Data centers are moving beyond Earth as companies race to solve the energy and cooling crisis plaguing AI infrastructure on the ground. Sophia Space, a Pasadena-based startup founded by Dr. Leon Alkalai, is leading this orbital frontier with plans to begin testing space-based data centers by next year, leveraging technology developed at NASA's Jet Propulsion Laboratory and Caltech to create lightweight, passively cooled computing systems that can operate in the harsh vacuum of lower Earth orbit.

Why Are Companies Moving Data Centers to Space?

Terrestrial data centers face mounting pressure from municipalities and environmental concerns. The City of Los Angeles temporarily banned new data center construction due to energy requirements, and other cities have followed suit, citing water usage, noise pollution, and grid strain. Space offers a fundamentally different solution: no air or water means companies can abandon traditional cooling systems that consume enormous amounts of power and resources.

Sophia Space's innovation centers on a passive cooling system built into its "Tile" design, which eliminates the need for bulky, heavy liquid-cooled radiators that would be impractical to launch and maintain in orbit. The company raised $3.5 million in pre-seed funding and closed an $11.5 million seed round earlier this year led by Alpha Funds, KDDI Green Partners Fund, and Unlock Venture Partners, and now employs about 25 engineers in Pasadena.

"Data centers in space are foundational infrastructure for the emerging space economy, for future national defense, and for enabling responsive disaster management and saving lives on Earth," said Dr. Leon Alkalai, chief technical officer and founder of Sophia Space.

Dr. Leon Alkalai, Chief Technical Officer and Founder, Sophia Space

What Computing Problems Could Orbital Data Centers Solve?

The real-world applications are compelling. Satellites already monitor weather patterns, crop growth, disaster zones, and supply chains, but they're limited by their ability to transmit raw data back to Earth for processing. Moving computation into orbit means satellites can process information locally and send only the most critical insights downward, dramatically reducing transmission bottlenecks and enabling real-time disaster response, agricultural optimization, and maritime tracking.

Google is pursuing this vision through Project Suncatcher, a partnership with Planet Labs to test AI hardware in space. The companies plan to launch two prototype satellites by early 2027 to validate Tensor Processing Unit (TPU) hardware in orbit. James Mason, chief space officer of Planet, explained the strategic advantage: processing raw imagery near the sensor allows rapid extraction and delivery of urgent information without waiting for massive file transfers.

"As optical communications improve and AI workflows become more integrated, shifting to edge compute and orbital data centers is going to become the standard way we operate," said James Mason, chief space officer of Planet.

James Mason, Chief Space Officer, Planet Labs

When Will Orbital Data Centers Become Economically Viable?

The biggest barrier is launch cost. Currently, sending equipment to orbit costs upwards of $20,000 per kilogram, though this represents a dramatic decline from a decade ago. Google estimates that launch costs need to fall to $200 per kilogram for orbital data centers to compete financially with terrestrial facilities on a per-kilowatt-per-year basis. The company projects this price point could be reached by the mid-2030s as SpaceX, Rocket Lab, Blue Origin, and Relativity Space expand launch capacity.

Rob DeMillo, chief executive of Sophia Space, acknowledged the current economics: "Eventually, the cost will be lower than on Earth, but in the early days, you are getting computing to a place and at a scale that has never done it before. They do it because there is not enough compute aboard the spacecraft. You can't do anything real time".

How to Evaluate Orbital Data Center Opportunities

• Launch Capacity Constraints: Companies cannot solve the launch availability problem alone; SpaceX and Rocket Lab currently dominate the market, and delays like Blue Origin's recent New Glenn rocket explosion demonstrate the challenges in expanding capacity.
• Thermal Engineering Innovation: Success depends on novel passive cooling solutions that avoid heavy, power-hungry radiators; Sophia Space's JPL-derived technology represents a critical competitive advantage in this space.
• Timeline Expectations: Small but functional orbital computing clusters are expected within two to three years, with broader deployment unlikely until launch costs drop significantly in the mid-2030s.

The venture capital community is taking notice. Space-tech investing surged to $11.1 billion last year, a 65 percent increase from $6.7 billion in 2024, according to PitchBook data. By comparison, terrestrial data centers have hundreds of billions of dollars committed to their development, suggesting that orbital infrastructure remains in early-stage exploration.

Other companies are also exploring the orbital frontier. Varda Space Industries in El Segundo has already begun using microgravity for commercial pharmaceutical development, while Vast Aerospace in Long Beach is developing commercial space stations for orbital manufacturing. These parallel efforts suggest that space-based computing may be part of a broader shift toward distributed, orbital infrastructure.

The convergence of AI demand, terrestrial infrastructure constraints, and improving space access is creating a genuine opportunity for orbital data centers. Whether they become mainstream depends on solving the launch cost equation and proving that the engineering challenges of operating in space are worth the environmental and operational benefits on Earth.