Is a data center in space cheaper than one on the ground? Not at any launch price on offer in 2026, and probably not at Starship-era prices either. Starcloud's 2024 white paper assumes launch at about $30 per kilogram and puts the ten-year power, cooling and launch cost of a 40-megawatt cluster in orbit at about one twentieth of the comparable costs on Earth ($8.2 million against $167 million, or just under 5 cents per kilowatt-hour on the ground). By our reading, it also assumes a spacecraft delivering about 200 kilowatts per metric ton.
In 2026, a Falcon 9 lists at $74 million, outside cost models use about $1,500 per kilogram as today's effective launch price, and SpaceX has not published a Starship price. SpaceX cited about 70 kilowatts per ton for its AI1 compute satellite in June, Starcloud's own planned Starcloud-3 is about 67 (200 kilowatts in about three metric tons), and an April 2026 preprint by Jet Propulsion Laboratory physicist Slava Turyshev works out to roughly 17 to 30. Using Turyshev's power and thermal mass and a Starship-era launch price of $250 to $500 per kilogram, the paper's own method gives orbital power at about 8.5 to 17 cents per kilowatt-hour, roughly two to three and a half times the paper's ground cost.
Starcloud's chief executive, Philip Johnston, told TechCrunch in March 2026: "We're not going to be competitive on energy costs until Starship is flying frequently." Boston Consulting Group (summer 2026) puts orbital data centers at 2.5 to 3 times terrestrial cost today; SemiAnalysis (June 2026) at 3.5 to 4.5 times, with base-case parity around 2040; Wood Mackenzie at least 3 times for a 1-gigawatt build; ABI Research expects convergence around 2035.
What holds up: Starcloud-1's Nvidia H100 has been in orbit since November 2025, Google's Project Suncatcher prototype began running artificial intelligence (AI) models in orbit in October 2026, in-orbit processing of satellite data is an early market, and grid scarcity on the ground is a real argument for orbit. The milestone to watch is Starcloud-2, two 8-kilowatt satellites on rideshares in 2027.
Full Deep Dive, with every input and the arithmetic: https://www.satelliteinsights.com/starcloud-orbital-data-center-economics/?utm_source=podcast&utm_medium=audio&utm_campaign=starcloud-orbital-data-center-economics&utm_content=show-notes
Chapters
0:00 Is a data center in space cheaper than one on the ground?
1:05 What has Starcloud actually flown?
2:14 What does Starcloud's white paper claim?
3:15 What does launch cost today?
4:44 How much spacecraft does a kilowatt in orbit need?
6:22 What do Starcloud's chief executive and outside analysts say?
7:39 What part of Starcloud's case holds up?
8:47 When could orbital compute be cheaper than the ground?
Glenn Canales has spent four decades in satellite communications, including running enterprise broadband services for a major satellite operator, and writes Satellite Insights Weekly. No position in any company discussed.