Building the First Data Centers in Space
Philip Johnston, CEO of StarCloud, discusses building data centers in space to address Earth's energy bottleneck for AI. The company launched its first satellite (StarCloud 1) in November 2025 with an Nvidia H100 GPU, and has raised $170 million to deploy a constellation of 88,000 satellites that could provide 20 gigawatts of compute capacity.
Summary
Philip Johnston founded StarCloud in January 2024 with the core insight that dramatically falling launch costs (driven by SpaceX's Starship program) would make space-based data centers economically viable. After visiting Starbase Texas in early 2023, he realized that reusable Starship vehicles could eventually provide thousands of times more launch capacity than currently exists, making previously impractical ideas suddenly feasible.
Johnston initially explored space-based solar power but pivoted to data centers after calculating that the launch cost breakeven for orbital solar was $50/kg (unrealistic), while space data centers needed only $500/kg (achievable with Starship). The company faced significant skepticism—rejected by approximately 100 VCs during early fundraising attempts and 20+ VCs before securing its first check post-YC. Sentiment shifted dramatically when SaaS valuations declined and the AI compute demand bottleneck became apparent, alongside regulatory pushes against terrestrial data centers.
StarCloud 1 launched in November 2025 with five GPUs including an Nvidia H100, which many claimed was physically impossible to operate in space. The team solved thermal management through immersion cooling in phase-change material and addressed radiation concerns through extensive testing at particle accelerators (Brookhaven National Lab and Knoxville cyclotron). The satellite was deployed for $2 million total, versus a $75-100 million estimate from traditional aerospace contractors.
The company's roadmap progresses from StarCloud 1 (proof of concept) to StarCloud 2 (10 kilowatt commercial product for government/military customers) to StarCloud 3 (200 kilowatt spacecraft targeting hyperscale data centers). StarCloud has filed FCC applications for 88,000 satellites, which could deliver 20 gigawatts of compute capacity—20 times the entire U.S. power grid. The company plans to deploy Bitcoin mining ASICs and AWS Outpost hardware alongside GPU compute.
Key engineering challenges include thermal dissipation in vacuum (no convection cooling) and radiation-induced bit flips in processors. StarCloud partnered with Nvidia on a custom 'Reuben' space chip and with SpaceX on Starlink laser terminals for high-bandwidth connectivity. Johnston emphasizes that the strongest competitive advantage is the engineering team—the company remains extraordinarily selective with hiring, maintaining only 20 engineers despite raising hundreds of millions.
Regulatory backlash against terrestrial AI data centers (New York's ban, water usage concerns) actually benefits StarCloud's economics and geopolitical positioning, as building compute capacity in space avoids terrestrial environmental and political constraints entirely.
Key Insights
- Johnston calculated that space-based solar required a $50/kg launch cost breakeven (unrealistic), but space data centers needed only $500/kg—a number approaching feasibility with Starship, which fundamentally changed the business model viability
- The company booked its first SpaceX launch before determining what hardware to fly, using the launch date as a forcing function to drive engineering decisions rather than building theoretically then seeking launch capacity
- Running an Nvidia H100 in space was claimed by many to be physically impossible due to thermal density and radiation; StarCloud solved this through immersion cooling in phase-change material and radiation testing at particle accelerators, proving the skeptics wrong
- Investor sentiment on space tech and hard tech shifted dramatically and rapidly once SaaS stock valuations fell and AI compute demand created clear bottlenecks, moving from 100+ rejections in 2023 to major venture backing by 2024
- Regulatory bans on terrestrial AI data centers (New York, other states) actually position StarCloud advantageously by making space-based compute the politically viable alternative for national security AI infrastructure
Topics
Transcript
[0:00] First thing every space company should do is book the first available launch they can [laughter] >> before they built the thing that they're going to launch >> before they built the thing. Before they probably even know what they're going to launch. Booking a launch is such a good forcing function for a space company. So we founded the company January 1st 2024. January 2nd we booked the first available SpaceX ride share uh launch and we were like okay something is going to be on that rocket. I'm not sure 100% sure what it's going to be at this point but something is going to be [laughter] on there. >> Sounds like the same advice for all…
Full transcript available for MurmurCast members
Sign Up to AccessMore from Y Combinator
Patrick Collison: "What If You Succeed?"
Patrick Collison discusses Stripe's founding story, the importance of concrete customer problems, and shares data suggesting it's the best time ever to start a business. He argues that despite AI advancement, there will be many winners rather than centralization, and emphasizes the value of deep learning alongside AI tools.
Multi-GPU Kernels, Intelligence per Watt, Heterogeneous Inference, and More | YC Paper Club
This YC Paper Club event focused on chip and kernel specialization in AI systems, featuring discussions on multi-GPU kernel optimization, energy-efficient inference, AI-generated kernel code, heterogeneous inference hardware, and GPU-accelerated game engine simulation. Speakers explored how specialization across hardware, software, and algorithms can dramatically improve efficiency and performance across different AI workloads.
Blake Scholl: The Future Was Supposed to Be Faster
Blake Scholl, founder of Boom Aerospace, discusses building the first independently developed supersonic jet by applying software development principles to hardware manufacturing. He argues that startups can compete in regulated industries by building solutions that solve regulatory concerns, and emphasizes that passion, iteration, and vertical integration are key to breakthrough innovation in deep tech.
Boris Cherny: Building Claude Code
Boris Cherny, creator of Claude Code, discusses how Anthropic's Opus 5 model represents a major leap in AI capabilities, requiring a fundamental rethinking of how to build AI products. Rather than accumulating complexity, successful builders must regularly delete system prompts and code to "unhobble" the model, allowing it to tackle increasingly complex tasks like rewriting entire codebases and maintaining software automatically.
Self-Maintaining APIs
The speaker argues that API communication is fundamentally broken across the industry, with breaking changes and new features going unnoticed. Since agentic coding tools have proven viable and developers accept external tool access to codebases, API providers should automatically apply changes to customer code rather than just announcing them—similar to how Dependabot works for dependencies.