Space Foundation's Pringle Targets Orbital Data Centers in Five to Seven Years
Ai4
Software-defined satellite constellations and AI foundation models like NASA's Prithvi are replacing static orbital hardware to handle massive space data volumes and counter debris risks.
## Background
Published by Ai4 on YouTube on August 22, 2026, this video features a fireside chat with Space Foundation CEO Heather Pringle and Financial Times correspondent Rafe Uddin. According to the page description, the discussion explores how AI serves as a force multiplier for space data, driving transformation across exploration, national security, global commerce, and the $1 trillion space economy.
## TL;DR
Retired Maj. Gen. Heather Pringle and Rafe Uddin discuss the central role of AI and software in modern space operations. Pringle explains how AI processes massive Earth observation datasets, improves spacecraft design, and enables real-time orbital applications like NASA's Prithvi model. The conversation covers the transition toward software-defined satellite constellations, commercial-first government procurement strategies, and next-generation capabilities such as space-based Moving Target Indication and orbital data centers. Finally, Pringle emphasizes the necessity of resilient space architectures and international partnerships to defend against kinetic debris and evolving orbital threats.
## Atmosphere & style
The chat maintains an authoritative, forward-looking, and intellectual tone, blending deep military command experience with sharp journalistic inquiry. Pringle speaks with polished expertise, conveying genuine enthusiasm for space exploration and AI integration while grounding the discourse in practical engineering and security realities. The interaction feels like an insider executive
## Key takeaways
- [00:04](https://www.youtube.com/watch?v=fsUIQncBliE&t=4s) Data and AI serve as force multipliers across global space operations, enhancing exploration, national security, and commercial ecosystems.
- [00:04](https://www.youtube.com/watch?v=fsUIQncBliE&t=4s) Space Foundation brings together civil space agencies, defense organizations, and commercial entities to collaborate on technology and policy.
- [00:04](https://www.youtube.com/watch?v=fsUIQncBliE&t=4s) AI was heavily integrated into mission management, launch, and splashdown operations during the Artemis program.
- [03:41](https://www.youtube.com/watch?v=fsUIQncBliE&t=221s) Initial AI applications in space focused on curating and structuring massive data volumes, such as NASA's Earth observation libraries.
- [03:41](https://www.youtube.com/watch?v=fsUIQncBliE&t=221s) Commercial space enterprises universally adopt AI for mission management, spacecraft design optimization, and conjunction avoidance to prevent orbital collisions.
- [03:41](https://www.youtube.com/watch?v=fsUIQncBliE&t=221s) NASA and IBM developed Prithvi, an open-source foundational geospatial AI model running in orbit to process decades of satellite data.
- [07:38](https://www.youtube.com/watch?v=fsUIQncBliE&t=458s) Space architecture is shifting from heavy, hardware-bound satellites to software-defined, interconnected satellite constellations.
- [07:38](https://www.youtube.com/watch?v=fsUIQncBliE&t=458s) Government agencies are adopting 'commercial-first' procurement strategies to keep pace with rapid commercial AI innovation.
- [09:40](https://www.youtube.com/watch?v=fsUIQncBliE&t=580s) The U.S. Space Force is transitioning Moving Target Indication (MTI) from aerial platforms to space for real-time dynamic tracking.
- [09:40](https://www.youtube.com/watch?v=fsUIQncBliE&t=580s) Orbital commercialization includes microgravity 3D printing, in-space pharmaceutical manufacturing, and private space stations.
- [09:40](https://www.youtube.com/watch?v=fsUIQncBliE&t=580s) Functional orbital data centers are projected within 5 to 7 years to bypass terrestrial power and land constraints, despite thermal and bandwidth challenges.
- [15:46](https://www.youtube.com/watch?v=fsUIQncBliE&t=946s) Modern space resiliency relies on high-attrition, multi-orbit constellation architectures to withstand space debris and anti-satellite threats.
- [15:46](https://www.youtube.com/watch?v=fsUIQncBliE&t=946s) High kinetic speeds exceeding 17,000 mph make orbital debris a severe risk requiring physical shielding for spacecraft.
- [15:46](https://www.youtube.com/watch?v=fsUIQncBliE&t=946s) International partnerships and global ground-station networks provide strategic redundancy and operational success for space infrastructure.
## Detailed walkthrough
### [00:04](https://www.youtube.com/watch?v=fsUIQncBliE&t=4s) Introduction and the Role of Space Foundation
Financial Times San Francisco correspondent Rafe Uddin opens the fireside chat by introducing Maj. Gen. Heather Pringle (ret.), Chief Executive Officer of Space Foundation. Uddin highlights her distinguished military career spanning over three decades in the U.S. Air Force, which included serving as Director of Strategic Plans at the Pentagon, commanding the Air Force Research Laboratory (AFRL), and managing a multi-billion-dollar science and technology budget alongside a team of over 10,000 personnel. Pringle thanks the audience for staying until the end of the event, noting that her career was dedicated to science and technology across both the Air Force and the Space Force.
After retiring three years prior, Pringle joined Space Foundation—a 40-year-old nonprofit organization operating at the intersection of technology, policy, workforce development, and global partnerships. Space Foundation brings together civil space agencies like NASA and international allies, defense organizations such as the U.S. Space Force, and a broad commercial ecosystem leveraging AI technologies. Expressing enthusiasm for recent achievements like the Artemis program, Pringle highlights how over 2,000 individuals worldwide contributed to building and managing the mission, which heavily integrated AI into its operations.
- **Speaker Background**: Maj. Gen. Heather Pringle (ret.), CEO of Space Foundation; former AFRL commander managing a multi-billion-dollar budget and 10,000+ personnel.
- **Organization Scope**: Space Foundation unites civil space agencies (NASA, allied programs), military sectors (U.S. Space Force), and commercial enterprise.
- **Artemis Program Scale**: Involved over 2,000 international contributors across Europe and the United States, utilizing AI applications during launch and splashdown operations.
[03:41](https://www.youtube.com/watch?v=fsUIQncBliE&t=221s) AI Capabilities, Earth Observation, and On-Orbit Processing
Addressing Uddin's question about strategic AI deployment in space, Pringle explains that initial applications focused heavily on curating, simplifying, and structuring raw data for operational decision-makers. Earth observation generates vast volumes of imagery and pixel data. For instance, a single NASA data library holds an volume of information comparable to the entire Library of Congress. Machine learning algorithms process this data for predictive analysis, change detection, and monitoring population growth and areas of global concern.
Pringle emphasizes that AI usage has expanded far beyond simple data curation. Industry adoption is universal, with every major space enterprise incorporating AI into mission management, spacecraft design optimization, and conjunction avoidance to prevent collisions in orbit. Highlighted during the discussion is NASA's rollout of "Prithvi," a foundational geospatial AI model developed in partnership with IBM. Running on orbit across satellite architectures, Prithvi synthesizes decades of Earth science data, accelerating data accessibility for researchers and unlocking new observational capabilities.
```
RAW SPACE DATA AI AGGREGATION & IN-ORBIT
(Earth Observation, Sensors) PROCESSING (e.g., Prithvi)
+----------------------------+ +--------------------------+
| NASA Data Libraries | | On-Orbit Foundation |
| (Library of Congress vol) | ------------> Models & Processing |
| Satellite Constellations | | Conjunction Avoidance |
+----------------------------+ +--------------------------+
|
v
OPERATIONAL OUTPUTS
+--------------------------+
| Predictive Analytics |
| Change Detection |
| Mission Management |
+--------------------------+
```
- **Data Volume Benchmark**: NASA's Earth observation libraries match the information volume of the entire Library of Congress.
- **Key Operational Uses**: Spacecraft design, autonomous mission management, on-orbit data processing, and collision prevention (conjunction avoidance).
