Philip Johnston of Starcloud at RAAIS 2026

2025-10-09 · Source: Air Street Press · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure, Emerging Technologies & Innovation · Depth: Advanced, quick

Summary

Starcloud, co-founded by Philip Johnston, is developing orbital data centers to overcome the energy and compute limitations facing terrestrial AI infrastructure. Terrestrial data centers struggle with grid capacity, cooling, land use, and permitting, while space offers abundant solar power and natural radiative cooling for scalable compute. In November 2025, Starcloud launched Starcloud-1, a 60-kilogram satellite featuring the first NVIDIA H100 GPU in space, delivering 100 times more powerful compute than prior orbital deployments. This mission successfully trained a GPT-style language model using NanoGPT and ran Google's Gemma model in orbit, demonstrating the ability to reduce data processing latency from hours to minutes for applications like synthetic aperture radar. The company's long-term vision includes a 5-gigawatt solar-powered orbital data center. Starcloud-2, planned for October 2026, will be its first commercial mission, integrating multiple NVIDIA H100 and Blackwell GPUs, persistent storage, and Crusoe's cloud platform for direct customer access to orbital AI workloads.

Key takeaway

For AI Architects evaluating long-term compute infrastructure strategies, Starcloud's progress with orbital data centers suggests a viable path to overcome terrestrial energy and cooling limitations. You should monitor developments in space-based compute, particularly for high-latency data processing needs like satellite imagery, as this approach could significantly reduce costs and accelerate insights for your future AI deployments.

Key insights

Orbital data centers can overcome terrestrial AI compute constraints by leveraging space-based power and cooling.

Principles

• Space offers scalable, grid-independent compute.
• In-orbit processing reduces data downlink latency.
• Radiative cooling is efficient in space.

In practice

• Deploy NVIDIA H100/Blackwell GPUs in orbit.
• Train GPT-style models on orbital hardware.
• Process synthetic aperture radar data in space.

Topics

• Orbital Data Centers
• AI Infrastructure
• Space Compute
• NVIDIA H100
• Satellite Data Processing
• Energy Efficiency

Best for: AI Architect, AI Scientist, Entrepreneur

Related on AIssential

• XAI's Radical Plan: Data Centers In Space — XAI plans orbital AI data centers and advanced automation, leveraging SpaceX's capabilities to overcome terrestrial compute limitations.
• World's First Space Data Center — $100B Disaster — Space data centers face immense physics and economic challenges despite theoretical advantages over terrestrial facilities.
• Space-Based Data Centres: The Future of AI Computing in 2025 — Orbital data centers offer sustainable, scalable AI computing by harnessing space's unique environment and unlimited solar energy.
• Delian Asparouhov and Philip Johnston on making the case for orbital data centers — Orbital data centers are emerging as a viable, long-term solution for edge compute and inference, driven by space technology advancements.
• Why Scaling AI is Underestimated ⚡ — Orbital data centers powered by space-based solar energy offer a scalable solution to AI's escalating compute and energy demands.
• SpaceX wants to put data centers in orbit, and Musk says it's no big deal — SpaceX aims to put AI data centers in orbit, but current technology limits them to inference, not complex training.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Air Street Press.