Fast on the heels of our conversation with Maxie Reynolds at OODAcon 2025 in the session entitled “Data at the Edge: Subsea and Sustainable Next-Gen Datacenter“, Alphabet recently released the details of X Development’s Project Suncatcher – an audacious vision for scaling artificial intelligence (AI) infrastructure into orbit by coupling solar-powered satellites equipped with Tensor Processing Units (TPUs) and free-space optical networks.

The project envisions future constellations of networked satellites operating as machine learning data centers in space, harvesting the Sun’s abundant energy to power AI computation beyond terrestrial limits.

Why This Matters

AI compute demand is growing exponentially, driving massive increases in energy consumption. Terrestrial data centers—already significant energy consumers—face limits due to cost, sustainability, and physical infrastructure constraints.

Project Suncatcher represents a paradigm shift: relocating computation to space, where solar exposure is continuous and energy capture can be up to eight times higher than on Earth.

• Strategic Energy Sovereignty: By decoupling AI compute from terrestrial grids, Suncatcher aligns with Google’s broader clean-energy initiatives (nuclear, fusion, geothermal).
• Geopolitical Implications: As noted in the recent OODA Loop analysis “Subsea, Space, and AI Sovereign Power: A Future That Does Not Suffer from a Failure of Imagination”, shifting compute to space will redefine infrastructure sovereignty and national competitiveness.
• Moonshot Continuity: Following the ethos of Google X projects, this research builds on interdisciplinary advances across AI efficiency, sustainable energy, and orbital logistics.

Key Points

• Foundational Premise: If AI is a general-purpose technology akin to electricity, its long-term scaling must harness the largest available power source—the Sun.
• System Architecture: Fleets of solar-powered satellites in sun-synchronous low-Earth orbit (LEO), equipped with Trillium TPUs and inter-satellite free-space optical (FSO) links, could function as a distributed compute fabric.
• Inter-Satellite Networking: Dense Wavelength Division Multiplexing (DWDM) and close-proximity formation flight enable 10+ Tbps optical connectivity—comparable to terrestrial TPU clusters.
• Radiation Resilience: TPUs have demonstrated survivability in proton-beam radiation tests equivalent to five-year missions without hard failures.
• Economic Viability: Launch costs projected to drop below $200/kg by mid-2030s make orbital compute economically comparable to terrestrial data centers.
• Sustainability Convergence: Google’s investments in Kairos Power (advanced nuclear), Commonwealth Fusion Systems (fusion energy), and Fervo (geothermal) provide complementary models for decarbonized compute infrastructure.

For the full white paper on Project Suncatcher from Travis Beals, Senior Director, Google’s Paradigms of Intelligence, and his colleagues, see: Exploring a space-based, scalable AI infrastructure system design.

What Next?

• Prototype and Testing Phase: Google plans in-orbit prototype missions to validate formation flight, optical communication, and radiation tolerance.
• Thermal and Fault Tolerance Research: Key milestones include developing passive cooling systems and fault-tolerant compute architectures for orbital clusters.
• AI-Governance Crossroads: Space-based compute raises new policy frontiers—sovereignty over orbital data centers, sustainability of orbital debris management, and militarization risks in dual-use AI infrastructure.
• Strategic Inflection: As Maxie Reynolds framed at OODAcon 2025 in the session entitled “Data at the Edge: Subsea and Sustainable Next-Gen Datacenters, infrastructure relocation—whether undersea or orbital—is an emerging frontier in the AI race.

For Your Consideration

1. For Industry:
   • Begin developing interoperable orbital data standards and hardware provenance frameworks (e.g., HBOM) for space-based compute.
   • Invest in AI-optimized optical networking technologies adaptable to both terrestrial and orbital environments.
2. For Policymakers:
   • Develop early frameworks for Orbital Compute Governance akin to cloud sovereignty laws, ensuring space compute aligns with ethical and security standards.
   • Expand public-private partnerships in sustainable AI infrastructure, leveraging fusion, nuclear, and space-based solar programs.
3. For Strategic Foresight Analysts:
   • Integrate space-based compute into scenario planning for global AI infrastructure dependencies.
   • Monitor launch-cost learning curves and energy price parity thresholds as key strategic indicators for orbital compute feasibility.

Additional OODA Loop Resources

Subsea, Space, and AI Sovereign Power: A Future That Does Not Suffer from a Failure of Imagination: The global race to expand AI compute is pushing data centers underwater, into orbit, and adjacent to new energy sources — reshaping critical infrastructure and geopolitical risk.

Underwater and Orbital: The Next Frontier for Data Centers: This piece surveys subsea and orbital data-center concepts as emerging responses to land, cooling, and resiliency constraints. It frames these unconventional siting strategies as part of a broader shift toward energy- and environment-aware infrastructure for AI-era compute.

The Future of the Energy Sector and AI Data Centers: An analysis of how electricity availability, grid stability, and fuel choices now shape the feasibility and location of AI data centers. It highlights the tightening linkage between compute growth and long-horizon energy investments, including renewables and advanced nuclear.

Constraints to Growth Across Exponential Technologies: Outlines non-silicon bottlenecks—electric power, skilled trades, and large-scale infrastructure mobilization—that limit deployment speed across emerging technologies. For data centers, these constraints translate into extended timelines and higher execution risk.

The Commercial Convergence of AI Compute and Nuclear Technology: Discusses the growing commercial logic of pairing high-density AI compute with nuclear generation. It points to small modular reactors and novel procurement structures as potential pathways to off-grid or firm, low-carbon power.

The Climate Crisis, Texas’ Bitcoin Miners, and European Farmers: Explores how rising electricity demand from digital industries collides with climate impacts and competing sectoral needs. The piece foreshadows policy and market friction over power allocation as AI loads ramp.