- **Prithvi Foundation Model**: Open-source geospatial AI model designed to make decades of historical satellite data immediately actionable for scientists.
[07:38](https://www.youtube.com/watch?v=fsUIQncBliE&t=458s) Acquisition Shifts and Government Adoption Strategies
Discussing the operational shift within space architecture, Pringle outlines how space activities are transitioning from a historical reliance on heavy hardware to a software-defined, service-centric model. Rather than launching single satellites executing isolated orbital loops, modern strategies prioritize massive, interconnected constellations. Because hardware in orbit cannot easily be modified after launch, software-defined systems are crucial for upgrading assets continuously.
When addressing concerns regarding the speed of government procurement compared to rapid commercial AI releases, Pringle notes that agencies like NASA, the U.S. Space Force, the Federal Communications Commission (FCC), and the Federal Aviation Administration (FAA) are actively adopting "commercial-first" acquisition strategies. Rather than attempting to match the fast-moving commercial sector step-by-step, government agencies aim to leverage existing commercial innovation to maintain cutting-edge capabilities for national space assets.
- **Architecture Shift**: Transition from single hardware-bound satellites to software-defined, multi-satellite constellations.
- **Agency Participation**: NASA, U.S. Space Force, FCC, and FAA are framing commercial-first integration strategies.
- **Strategic Goal**: Fast-tracking advanced commercial software and AI tools into defense and civil space infrastructure.
[09:40](https://www.youtube.com/watch?v=fsUIQncBliE&t=580s) Next-Generation Missions and the Frontier of Orbital Data Centers
Uddin and Pringle explore emerging, novel capabilities being readied for deployment. While NASA's immediate focus centers on returning astronauts safely to the Moon via the Artemis program timeline, the Space Force is developing space-based Moving Target Indication (MTI). Historically restricted to aerial platforms like specialized aircraft, tracking moving targets from space represents a major shift away from static imagery toward real-time intelligence. Other emerging commercial initiatives include 3D printing in microgravity, in-space pharmaceutical manufacturing, and commercial space stations designed to replace legacy low-Earth orbit (LEO) outposts.
The discussion shifts to orbital data centers, a concept gaining traction due to terrestrial power, land, and water constraints. Pringle acknowledges the immense technical hurdles—particularly thermal dissipation and bandwidth constraints for downlinking data—but suggests that infinite solar energy makes the concept viable. In response to commercial filings from companies like SpaceX (filing for up to a million satellites) and other firms proposing constellations of 88,000 satellites, Pringle estimates functional orbital data centers could emerge within 5 to 7 years, heavily driven by economic incentives and heavy-lift launch vehicles like SpaceX's Starship.
- **Space-Based MTI**: Transitioning Moving Target Indication capabilities from aircraft to orbital platforms for dynamic tracking.
- **Orbital Data Center Projections**: Commercial models target vast constellations (from tens of thousands up to 1,000,000 satellites); operational timeline estimated at 5–7 years.
- **Key Engineering Barriers**: Spacecraft heat dissipation (cooling) and ground-to-space data bandwidth bottlenecks.
- **LEO Commercialization**: Microgravity manufacturing, orbit-based pharmaceutical development, and private space station infrastructure.
[15:46](https://www.youtube.com/watch?v=fsUIQncBliE&t=946s) Geopolitical Resiliency, Space Threats, and International Cooperation
In the final segment, Pringle discusses the evolving threat matrix facing space assets. For decades, space operated as a benign environment where satellites required minimal built-in resiliency against intentional threats or collision debris. Today, space assets face risks from space debris as well as anti-satellite mechanisms designed to disrupt or de-orbit platforms. Because space infrastructure underpins modern daily life—from commercial aviation to mobile connectivity—protecting these assets is a central focus for the Space Force.
To build system resiliency, engineers utilize high-attrition, multi-orbit constellation architectures where lost individual satellites do not trigger system-wide operational failure. Pringle notes that physical shielding is essential given that even small debris in orbit carries immense kinetic energy at speeds exceeding 17,000 mph. Pringle concludes by emphasizing that international partnerships—such as global ground-station networks for continuous data downloading and multi-nation human spaceflight initiatives—provide the ultimate layer of operational redundancy, capital distribution, and strategic success in space.
- **Orbital Velocities**: Orbital debris traveling at speeds around 17,000 mph poses catastrophic kinetic impact risks to spacecraft.
- **Resiliency Architecture**: Large-scale constellation designs built for attrition, multi-orbit redundancy, and hardened spacecraft hulls.
- **International Alignment**: Partnership strategies across data sharing, ground-station placement, and joint crewed spaceflight missions.
## Mentioned
Heather Pringle, Rafe Uddin, Space Foundation, Financial Times, U.S. Air Force, Pentagon, Air Force Research Laboratory, AFRL, U.S. Space Force, NASA, Artemis program, Library of Congress, IBM, Prithvi, Federal Communications Commission, FCC, Federal Aviation Administration, FAA, Moving Target Indication, SpaceX, Starship
## Video description
Heather Pringle - CEO, Space Foundation
Rafe Uddin - San Francisco Correspondent, Financial Times
Data has always been part of space activities. It is the ultimate ingredient by which planners and participants can take the next giant leaps for exploration, enhanced Earth and universe understanding, increased infrastructure and operating performance and ultimately critical mission success. With the rise of AI, all of these space data dividends have force multipliers that expand greater possibilities and impact as more countries, companies and communities engage with space.
As CEO of Space Foundation, Heather Pringle, Maj General, USAF, (ret.), PhD is a leading voice on the issues and environments that are shaping today’s and tomorrow’s global space ecosystem.
Today there are more than 90 countries with active space operations and more are looking to define and expand their place with space every day. Further expanding the aperture of space capabilities are the hundreds, if not thousands of space companies supplying critical services, providing capabilities, driving innovations – all impacting and benefiting life on Earth. But of all of the revolutionary technologies that are shaping our present and future with space, none of them will be more transformative than AI.
With the global space economy forecast to surpass $1Trillion in the early 2030’s, countries and companies will find themselves even more dependent on space reliant operations, architectures and providers for security, commerce, innovation, job creation, education and more. AI will be what distinguishes and transforms all of these activities at every imaginable level.
In this fireside chat, Heather will share those drivers and challenges and how they will shape our return to the Moon; support national security; expand global commerce and competition; and forever transform our connection to space.
Maj. Gen. Heather Pringle (ret.), CEO of Space Foundation, contends that space operations are transitioning from isolated, hardware-bound platforms to interconnected, software-defined systems accelerated by commercial artificial intelligence. To adapt, civil and military space agencies are moving toward commercial-first acquisitions and distributed, multi-orbit architectures built for operational resiliency.
On-Orbit AI Processing and Earth Observation (3:41)
Raw data management in space has moved beyond manual curation as satellite networks generate unprecedented data volumes. Pringle points out that a single NASA Earth observation library now holds an information volume comparable to the entire Library of Congress. To process these massive data streams for change detection, predictive analytics, and population monitoring, space enterprises are deploying machine learning models directly into orbit.
A key milestone highlighted is NASA and IBM's open-source geospatial foundation model, Prithvi, which synthesizes decades of Earth science data across satellite architectures to accelerate scientific access. Beyond data analysis, Pringle notes that major space operators now universally integrate AI into spacecraft design optimization, autonomous mission management, and conjunction avoidance algorithms to prevent collisions in crowded orbital environments.
Commercial-First Procurement and Software-Defined Satellites (7:38)
Space architecture is shifting away from historical reliance on heavy, individual satellites executing isolated orbital loops. Because physical hardware cannot be easily altered once launched, modern strategies prioritize massive, software-defined constellations that allow continuous capability upgrades through software patches.
