Sophia Space secures $10M seed to demonstrate novel space computers

As space companies race to deploy cutting-edge chips in orbit, cooling these high-powered processors has become a critical challenge.

“It’s cold in space … but without airflow, the only way to dissipate heat is through conduction,” Nvidia CEO Jensen Huang remarked during the company’s latest earnings call when questioned about space-based data centers.

Sophia Space has secured $10 million from investors including Alpha Funds, KDDI Green Partners Fund, and Unlock Venture Partners. The company intends to validate a passive cooling method for space computers on Earth, then purchase a satellite bus from Apex Space to demonstrate its effectiveness in orbit by late 2027 or early 2028.

Companies such as SpaceX, Google, and Starcloud are exploring conventional satellite designs for their planned space data center networks, which depend on sizable radiators to maintain optimal chip temperatures. However, Sophia Space’s founders — CTO Leon Alkalai, CEO Rob DeMillo, and Chief Growth Officer Brian Monnin — are taking a distinct path.

The company’s technology originates from an unconventional source: a $100 million program at Caltech aimed at building orbital solar power stations that could transmit energy to Earth. The researchers eventually developed a sail-like design that is thin and flexible, unlike traditional boxy satellites.

Although technical and regulatory hurdles make generating power for Earth a challenge, Alkalai — a fellow at Caltech’s Jet Propulsion Laboratory — realized the design could be adapted to power processors in space. (Aetherflux, a space solar startup, has also reached a similar conclusion.)

Sophia, an Nvidia partner, has created modular server racks called TILEs that integrate solar panels. Each TILE measures 1 meter by 1 meter and is just a few centimeters thick. According to DeMillo, this thin design allows processors to be placed against a passive heat spreader, making active cooling unnecessary. He anticipates that 92% of the generated power will be used for processing — a substantial improvement over conventional designs. However, this setup demands a sophisticated software management system to balance workload across the processors.

Save up to $300 or 30% on TechCrunch Founder Summit

More than 1,000 founders and investors will gather at TechCrunch Founder Summit 2026 for a full day dedicated to growth, execution, and real-world scaling. Hear from founders and investors who have influenced the industry. Network with peers facing similar growth challenges. Leave with actionable strategies you can implement right away.

Offer expires March 13.

Save up to $300 or 30% on TechCrunch Founder Summit

More than 1,000 founders and investors will gather at TechCrunch Founder Summit 2026 for a full day dedicated to growth, execution, and real-world scaling. Hear from founders and investors who have influenced the industry. Network with peers facing similar growth challenges. Leave with actionable strategies you can implement right away

Offer expires March 13.

By the 2030s, Sophia aims to construct larger space data centers using thousands of TILEs, envisioning a 50-by-50-meter structure capable of delivering 1 MW of computing power. DeMillo contends that building space data centers with less efficient systems would not be cost-effective and that a single structure, rather than a distributed network connected by lasers, would be simpler to implement.

First, Sophia plans to start by offering its TILEs to satellite operators in need of on-orbit computing. Potential customers include Earth-observation satellites that gather vast amounts of sensor data, missile warning and tracking systems that the Pentagon is pouring billions into, and increasingly sophisticated communications networks.

“The dirty little secret of the satellite industry is that we have all these incredible sensors in orbit producing terabytes or even petabytes of data every few minutes, and they discard most of it because they can’t process it onboard and can't transmit it to the surface and back quickly enough,” DeMillo told TechCrunch.