In the burgeoning landscape of the New Space economy, the vision of a "data center in orbit" has long been the industry’s holy grail. Yet, for Kenyan American engineer Rama Afullo, the future of space computing isn’t about building bigger hardware—it’s about creating the software ecosystem that makes that hardware intelligent.
Afullo, a veteran of both Google’s cloud computing division and SpaceX’s Starlink network, has officially emerged from stealth with his company, Satlyt. Having secured an $8 million seed round led by Houston-based Non Sibi Ventures, Satlyt is positioning itself not as a spacecraft manufacturer, but as the foundational software layer—the "VMware or Snowflake of space"—that will allow disparate satellites to share computing workloads.
The Genesis of an Orbital Vision
The concept for Satlyt was born out of frustration. During his tenures at two of the most significant tech and aerospace companies in the world, Afullo saw the potential for orbital computing but found his internal pitches met with silence.
"When I was at SpaceX, I tried to pitch this internally. They said no. When I was at Google, I tried to pitch this internally. They said no," Afullo told TechCrunch.
His realization was simple: while SpaceX and Google were focused on their proprietary hardware, the industry lacked a horizontal, open-source-style ecosystem that could run AI models across any satellite, regardless of its manufacturer. Afullo left these giants to co-found Satlyt, establishing headquarters in both Sunnyvale, California, and Nairobi, Kenya. His mission is to turn the cold, isolated hardware of current satellites into revenue-generating, managed service providers.
Chronology: From Concept to Orbit
Satlyt’s journey has been rapid, characterized by a "test, fail, iterate" philosophy that defines the modern aerospace startup.
- 2024: Afullo departs SpaceX to co-found Satlyt, focusing on software-defined satellite operations.
- Early 2024: The company successfully deploys Google DeepMind’s Gemma AI model onboard a Momentus spacecraft. The mission acts as a proof-of-concept, demonstrating that AI can manage onboard software errors, reducing transmission data—a critical bottleneck—by over 60%.
- Late 2024: Satlyt executes two additional demonstration missions to refine operational efficiencies, moving away from the traditional, ground-dependent flight control models.
- This Week: Satlyt’s software launches on a SpaceX rocket. The payload includes hardware from Indian startup TakeMe2Space, with three distinct customers: NASA, space surveillance firm Stellerian, and TakeMe2Space itself.
- 2025 (Projected): The company aims to attempt the first "cloud" network spanning two different satellites, effectively creating a distributed computing system in space.
The Technical Architecture: Why Software Matters
Current satellite operations are notoriously inefficient. Typically, a spacecraft acts as a "bent pipe," collecting raw data and beaming it back to Earth for processing. This process, known as downlink, is slow, expensive, and subject to significant latency.
Satlyt changes this paradigm by moving the "brain" to the edge. By running AI models directly on the spacecraft, Satlyt allows satellites to perform real-time image processing, anomaly detection, and decision-making without waiting for instructions from a ground station.
Afullo compares his software stack to the Android ecosystem. "The folks like SpaceX who are doing orbital data centers—if they are the iPhone, we’ll build Android as a horizontally integrated, open ecosystem," he explains. By building a platform that functions across various companies’ satellites, Satlyt avoids the massive capital expenditures associated with spacecraft construction, focusing instead on the high-margin, scalable software business.
Supporting Data and Financial Backing
The $8 million seed round led by Non Sibi Ventures is a vote of confidence in a "software-first" approach to the space economy. Bernard Harris, a partner at the firm and a veteran NASA astronaut with over two decades of experience, was instrumental in the investment.
The financial logic is compelling: Satlyt does not rely on the successful deployment of massive, expensive orbital data centers, which are currently hampered by launch bottlenecks and unproven economics. Instead, Satlyt benefits from the sheer volume of small-sat launches. Whether a satellite is used for Earth observation, climate monitoring, or telecommunications, the inclusion of Satlyt’s software provides immediate operational savings.
As Afullo notes, the efficiency gains are not theoretical. In the Momentus mission, the software saved significant costs by resolving anomalies onboard. "The kind of efficiency we’re talking about can save hundreds of thousands of dollars per satellite each year," Afullo claims. For a fleet operator, that is a transformative impact on the bottom line.
Implications for the Space Economy
The implications of Satlyt’s technology extend far beyond cost savings. By enabling cloud computing in orbit, the company is laying the groundwork for a true "Space Internet."
1. Decentralization of Compute
If Satlyt succeeds in its goal to link two satellites into a single compute network by next year, it will effectively create a distributed cloud. This would allow smaller satellites—which lack the power or physical space for high-end GPUs—to "rent" processing power from larger, more capable neighbors. This democratization of computing power could allow academic and smaller commercial missions to perform tasks previously reserved for massive, government-funded spacecraft.
2. A New Revenue Model for Hardware Builders
Afullo’s pitch to satellite manufacturers is simple: "We turn your satellite into a revenue-generating managed service." By hosting Satlyt’s software, a manufacturer can provide compute-as-a-service to third parties. This creates a secondary revenue stream for companies like TakeMe2Space, turning idle satellite time into a valuable commodity.
3. The GPU Imperative
Afullo is bullish on the future of space hardware. He expects that within this decade, nearly every serious satellite builder will integrate GPUs or advanced AI processors into their designs. "If you’re putting up a satellite without putting up a GPU on it, at the very least, you’re doing yourself a disservice," he asserts. Satlyt is positioning itself to be the operating system for that eventual hardware standard.
Challenges and Future Outlook
Despite the optimism, significant challenges remain. The space environment is notoriously hostile to high-performance computing hardware, which is prone to radiation-induced errors and thermal management issues. Furthermore, the standardization of protocols across different satellite buses—each with unique power constraints and proprietary software—is a monumental engineering hurdle.
However, the timing is favorable. As space becomes increasingly congested and data-heavy, the traditional "ground-to-orbit" communication model is reaching its breaking point. The transition toward autonomous, AI-driven satellites is not just a trend; it is a necessity for the survival of the next generation of orbital infrastructure.
By targeting 20% of all satellites by the end of the decade, Satlyt is aiming for a ubiquitous presence. Whether they can navigate the complex regulatory and technical landscape remains to be seen, but their presence at this week’s TechCrunch Disrupt underscores a shift in the industry’s focus. The "New Space" race is no longer just about who can get the most hardware into orbit; it is about who can make that hardware the smartest.
As the industry gathers in San Francisco this week, Rama Afullo and Satlyt serve as a reminder that the most valuable real estate in the solar system may not be the orbit itself, but the software that runs within it.