To keep pace with commercial technology cycles, public agencies—including NASA, the U.S. Space Force, the Federal Aviation Administration (FAA), and the Federal Communications Commission (FCC)—are adopting "commercial-first" procurement frameworks. Pringle explains that rather than attempting to duplicate commercial research and development internally, government agencies are looking to buy and integrate off-the-shelf commercial AI software to modernise national security and civil space infrastructure.
Orbital Data Centers and Dynamic Target Tracking (9:40)
Next-generation space missions are expanding beyond static surveillance. The U.S. Space Force is developing space-based Moving Target Indication (MTI), moving a tracking capability historically restricted to specialized aircraft platforms directly into orbit to achieve real-time surface intelligence. Concurrently, commercial ventures are pursuing microgravity 3D printing, in-space pharmaceutical manufacturing, and private space stations to succeed legacy low-Earth orbit platforms.
Addressing terrestrial land, water, and power constraints, industry proposals envision hosting data centers in orbit, supported by continuous solar energy and heavy-lift launch vehicles like SpaceX's Starship. With commercial regulatory filings proposing constellations ranging from tens of thousands up to one million satellites, Pringle estimates that functional orbital data centers could emerge within five to seven years. However, she cautions that this deployment timeline remains constrained by unresolved engineering challenges, primarily thermal dissipation (cooling) in a vacuum and severe downlink bandwidth bottlenecks.
Orbital Threats and Architectural Resiliency (15:46)
The operating environment for space assets has transitioned from a benign sanctuary into a contested domain. Spacecraft face risks from anti-satellite technologies and orbital debris traveling at speeds around 17,000 mph, where impacts from tiny fragments can cause catastrophic kinetic destruction. Because space systems underpin civilian communications and defense functions, ensuring systemic survival is a primary focus for the U.S. Space Force.
To counter these vulnerabilities, satellite developers are abandoning single-point-of-failure designs in favor of high-attrition, multi-orbit constellation architectures that maintain continuity even if individual satellites are disabled. Pringle emphasizes that physical hull shielding must be paired with international alliances, which provide essential operational redundancy through globally distributed ground stations and shared crewed flight programs.
Pringle emphasizes that the space sector's future depends on rapid commercial integration and distributed constellation architectures. However, realizing frontier concepts such as orbital computing and securing critical space infrastructure against kinetic and orbital threats will require resolving significant physical cooling constraints, bandwidth limitations, and international operational coordination.
精炼阅读摘要
Space Foundation（太空基金会）首席执行官 Heather Pringle 指出，全球航天正在经历由软件定义架构与人工智能驱动的深层转型，政府与军用航天机构正通过“商业优先”采购加速整合私营技术，以应对激增的遥感数据处理需求与复杂的轨道威胁。
在轨人工智能与海量遥感数据处理 (3:41)
随着对地观测技术的普及，空间传感器产生的数据量呈现出爆炸式增长。Pringle 举例指出，仅 NASA 单个对地观测数据中心存储的信息体量就已相当于美国国会图书馆（Library of Congress）的全部藏书规模。早期的机器学习算法主要用于地面原始数据的整理、分类、变化监测以及对全球人口增长等关切领域的趋势预测。如今，人工智能的应用已全面扩展至任务自主管理、航天器结构优化设计以及规避空间碰撞的交会预警分析（conjunction avoidance）。
为了加速海量卫星数据的即时转化，航天领域正在向在轨运行的基础模型迈进。NASA 与 IBM 联合开发了名为“Prithvi”的开源地理空间基础 AI 模型，并将其部署于卫星星座架构中进行在轨测试。该模型通过综合梳理数十年来积累的地球科学观测数据，不仅大幅缩短了科研人员获取与调用高价值数据的时间，也标志着天基数据处理从单纯的地面被动接收转变为在轨自主合成。
软件定义架构与商业优先采购机制 (7:38)
航天装备正在打破传统上对重型专用硬件的依赖，转向以软件定义为核心、服务为导向的敏捷架构。Pringle 指出，以往发射入轨的单体卫星往往在孤立轨道上运行，且发射后受限于物理条件无法改装；现代航天则普遍采用互联的大型多星分布式星座。在硬件固化的客观限制下，软件定义模式允许系统在轨道寿命周期内通过算法OTA迭代与软件补丁持续升级作战和观测能力。
面对私营商业 AI 与软件技术的爆发式迭代，传统政府采办模式难以按部就班地跟上步伐。为此，NASA、美国太空军（U.S. Space Force）、联邦通信委员会（FCC）以及联邦航空管理局（FAA）等机构正积极推进“商业优先”（commercial-first）战略。官方机构不再试图由内部从零研发每一套系统，而是倾向于直接采纳商业领域的现成创新，将前沿算法与软件工具直接整合进国家防御与民用航天基础设施之中。
轨道数据中心构想与新型在轨任务 (9:40)
