683. In the New Space Race, Who Makes the Rules?

Episode Summary

Executive Summary: The episode examines the push to move AI computing into space, weighing its economics, regulatory gaps, and long history of private space patronage. Through NASA economist Alice McDonald, UN space-law chief Rosanna Hoffman, and Planet CEO Will Marshall, it argues that space is becoming a commercial, strategic, and legal battleground where AI could be both enabled by and reshape Earth observation, orbital infrastructure, and the future of the solar system.

Main Topics: AI data centers in space (Priority: 5/5): The core focus is Google’s Project Suncatcher and the broader idea of orbital data centers: using solar-powered satellites to host AI workloads off Earth, potentially reducing terrestrial power and land demands while creating major new space risks. The economics of space investment (Priority: 5/5): Alice McDonald explains that wealthy patrons have long funded high-cost space projects, and that modern privately financed missions and commercial contracts reflect a continuation of this pattern rather than a new phenomenon. Space as a strategic signaling tool (Priority: 4/5): The conversation links spaceflight to national security and geopolitical signaling, arguing that launches and lunar missions have historically demonstrated technological and economic power to allies, rivals, and the developing world. International space law and regulation (Priority: 5/5): Rosanna Hoffman describes how the UN and member states manage outer space through treaties, guidelines, and negotiations, highlighting the lack of a global space traffic management system and the legal challenges posed by mega-constellations. Private companies and commercialization of exploration (Priority: 4/5): NASA’s shift toward fixed-price, milestone-based commercial partnerships—especially for lunar landings—shows how SpaceX, Blue Origin, and others now shoulder more upfront risk and capital in space development. AI, Earth observation, and planetary intelligence (Priority: 5/5): Will Marshall argues that AI becomes more useful when fused with Earth data from satellites and sensors, enabling a form of 'planetary intelligence' that better understands real-world systems and may improve AI alignment. Risk, debris, and sustainability in orbit (Priority: 5/5): The episode warns that hundreds of thousands of new satellites could intensify collision risks, debris, environmental degradation, and congestion in low Earth orbit, making sustainability a central policy problem.

Key Arguments: Space is not a new domain for billionaire-backed ambition; observatories, telescopes, and early rocketry were also funded by private wealth and could cost modern-equivalent billions. Government space spending has historically served two goals: national security and geopolitical signaling rather than a neat optimization function. NASA’s newer procurement model for lunar landings shifts risk and upfront costs to private firms through fixed-price milestone payments. International space law is increasingly relying on nonbinding guidelines because treaties are too slow and inflexible for fast-changing technical issues like debris and traffic management. There is currently no global space traffic coordination mechanism, making close-approach and collision response ad hoc and often dependent on diplomatic channels. AI data centers in orbit may be technically feasible, but their success depends on economics, demand, and the ability to regulate thousands or millions of satellites sustainably. Earth observation data could become foundational for next-generation AI models by giving them real-world, time-varying understanding of the physical planet. Training AI on the living Earth may help align it with human and ecological interests by increasing its familiarity with the biosphere and human activity. The space economy is already much larger than NASA’s budget and is dominated by telecommunications, suggesting orbital AI will emerge within a substantial commercial ecosystem. The long-run trajectory may be a solar-system-scale shift where most energy is used for computation off Earth, though this is still speculative and depends on civilization’s stability.

Data Points: Space economy size: $500 billion to $650 billion per year - Estimated global space economy compared with NASA’s annual budget NASA budget: About $25 billion per year - Used as a benchmark against the size of the overall space economy Space economy share from telecommunications: 75%–80% - Estimated share of the space economy coming from telecom, especially satellites Outer Space Treaty ratifications: 118 countries - Foundational 1967 treaty prohibiting nuclear weapons in orbit and asserting space belongs to everyone Moon Agreement ratifications: 17 ratifications - Illustrates the decline in willingness to sign binding space treaties Von Kármán line: 100 kilometers - Internationally recognized definition of space mentioned in the discussion of SpaceShipOne Lick Observatory era: 1870s - Example of a privately funded observatory with a wealthy patron legacy story Project Suncatcher expected orbital data center revenues: Zero today; single-digit million-dollar research revenues at best - Current market reality for orbital data centers Projected cost to test orbital data centers: Hundreds of millions to low billions of dollars - Estimated cost range to see whether GPU-based space computing works and is efficient Earth imaging scale (Planet + Landsat): About 5,000 full-land-mass images - Combined historical imagery stock used to train Earth-observation AI systems Planet imagery cadence: Every day at 3-meter resolution - Planet’s recent Earth observation program Landsat cadence: Every month since 1972 - Long-running U.S. Earth-observation archive Projected likelihood of orbital AI data centers in 40 years: 90% - Blaise Agüera y Arcas’s estimate that AI computing in space will come to fruition Time horizon for AI energy use in solar system: By 2100 - Prediction that most energy in the solar system will be used for AI and consumed off Earth Potential Earth-like planets: About 1,000 billion - Used in the Fermi paradox discussion about the rarity of life Earth-like planets per human: About 100 billion per human - Same Fermi paradox framing, emphasizing how vast the universe is

Pivotal Quotes: "Space is extremely dangerous. And you really don't know what the consequences of your actions in space will be." — Rosanna Hoffman: Introduced during the discussion of why regulation and coordination matter so much in orbit "I fully expect that by the year 2100, the great majority of energy used in the solar system will be going toward AI and will be both harvested and consumed off Earth." — Blaise Agüera y Arcas: A long-term forecast about the future energy footprint of AI and orbital computation "The real core incentive is an internal psychological one." — Alice McDonald: Explaining why space workers and pioneers are often motivated by belief and vision rather than immediate economic payoff

Implications: Orbital AI could become a major industry, but it will raise urgent issues around debris, traffic control, treaties, liability, and equitable access to low Earth orbit. The winners will likely be firms that solve both engineering and governance at scale.