟瓶颈,太空数据中心可能成为未来算力供给的重要补充,特别是在高带宽、低延迟要求的特定场景(如遥感实时处理)。
  • 航天工程能力的商业化转化:北斗三号等国家级工程积累的“低成本、高质量、快速交付”能力及供应链体系,是商业航天公司切入高门槛赛道(如太空算力)的核心竞争力,证明了体制内技术团队向市场化转型的潜力。
  • 安全与战略价值驱动:除了经济效益,太空算力在极端情况下的生存能力(如战时地面设施易受损)具有显著的战略安全价值。这种双重属性(商业+安全)可能促使更多政策支持和长期资本投入该领域。

Disclaimer: The above content is generated by AI and is for reference only. 免责声明:以上内容由 AI 生成,仅供参考。

Funding 融资 Product Launch 产品发布 Research 科学研究