新一代航天任务正在拓展空间作业的边界。除了 NASA 阿尔忒弥斯（Artemis）计划聚焦宇航员重返月球外，美国太空军正在研发天基移动目标指示（MTI）能力，旨在将原先依赖侦察机等空中平台的动态追踪功能搬上轨道，实现对高动态目标的实时监控。与此同时，微重力 3D 打印、在轨生物制药以及旨在接替老旧近地轨道空间站的商业载人设施也进入实质筹备阶段。
受地面数据中心能耗、土地占用和水资源冷却瓶颈的制约，“轨道数据中心”概念引发了产业界的密集布局。商业备案显示，已有企业提出了多达 8.8 万颗卫星的构想，SpaceX 甚至提交了包含多达 100 万颗卫星的设想申请。尽管轨道拥有无尽的太阳能资源，但 Pringle 强调，真空环境下的航天器散热（thermal dissipation）困难与空地高通量数据下行带宽受限是当下关键的技术瓶颈。结合 SpaceX 星舰（Starship）等重型运力带来的成本下降曲线与经济驱动力，她预测可运行的轨道数据中心可能在 5 至 7 年内初见端倪。
空间碎片威胁与多轨星座韧性防护 (15:46)
现代社会的基础运转高度依附于天基基础设施，涵盖商业航空调度乃至日常移动网络通信，但空间运作环境已由以往的和平温床变为充满潜在对抗与碎片的复杂空域。Pringle 警告称，除针对卫星平台实施干扰或破坏的恶意反卫星手段外，空间轨道中还密布着大量运行速度超过每小时 17,000 英里（约每秒 7.6 公里）的高速碎片。在超高动能碰撞面前，即使极微小的碎片撞击也会造成灾难性损伤。
为确保空间体系在遭受打击或撞击时维持基本运转，系统设计正在向多轨道、具备高容损能力（high-attrition）的分布式弹性星座演进，确保个别卫星的损毁不致引发系统级崩溃，并在硬件层面强化物理屏蔽外壳。Pringle 强调，跨国合作是建立系统冗余的关键支柱，例如通过全球分布的地面测控站网络保障数据回传不断联，以及类似 Artemis 计划动员欧美逾 2,000 名国际团队成员联合攻关，以分散地缘风险与建设资本。
本场访谈展现了太空技术从硬件孤岛向“云+AI+多轨星座”深度融合的演进逻辑，但轨道数据中心在真空热控与空地传输上的工程瓶颈、百万级近地卫星引发的轨道拥堵与碰撞隐患，以及传统政府机构对颠覆性商业技术的真实吸收转化效率，仍是未来数年内有待验证的核心难题。
Detailed notes — zh
## Background
本视频由科技会议频道“Ai4”于 2026 年 8 月 22 日发布在 YouTube 平台。内容为《金融时报》旧金山特派记者雷夫·乌丁（Rafe Uddin）与太空基金会首席执行官海瑟·普林格尔（Heather Pringle）的炉边对谈，页面介绍重点阐述了在预计于 2030 年代初突破万亿美元的太空经济背景下，AI 如何作为核心力量重塑重返月球、国家安全与全球商业航天生态。
## TL;DR
对话系统梳理了太空产业从传统地面数据后处理向“软件定义”与全流程在轨 AI 赋能的范式演进，指出 AI 已深度融入航天器设计、遥感基础模型和轨道避撞系统。普林格尔分析了政府机构依托“商业优先战略”吸纳私营创新的趋势，并探讨了太空移动目标指示（MTI）与轨道数据中心等前沿技术，预测商业驱动的太空算力将在未来 5 至 7 年内取得实质突破。最后，针对高危轨道碎片与地缘对抗威胁，对谈强调了通过超大规模分布式星座吸收损耗的韧性设计，以及依托国际伙伴关系共建全球地面站网络的必要性。
## Atmosphere & style
整场对谈节奏明快且充满战略高度，记者提问精准敏锐，普林格尔以退役将领兼科技掌舵人的严谨务实、亲和幽默从容回应，两人互动默契自然。交流从晨起咖啡与卫星信号交互的生动日常切入，逐步升华至轨道算力集群与深空防御的宏大架构，在严峻的技术瓶颈与地缘风险剖析中始终贯穿着对万亿美元太空产业未来的理性乐观，现场科技变革氛围浓厚。
## Key takeaways
- [00:04](https://www.youtube.com/watch?v=fsUIQncBliE&t=4s) 太空产业正迎来爆发式增长，登月任务如阿尔忒弥斯（Artemis）汇集了全球数千名顶尖专业人才，AI 技术已深度渗透进任务规划与执行的各环节。
- [03:19](https://www.youtube.com/watch?v=fsUIQncBliE&t=199s) AI 在太空的应用范式已从早期的“地面遥感影像后处理与数据清洗”，跃升为覆盖航天器设计、轨道规划、在轨智能避撞的全生命周期赋能。
- [03:19](https://www.youtube.com/watch?v=fsUIQncBliE&t=199s) NASA 发布的 Prithvi 等遥感基础模型（Foundation Models）使在轨即时图像处理成为现实，大幅提升了对海量地球观测数据的利用效率。
- [07:43](https://www.youtube.com/watch?v=fsUIQncBliE&t=463s) 卫星架构正从以单一硬件为核心转向“软件定义”（Software-Defined）与服务导向，以克服硬件入轨后无法物理升级的限制并平衡星地带宽瓶颈。
- [07:43](https://www.youtube.com/watch?v=fsUIQncBliE&t=463s) 包括 NASA 和美国太空军在内的政府机构已全面推行“商业优先战略”（Commercial First Strategy），直接借力私营部门的快速 AI 创新而非从零自研。
- [10:43](https://www.youtube.com/watch?v=fsUIQncBliE&t=643s) 太空防御正将“移动目标指示”（MTI）从传统空中平台迁移至轨道，使卫星从静态拍摄跃升至对地空动态目标的实时连续追踪。
- [10:43](https://www.youtube.com/watch?v=fsUIQncBliE&t=643s) 轨道数据中心（Orbital Data Centers）能够利用近地轨道近乎无限的清洁太阳能，绕过地面算力面临的土地、电力与电网容量瓶颈。
- [13:14](https://www.youtube.com/watch?v=fsUIQncBliE&t=794s) 轨道数据中心落地的关键前提依赖 SpaceX Starship 等超重型可复用火箭实现发射成本的大幅下降，核心技术挑战集中在真空散热温控与高带宽下行。
- [13:14](https://www.youtube.com/watch?v=fsUIQncBliE&t=794s) 在强大商业资本与算力需求驱动下，太空数据中心预计将在未来 5 至 7 年内取得实质性的商业化突破与规模化部署。
- [15:46](https://www.youtube.com/watch?v=fsUIQncBliE&t=946s) 太空环境已从过去的平静良性转变为高危与对抗态势，轨道以高达每小时 17,000 英里运行的微小碎片即可对航天器造成毁灭性动能破坏。
- [15:46](https://www.youtube.com/watch?v=fsUIQncBliE&t=946s) 现代卫星系统韧性（Resiliency）设计依托大规模分布式星座与多轨道（Multi-Orbit）协同，通过吸收损耗实现单点故障下的全局任务连续性。
- [18:26](https://www.youtube.com/watch?v=fsUIQncBliE&t=1106s) 跨国伙伴关系与全球分布式地面站网络是分摊巨额太空基建成本、确保全天候低延迟数据通信和应对主权 AI 竞争的核心支柱。
## Detailed walkthrough
### [00:04](https://www.youtube.com/watch?v=fsUIQncBliE&t=4s) 嘉宾背景与太空基金会的 AI 时代使命
对谈在《金融时报》（Financial Times）旧金山特派记者雷夫·乌丁（Rafe Uddin）对海瑟·普林格尔（Heather Pringle）的介绍中拉开序幕。普林格尔曾在美国空军服役逾三十年，军衔至少将，曾任五角大楼战略规划主管以及空军研究实验室（AFRL）指挥官，统领过超万人的庞大团队与数十亿美元的科技研发预算。三年前退役后，她出任太空基金会（Space Foundation）首席执行官。普林格尔首先幽默地向坚持到会议最后一刻的现场听众致谢，并表示太空基金会成立 40 年来，始终致力于连接技术、政策、人才培养和伙伴关系，汇聚了 NASA、盟国民用太空机构、美国太空军（USSF）以及各类太空企业。
普林格尔在开场中提到了近期备受瞩目的阿尔忒弥斯（Artemis）登月任务发射与溅落回收。她与现场观众互动询问是否有人参与其中，并特别指出该任务汇聚了来自欧美超过两千名专业人员的协作，其中广泛部署了 AI 技术。乌丁也补充道，随着发射成本降低与商业火箭常态化，公众如今几乎每周都能看到火箭发射、回收、部署卫星星座的直播，Blue Origin、Rocket Lab 等商业公司与 NASA 及盟友机构紧密协作，整个太空生态正在以前所未有的速度升温。
- **海瑟·普林格尔资历**：退役美国空军少将（Ph.D.），前空军研究实验室指挥官，领导过 10,000 余人团队并管理数十亿美元科技预算。
- **太空基金会定位**：成立 40 年的非营利组织，横跨技术、政策、人才发展和产业联盟，协同 NASA、美国太空军及全球商业太空实体。
- **Artemis 任务规模**：全球 2,000 多名专业人员协同研发，AI 技术在任务各环节深度渗透。
---
### [03:19](https://www.youtube.com/watch?v=fsUIQncBliE&t=199s) 从地球观测数据洪流到全流程 AI 赋能
当被问及作为前战略决策者最关注 AI 在太空中的哪些方向时，普林格尔回顾了 AI 应用的演进路径。早期 AI 主要用于“数据治理与简化”，帮助操作员和决策者处理海量的地球观测（Earth Observation）影像数据。例如，NASA 单个数据库所包含的信息量就堪比整个美国国会图书馆，AI 能够快速进行变化检测、趋势预测并标定人口增长或重点关注区域。
