Rama Afullo reacted to a successful 13-month project with the simple phrase, "Let's GOOOO," as his startup Satlyt moved closer to converting operating satellites into distributed computing networks akin to orbital data centers. Earlier this month, a test conducted alongside the University of Houston on a NASA-funded research initiative demonstrated how the firm's software splits tasks among computers in orbit depending on available compute power and communication.
Afullo explained that the objective involves building software linking multiple spacecraft to distribute computing workloads. Users "will only need to define permissions, priorities, and operating limits," while the program independently makes choices within those boundaries, he noted. The technology currently runs on three orbiting satellites, and this trial represented the initial phase of that broader vision. Although the trial used NASA-funded research, the space agency has not yet picked the company for a specific mission or technology deployment.
According to Afullo, NASA missions experience the exact limitations Satlyt aims to solve, which include "limited computing resources, intermittent communication, and the need to get useful information back to people."
During the evaluation, the firm established a simulated network containing four users, four computers, and eight services. The software then calculated how to distribute and execute tasks across those machines based on communication and computing capacity. Only one physical spacecraft computer ran the program in orbit, while the remaining three acted as simulations before transmitting results back to Earth. The trial confirmed that Satlyt's software creates execution plans on hardware in space, though future tests will examine how multiple physical machines exchange data in practice.
"The finding is a trade-off," Afullo stated, noting that one specific method performed best despite demanding higher computing power than alternate options.
Satlyt, which ranked in the top five of the TechCrunch Disrupt Battle this year, views the outcome as a victory. "It's a building block toward satellites sharing computing work," he said.
NASA currently operates a network intended for moving space data despite communication delays, having launched its use earlier this year and tested it on aircraft.
Afullo established Satlyt in 2024 after prior experience at SpaceX and Google to convert spacecraft into virtual data centers. The software enables operators to run artificial intelligence and other programs onboard, processing information directly where it is gathered to lower reliance on ground analysis and satellite downlinks.
For instance, if a satellite detects a wildfire, the system assists in transmitting critical data much faster. "The goal is to get useful answers to people sooner and make better use of limited communication windows," Afullo remarked. "This experiment tested one planning component of that broader capability."
The company acknowledged that further testing is required across various connectivity, hardware, and workload environments. Developers also discovered a time-referencing error regarding delays, noting the system recorded arrival time rather than elapsed delay.
However, Afullo clarified that the experiment did not fail. "The software development, onboard execution, and return of outputs succeeded," he said.
Significant challenges remain, including securing usable communication links, obtaining operator permissions, managing resources, ensuring secure transfers, and handling network recoveries. The company also intends to expand into environments beyond low Earth orbit soon.
"Our next technical steps are broader validation and integration, followed by demonstrations of computing work shared across separate spacecraft," Afullo concluded. "Longer term, we want developers to build applications for space without having to rebuild the infrastructure for every mission. That's what virtual AI data centers in space mean to us."