然而，当前的太空产业已发生根本性跃迁。普林格尔直言：“现在没有任何一家太空公司不在使用 AI。”从在无重力环境下进行更多元化的航天器结构设计，到任务调度管理以及防止轨道碰撞的“交会避撞”（conjunction avoidance），AI 正全面融入全生命周期。乌丁提到 NASA 近期推出的 Prithvi 等遥感基础模型（Foundation Models），实现了在轨卫星图像的即时处理。普林格尔指出，这类模型整合了数十年沉淀的历史数据并开放给科学界，不仅大幅提升了效率，更为人类以全新方式探索和理解地球打开了空间。
```mermaid
flowchart TD
accTitle: AI 在太空领域的技术演进
subgraph S1["早期阶段：地面数据后处理"]
A["海量地球观测数据<br/>(像素/遥感影像)"] --> B["地面 AI 清洗与变化检测<br/>(简化与辅助决策)"]
end
subgraph S2["现阶段：端到端与在轨赋能"]
C["AI 辅助航天器设计<br/>与任务轨道规划"] --> D["星载基础模型<br/>(如 NASA Prithvi)"]
D --> E["在轨边缘计算与智能避撞<br/>(Conjunction Avoidance)"]
end
B -.->|范式演进| C
```
---
### [07:43](https://www.youtube.com/watch?v=fsUIQncBliE&t=463s) “软件定义”转型与政府的商业优先战略
在探讨政府与私营部门在 AI 竞速中的角色时，乌丁指出私营领域几乎每周都在迭代新模型，而政府机构常被诟病决策迟缓。对此，普林格尔阐述了太空架构底层的范式转移：太空领域过去数十年严重依赖硬件，但如今正快速转向以服务为中心（service-centric）。在单颗卫星演进为大规模卫星星座的背景下，核心考量变成了“星载处理能力与下行链路带宽的权衡”，而发射入轨的硬件无法物理升级，唯有依靠“软件定义”（software-defined）能力来实现持续更新。
针对政府节奏的问题，普林格尔澄清道，包括 NASA、太空军、FCC 以及 FAA 在内的各联邦机构目前普遍采取了“商业优先战略”（Commercial First Strategy）。面对飞速演进的 AI 技术，政府机构的目标并非从零自主构建去追赶私营企业，而是制定策略直接借力商业生态系统的快速创新，从而确保国家资产始终处于技术最前沿。对于 NASA 而言，当前核心是确保按期且安全地实现重返月球并返回；而太空军则更侧重于在对抗性环境中加快能力部署。
- **太空架构范式转移**：从单一硬件卫星向大规模星座与“软件定义”在轨可升级架构演进。
- **星地算力权衡**：重点解决在轨数据处理比例、下行通信带宽限制与实时性需求之间的平衡。
- **商业优先战略（Commercial First Strategy）**：政府机构直接吸纳私营部门顶尖 AI 成果，降低自主研发滞后风险。
---
### [10:43](https://www.youtube.com/watch?v=fsUIQncBliE&t=643s) 太空移动目标指示（MTI）与轨道数据中心的前景
谈及尚未大规模商用但极具潜力的前沿技术时，普林格尔指出了若干关键突破方向。在国防安全领域，太空军正推进将“移动目标指示”（Moving Target Indication, MTI）能力从传统空中平台（如侦察机）迁移到太空轨道。过去的卫星主要提供静态图像，而太空 MTI 将赋予轨道资产实时追踪动态目标的能力，彻底颠覆传统的太空侦察范式。在商业领域，商业空间站的加速研发正在催生新的微重力制造市场，涵盖 3D 打印制造和高精尖生物医药合成。
此外，对话转向了当下引发热烈讨论的“轨道数据中心”（Orbital Data Centers）。面对乌丁对其可行性与落地周期的提问，普林格尔表示虽然轨道数据中心在技术上仍需攻关，且面临极高的工程挑战（例如有企业构想发射多达 88,000 颗卫星组成太空算力网络），但绝不能低估其前景。地面数据中心受到土地、电力和冷却资源的严苛限制，而太空拥有取之不尽的太阳能资源，随着算力需求爆炸，将算力节点部署在近地轨道正成为一种极具潜力的破局路径。
- **太空移动目标指示（Space-based MTI）**：打破卫星仅能拍摄静态画面的局限，实现从轨道对地表/空中动态目标的连续追踪。
- **微重力轨道制造**：基于商业空间站的在轨 3D 打印与尖端生物医药制备。
- **轨道数据中心（Orbital Data Centers）**：
- 核心驱动：绕过地面土地占地、电网负荷与散热瓶颈，利用太空无限太阳能。
- 技术现状：属于高难度前沿课题，已有商业方案规划多达 88,000 颗卫星的星座集群。
### [13:14](https://www.youtube.com/watch?v=fsUIQncBliE&t=794s) 太空数据中心的优势、技术瓶颈与落地时间表
在轨数据中心具备独特的天然优势，尤其是能够利用太空中取之不尽的太阳能资源，但要真正将其变为现实，必须攻克下行数据传输、地空高带宽通信机制以及空间碎片防护等一系列难题。随着驱动 AI 智能体（AI Agents）所需算力的指数级爆发，无论是在航天探索还是商业应用领域，建设更多数据中心已成为刚性需求。当主持人提到一些公司规划发射多达 88,000 颗卫星时，指出这个数字在当下甚至显得有些保守——SpaceX 等行业巨头以及全球多个国家已经提交了数以百万计的卫星发射申请。这使得轨道空间变得前所未有地拥挤，而整个架构的实现深度依赖于发射能力的突破，尤其是像 SpaceX Starship 这样的重型可重复使用运载火箭，只有大幅降低入轨成本，才能经济高效地组装和部署庞大的轨道数据中心。
除了运载能力，冷却系统（Cooling）是目前在轨运行高性能算力面临的最核心技术瓶颈。当被问及何时能真正看到太空数据中心成熟运转时，Heather Pringle 笑着透露午餐时就有人问过这个问题，作为一名喜欢关注前沿科技发展的科学家，她预测商业驱动将在未来 5 到 7 年内极大加速这一进程。主持人也补充道，随着商业模式的转变，不仅初创企业开始转向在轨算力服务，包括 Blue Origin 也在积极探索轨道数据中心架构，太空算力正在形成不可逆转的发展势头。
```mermaid
flowchart TD
accTitle: 太空数据中心的发展驱动与关键挑战
subgraph Drivers["核心发展驱动"]
D1["AI 智能体算力激增"]
D2["无限太空太阳能资源"]
D3["可重复使用重型运载降本"]
end
subgraph Challenges["核心工程与运营挑战"]
C1["散热与温控瓶颈"]
C2["地空超高带宽与下行链路"]
C3["超高速微小碎片防护"]
end
subgraph Architecture["轨道韧性架构"]
A1["超大规模星座容错 (吸收损耗)"]
A2["多轨道异构协同 (Multi-Orbit)"]
A3["全球分布式地面站网络"]
end
Drivers --> Architecture
Challenges --> Architecture
```
- **太阳能优势**：太空可提供无日夜交替中断、近乎无限的清洁太阳能供应。
- **关键技术瓶颈**：高发热芯片在真空环境下的被动与主动散热机制、海量数据下行带宽。
- **商业与运载支撑**：依靠 Starship 等完全可复用火箭提供规模化、可承受的入轨成本。
- **落地时间预测**：在强大商业资本与经济利益驱动下，预计将在 5 至 7 年内取得实质性突破。
---
### [15:46](https://www.youtube.com/watch?v=fsUIQncBliE&t=946s) 从平静空间到高危环境：太空防御与韧性系统设计
过去的数十年里，太空一直被视为一个相对“良性且平静”的环境，早期发射的卫星和载荷基本不需要在设计中考虑过多的冗余度和抗打击能力，唯一的外部风险几乎仅来自于恶劣的太空天气。然而今天的地缘政治与轨道态势发生了剧烈变化，密集的太空碎片以及潜在的轨道机动反卫星能力构成了现实威胁。现代社会对太空基础设施的依赖已深入骨髓——绝大多数普通人在早晨喝第一杯咖啡之前，手机就已经调用了大约 20 次太空资产与卫星信号。太空已然成为全球商业与通信的基石，因此太空军（Space Force）的首要任务之一就是保卫这些资产并确保任务连续性。
这种外部威胁的转变倒逼航天工程从底层彻底重构设计理念。Heather Pringle 强调，在轨防御与韧性设计（Resiliency）主要从物理防护与架构分布两个维度展开：
1. **物理抗冲击防护**：在轨道速度下，即使是一枚小小的硬币大小的微小碎片，以高达每小时 17,000 英里的速度撞击航天器，其释放的动能也是毁灭性的，因此关键部件必须具备极高的抗撞击耐受能力；
2. **分布式星座容错**：现代巨型星座之所以部署成千上万颗卫星，核心逻辑之一就是“吸收损耗”（Attrition）。通过多星互联与多轨道（Multi-orbit）架构，单颗甚至部分卫星的失效或被击毁不会瘫痪整体系统，冗余网络能即时绕过故障节点保持业务不中断。
- **轨道动能冲击**：以 17,000 mph（约 27,358 km/h）运行的极微小物体即可摧毁航天器。
- **日常依赖度**：现代日常生活与商业体系对卫星导航、定时与通信高度依赖，清晨即可达数十次交互。
- **系统级韧性设计**：通过超大规模去中心化星座与多轨道层级部署，确保单点失效不影响全局连续性。
---
### [18:26](https://www.youtube.com/watch?v=fsUIQncBliE&t=1106s) 主权 AI 争论与国际联盟合作生态
在谈及主权 AI（Sovereign AI）以及大国在太空领域的竞争与合作时，Heather Pringle 指出，面对日益拥挤和复杂的太空环境，跨国联盟与伙伴关系本身就是增强系统韧性与冗余度的关键路径。单靠任何一个国家都无法在浩瀚、高成本且高风险的太空中独自承担所有基础设施建设。
美国与其盟友正在多个层面上深化合作，不仅体现在像阿耳忒弥斯（Artemis）计划这样具有里程碑意义的国际载人航天与深空探索任务中，更广泛地渗透在日常的数据共享机制以及遍布全球的地面站接收网络建设上。通过在全球范围内分布式建立地面站，能够确保海量卫星数据下载与地空通信的连续性与稳定性。在对谈接近尾声时，主持人 Rafe Uddin 感叹话题意犹未尽，Heather Pringle 随后向全场观众致谢，并祝愿参会者在后续的 Ai4 大会活动与交流中收获充实。
- **多边合作构建韧性**：跨国伙伴关系是分摊巨额资本开支、扩大全球基础设施覆盖的关键方式。
- **深空与人类探索标杆**：Artemis 计划展示了多国联合开展载人航天与深空探索的协作范式。
- **全球分布式地面网络**：联合盟友在全球范围内建立接收站，保障全天候、不间断的低延迟数据下行与链路连通。
## Notable quotes
> [03:19](https://www.youtube.com/watch?v=fsUIQncBliE&t=199s) 现在没有任何一家太空公司不在使用 AI。 — 海瑟·普林格尔
> [03:19](https://www.youtube.com/watch?v=fsUIQncBliE&t=199s) 这类模型整合了数十年沉淀的历史数据并开放给科学界，不仅大幅提升了效率，更为人类以全新方式探索和理解地球打开了空间。 — 海瑟·普林格尔
> [07:43](https://www.youtube.com/watch?v=fsUIQncBliE&t=463s) 太空领域过去数十年严重依赖硬件，但如今正快速转向以服务为中心（service-centric）。在单颗卫星演进为大规模卫星星座的背景下，核心考量变成了“星载处理能力与下行链路带宽的权衡”，而发射入轨的硬件无法物理升级，唯有依靠“软件定义”（software-defined）能力来实现持续更新。 — 海瑟·普林格尔
> [07:43](https://www.youtube.com/watch?v=fsUIQncBliE&t=463s) 面对飞速演进的 AI 技术，政府机构的目标并非从零自主构建去追赶私营企业，而是制定策略直接借力商业生态系统的快速创新，从而确保国家资产始终处于技术最前沿。 — 海瑟·普林格尔
> [10:43](https://www.youtube.com/watch?v=fsUIQncBliE&t=643s) 地面数据中心受到土地、电力和冷却资源的严苛限制，而太空拥有取之不尽的太阳能资源，随着算力需求爆炸，将算力节点部署在近地轨道正成为一种极具潜力的破局路径。 — 海瑟·普林格尔
> [13:14](https://www.youtube.com/watch?v=fsUIQncBliE&t=794s) 商业驱动将在未来 5 到 7 年内极大加速这一进程。 — 海瑟·普林格尔
> [15:46](https://www.youtube.com/watch?v=fsUIQncBliE&t=946s) 绝大多数普通人在早晨喝第一杯咖啡之前，手机就已经调用了大约 20 次太空资产与卫星信号。 — 海瑟·普林格尔
> [15:46](https://www.youtube.com/watch?v=fsUIQncBliE&t=946s) 在轨道速度下，即使是一枚小小的硬币大小的微小碎片，以高达每小时 17,000 英里的速度撞击航天器，其释放的动能也是毁灭性的，因此关键部件必须具备极高的抗撞击耐受能力。 — 海瑟·普林格尔
> [15:46](https://www.youtube.com/watch?v=fsUIQncBliE&t=946s) 现代巨型星座之所以部署成千上万颗卫星，核心逻辑之一就是“吸收损耗”（Attrition）。通过多星互联与多轨道（Multi-orbit）架构，单颗甚至部分卫星的失效或被击毁不会瘫痪整体系统，冗余网络能即时绕过故障节点保持业务不中断。 — 海瑟·普林格尔
> [18:26](https://www.youtube.com/watch?v=fsUIQncBliE&t=1106s) 面对日益拥挤和复杂的太空环境，跨国联盟与伙伴关系本身就是增强系统韧性与冗余度的关键路径。单靠任何一个国家都无法在浩瀚、高成本且高风险的太空中独自承担所有基础设施建设。 — 海瑟·普林格尔
## Mentioned
海瑟·普林格尔 (Heather Pringle), 雷夫·乌丁 (Rafe Uddin), 太空基金会 (Space Foundation), 金融时报 (Financial Times), NASA, 美国太空军 (USSF), 美国空军研究实验室 (AFRL), 阿尔忒弥斯任务 (Artemis), 蓝色起源 (Blue Origin), 火箭实验室 (Rocket Lab), SpaceX, 星舰 (Starship), Prithvi 基础模型, 美国联邦通信委员会 (FCC), 美国联邦航空管理局 (FAA), Ai4
## Video description
Heather Pringle - CEO, Space Foundation
Rafe Uddin - San Francisco Correspondent, Financial Times
Data has always been part of space activities. It is the ultimate ingredient by which planners and participants can take the next giant leaps for exploration, enhanced Earth and universe understanding, increased infrastructure and operating performance and ultimately critical mission success. With the rise of AI, all of these space data dividends have force multipliers that expand greater possibilities and impact as more countries, companies and communities engage with space.
As CEO of Space Foundation, Heather Pringle, Maj General, USAF, (ret.), PhD is a leading voice on the issues and environments that are shaping today’s and tomorrow’s global space ecosystem.
Today there are more than 90 countries with active space operations and more are looking to define and expand their place with space every day. Further expanding the aperture of space capabilities are the hundreds, if not thousands of space companies supplying critical services, providing capabilities, driving innovations – all impacting and benefiting life on Earth. But of all of the revolutionary technologies that are shaping our present and future with space, none of them will be more transformative than AI.
With the global space economy forecast to surpass $1Trillion in the early 2030’s, countries and companies will find themselves even more dependent on space reliant operations, architectures and providers for security, commerce, innovation, job creation, education and more. AI will be what distinguishes and transforms all of these activities at every imaginable level.
In this fireside chat, Heather will share those drivers and challenges and how they will shape our return to the Moon; support national security; expand global commerce and competition; and forever transform our connection to space.
[Source](https://www.youtube.com/watch?v=fsUIQncBliE)
[0:04] I'll introduce Heather. So obviously Heather
is the chief exec of the Space Foundation and
[0:09] a retired US Air Force major general. Sort
of three decades, more than three decades
[0:15] uh in service. Um served as director
of strategic plans at the Pentagon. Uh
[0:20] commanded the Air Force Research Lab and oversaw
a multi-billion dollar science and technology
[0:27] uh budget. Um and a quite sizable team as well.
I think you led more than 10,000 people. Sure.
[0:34] Um back in the day. But obviously it'd be great
to get a sense initially from you what you're
[0:40] currently working on. So tell us a bit more about
the space foundation. Well uh first I just want to
[0:45] say thank you all for staying till the very bitter
end. So, I appreciate you being here. And yes, I
[0:52] had three decades doing science and technology in
the Air Force for the Air Force and for the Space
[1:00] Force. And when I retired 3 years ago, I had the
really great opportunity to join a nonprofit that
[1:09] is really exciting because it is the intersection
of technology and policy and workforce development
[1:17] and partnerships. So I really enjoy all aspects
of it. Uh the space foundation has been around for
[1:25] 40 years and uh we bring together all sectors of
the space community. It includes NASA and all our
[1:36] uh allies around the world who have civil
space programs as well as the United States
[1:42] Space Force and a whole industry out there that
is uh leveraging AI technologies. And I'm excited
[1:51] to really tell you about how they're doing it, how
AI is powering what we've been achieving in space,
[1:58] especially recently. Uh, how many of you watched
the Aremis launch or splashdown? Wasn't that
[2:05] cool? Uh, but definitely, yeah, absolutely. Did
anybody uh help build it or contribute to it to
[2:13] the mission? One person or your son. Awesome. uh
that it's amazing how many people were part of
[2:23] this mission over 2,000 individuals around
the world from Europe to United States and
[2:30] uh lots of AI employed there as well. So so much
to talk about Rafe I really appreciate you having
[2:36] me here on stage. Yeah I mean I feel like is you
know was this huge moment and increasingly you
[2:43] and I have been talking about how you know
we had the SpaceX IPO quite recently. Yeah,
[2:49] seemingly every week now we get to watch a live
stream of a rocket getting launched to space,
[2:55] landing again safely, deploying satellites
in space, building up massive constellations,
[3:00] competitors like Blue Origin Rocket Labs and also
all working in partnership with both NASA but also
[3:08] other allied agencies as well. And it's just a
remarkable thing that you know I mean I think
[3:14] in the past year this topic has really sort of
live and dark. And so I'm curious I mean when
[3:19] we think about you know the the application of
AI when it comes to sort of space you given that
[3:26] you had a role as an operator somebody who made
strategic decisions you know where would you be
[3:34] looking right now? What what are the areas that
you're most intrigued by? Yeah. Um so the initial
[3:41] applications for AI in space really centered on
curating the data simplifying it making it usable
[3:49] for operators and decision makers. So if you think
about um earth observation and how much data is
[3:57] out there in terms of pictures and pixels um I
think one of the NASA uh libraries is carries as
[4:06] much information as the whole library of congress.
though there's a massive amount of information
[4:13] about the earth that AI has been simplifying
and predicting and looking for change detection
[4:21] and pointing individuals into areas of concern
and growth and population and so that's really
[4:28] been an early initial step but I have been at the
space foundation for three years and engaged with
[4:36] all kinds of industry and there's not a single
space company out there that isn't using AI and
[4:44] they're using it across the board and we're really
at the cusp of it taking off from everything for
[4:51] uh design of spacecraft cuz when you're in space
you have a lot more options and how you design it
[4:57] for mission management for conjunction avoidance
which means you know no crashing in space um and
[5:06] all all kinds of things. So, um there's there's so
much uh but um you know, I'm really excited about
[5:15] where it's going and how it's moving beyond just
space providing data for AI and having AI really
[5:23] uh advance and accelerate the capabilities, the
missions that we're taking and allowing us to go
[5:29] further in exploration. Yeah. And we recently saw
so NASA's team started developing or discussing
[5:36] the pry foundation models that they're deploying
as well. You know that that is sort of satellite
[5:42] imagery using foundation models to process
in orbit uh on two two satellite uh crafts.
[5:51] What does that sort of demonstrate about the
capabilities and applications that that NASA is
[5:57] currently building as well? Uh NASA's all in on
artificial intelligence as are you know all the
[6:03] space companies and um that one was so exciting
because it it took decades of data and made it
[6:12] uh available for uh users and scientists to really
take the next step. And so um I you know I was
[6:21] really excited and it just barely released, right?
So that's been that's been really good. Um but it
[6:28] is helping us do more uh do it more efficiently
but also do things we haven't been able to do.
[6:35] If you if you think about the vastness of space
and all the um missions that we could and should
[6:43] be doing to help us better understand our earth
um artificial intelligence is going to help us
[6:49] get there get there more efficiently and but not
just efficiently in different ways. And it's so
[6:57] fascinating, right? We're in this sort of massive
AI arms race. Things are moving so quickly and
[7:03] often times I think the complaint about government
is it doesn't move at the same pace as the private
[7:08] sector. And certainly in the AI race, uh, you
know, we're talking weekly releases of different
[7:13] models that are advancing and progressing so
rapidly. I'm just curious, I mean, how do you,
[7:18] you know, if you're if you're making sort of
strategic decisions around the technologies
[7:23] that you might adopt within a particular
government agency or within an organization,
[7:28] how do you make a decision around a particular
type of technology that you might invest resource
[7:32] in versus say bringing in something from a
third party? You know, like thinking about the
[7:38] decision-m behind sort of backing and providing
the resource for privy for example, like what
[7:43] goes into that? Well, there's a there's a couple
different ways to think about it. Um, you know,
[7:49] space has been very hardware dependent uh for for
decades and because of AI's capabilities, it's
[7:57] now becoming um more servicecentric and I think
that's a real shift in the community. But from
[8:05] an acquisition standpoint, like you asked, you
know, uh, you know, we're not thinking of just a
[8:12] single satellite going to space and doing one lap
around the Earth. It's more constellations and,
[8:18] uh, you know, less about the hardware. Some of the
questions are about how much are you processing in
[8:25] space, how much are you sending down to Earth,
what is the bandwidth that you have? And so
[8:31] um and typically once you send something to space
um it's not updatable unless it is software and so
[8:39] having softwaredefined capabilities is more and
more important and then your question about the
[8:45] government you know moving slowly um all as in in
my engagement with various government government
[8:53] agencies NASA space force and beyond FCC FAA
there's a commercial first strategy that all
[9:01] of them are taking that they're all um creating
strategies so that they can capitalize on how fast
[9:10] AI is moving in the ecosystem and they're not even
trying to keep up. They're just trying to leverage
[9:15] it in a way that keeps our uh national assets
at the cutting edge of technology. Yeah. And
[9:22] I I wonder I mean your discussions with various
different agencies now I mean where does the focus
[9:28] seem to be like what what is the thing that's
coming over the horizon that everyone is sort of
[9:32] either readying for preparing for or or trying to
just sort of like keep on top of in terms of yeah
[9:40] uh well I that's kind of a interesting question.
So for NASA, you know, it's all about getting to
[9:47] the moon, uh, getting there, uh, on the timeline
that they've set out to and doing it in a way
[9:54] that is safe and returns our astronauts back.
Space Force has it slightly different. And so,
[10:01] um, the bottom line is they're just trying
to move faster and they're trying to leverage
[10:06] commercial to get there, no matter what the
mission is or the capability. Yeah. Is that
[10:12] is that what you're asking or do you want me to
go a different way? Sort of. I think the thing I
[10:16] would love to understand a bit more is you know
in your engagement what are they talking about
[10:22] that seems particularly novel and interesting
things that you you you know you've not we've
[10:27] not necessarily seen deployed as yet. They're not
necessarily seen market in any way but things that
[10:32] problems either problems that they want to solve
or technologies that they're keeping an eye on
[10:37] that they think are particularly beneficial in the
future but aren't ready yet. Yeah. So, you know,
[10:43] in some sense it it is coming out like on the
Space Force side, they're they're looking at
[10:49] um moving target indication from space and that's
a new capability and it's uh in the past space has
[10:57] been leveraged for still pictures, right? And
not the moving aspect. Um and that has been a
[11:04] capability that's been more on aircraft. And so
moving that capability to space, using space for
[11:10] things, besides communications, besides pictures,
and um you know, that's really exciting. Um,
[11:19] I'm really excited about the commercial space
stations that are in development and trying to
[11:27] build a new market for where experiments can be
done, new capabilities can be built, whether it's
[11:33] 3D printed or um, you know, um, well 3D printed
is the one that comes to mind, but uh, pharmacy,
[11:44] pharmaceutical manufacturing, things like that.
So there there's a there's so much going on and
[11:50] you know the hot topic is data centers in space
as well and is that a reality and so so much
[11:57] yeah let's delve into that topic okay shall we
[laughter] I feel like um the discussion around
[12:03] or data centers has been a fascinating one to
try and cover as a reporter in large part because
[12:08] it's both trying to understand the technology
behind it the rationale behind it but also and
[12:15] and effectively whether it can even even be sort
of a possible technology in the near future to be
[12:23] deployed. I mean from your perspective I mean
how near term are orbital data centers? Well,
[12:31] I would say it's still a technology that needs to
be developed and um there are a lot of companies
[12:39] getting into it, not just uh not just SpaceX.
And uh there was a talk earlier today on a data
[12:48] center and you know they're planning on 88,000 uh
satellites for their data center. But, you know,
[12:55] there's some real um technological challenges, but
uh I wouldn't I wouldn't count it out. Uh there's
[13:03] a need. We we on Earth have a um uh you know a
space not a space but a uh territory limitation
[13:14] and um it uses a lot of resources and there are
advantages to having uh data centers in space.
[13:23] the unlimited uh energy from the sun. Um if we
can solve, you know, sharing that data down and
[13:31] getting new bandwidth uh mechanisms down to earth
and back and keeping them uh safe from debris. Um,
[13:40] I think it's a very it's it's worth exploring and
we're going to need it because we need more data
[13:46] centers to power the AI agents and the demand
for AI is only getting higher and it's really
[13:52] accelerating whether it's for space or commerce
or any application. Yeah. I mean 88,000 seems
[13:59] poultry. I think SpaceX Yeah. have filed for
a million. they did and you know countries
[14:04] are filing for that and so um it's going to get a
little messy and and of course it's all dependent
[14:11] on launch right and um how Starship brings the uh
satellites up and makes it reusable and affordable
[14:18] and the number of satellites you need to actually
um have an effective and efficient and affordable
[14:27] um data center it's it's going to take a minute
and obviously you know the cooling is the major
[14:34] um technological uh challenge right now, but you
know, I think we'll get there. I really do. Yeah.
[14:41] I mean, do do you want to bet on when or I I
was asked that at lunch, Rafe? And so, um Oh,
[14:51] uh five five maybe seven years. Interesting. I
don't know. maybe you know and I'm a scientist
[14:58] and I I you know I like watching the science
develop and you know there's especially when
[15:04] the economics are driving it um I think that'll
accelerate it yeah I mean I think a forlux pivoted
[15:12] into orital data centers suddenly it sort of made
economic sense sort of their business model had
[15:18] to pivot to make economic sense right do orbital
data centers instead blue origin are looking at
[15:24] orbital data centers 285 far constellation. So
I guess the sort of momentum is there. Um and
[15:33] we certainly saw with I think Starship with flight
13 starting to see some some early deployments of
[15:40] their most advanced satellites even if it went
up for about 20 minutes. Right. [laughter] But
[15:46] I mean I'm curious like when we start to sort
of build out these masses of constellations,
[15:52] what are some of the geopolitical constraints or
concerns that we should be thinking about as well?
[15:57] Because I I feel like that that feeds into some of
some of the things that you would have looked at
[16:02] when you were working on Space Force. Well, yeah.
And and you know, space, let's back up just a
[16:09] little bit before we get to that. you know, space
has been such a benign uh environment for decades,
[16:16] right? Um, you know, the capabilities you deployed
up there, you didn't have to um uh design in
[16:23] redundancy, resiliency, they weren't under threat
either from, you know, the only threat would have
[16:30] been meteorological, right? Uh but now with um
debris with um you know capabilities that could
[16:39] move them that could knock them off orbit. Um
you know all companies uh the governments around
[16:48] the world are thinking about resiliency and um
you know if something happens how do you keep
[16:54] capabilities uh going on and especially when
we are starting to depend more on things like
[17:01] um communications uh on our flights. Um, you know,
most of us probably used space on our cell phones
[17:08] about 20 times before we had our first coffee
this morning. So, it's it's part of our daily
[17:15] daily lives. It's a um underpinning our commerce
and so protecting it. Yes. Uh it's a job of the
[17:23] space force to protect all those assets to keep
continuity of operations, but um it's a design
[17:30] issue as well. Yeah. I mean I don't know if you
can delve into the sort of design issue aspect.
[17:35] So yeah um couple couple different ways. One you'd
want to be able to uh withstand impact cuz a small
[17:45] you know even a little penny in space is traveling
at 17,000 mph something like that you know just
[17:51] the impact of that would be extraordinary. Um
you know some are looking at uh resiliency in
[17:59] terms of that's why the constellations are so huge
right so you can have attrition in satellites and
[18:08] um you know work around a singular satellite
going down and having that operational capability
[18:15] continue or multi-orbit and so um lot lots of
different ways to get after it. It really depends
[18:22] on what you're trying to achieve. Yeah. And I I
think earlier you mentioned obviously, you know,
[18:26] it's not just the US, it's the US in tandem with
many of its allies. You and I have spoken about
[18:31] sovereign AI in the past and about the sort of
wider sovereignty debate. I'm just curious how
[18:38] you know the US and its allies can sort of align
on particularly on this matter particularly when
[18:45] we start to deploy ever larger constellations in a
much fter environment. uh and you know frankly um
[18:54] partnering with other nations is another source of
resiliency right and redundancy and so there are
[19:00] multiple paths and um there are there are lots of
ways that um you know the United States partners
[19:10] with countries to uh create capabilities you
know the Artemis was it the first international
[19:17] um uh crude mission uh that's just on the
human space flight but um sharing data um
[19:26] where the ground stations are uh around the world
so that um you know all the all the downloading is
[19:34] um consistent and continuous. So lots of different
ways there are so many um spaces so complex and so
[19:43] uh vast that and and it's takes a lot of
uh capital to deploy capabilities there.
[19:50] So partnering is really how you achieve success.
I think we're sort of at time which is unfortunate
[19:57] because I have a list of of questions to continue
asking you uh about uh but thank you so much
[20:03] Heather for sitting down and thank you to all
of you for sort of waiting through to the end
[20:07] of the day. Yeah, absolutely. Well, enjoy your uh
happy hour and hope you had a great week at AI4.
