For years, the aerospace industry viewed satellites as "bent pipes"—simple relays designed to capture data and beam it back to Earth for processing. Rama Afullo, a Kenyan American engineer with deep roots in the cloud computing giants of Silicon Valley, saw a different future. While working at Google Cloud and later at SpaceX’s Starlink division, Afullo saw the massive, untapped potential for localized computing in low Earth orbit (LEO).
"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.
Undeterred by the institutional skepticism, Afullo left the aerospace giants to found Satlyt, a company headquartered in both Sunnyvale, California, and Nairobi. His mission is simple but audacious: to create a universal, hardware-agnostic software layer that allows satellites to process data in space, rather than relying on costly and slow ground-based downlinks. With a fresh $8 million seed round in its coffers, Satlyt is positioning itself to become the "Android of space"—an open, horizontal ecosystem for orbital computing.
The Chronology of an Orbital Revolution
Afullo’s journey from corporate innovator to startup founder reflects the broader shift in the "New Space" economy.
2024: The Pitch and the Pivot
During his time at SpaceX, Afullo witnessed the internal debates regarding the future of the Starlink network. While SpaceX eventually pivoted toward fully embracing the concept of orbital data centers, the initial resistance to software-defined satellite architecture was high. Recognizing that a proprietary, closed-loop approach—like the one SpaceX might favor—would eventually face bottlenecks, Afullo decided to build a platform that could serve the entire industry.
Early 2024: Proving the Concept
Before securing major venture capital, Satlyt needed to prove that its software could survive the harsh, radiation-heavy environment of space. The company successfully executed two demonstration missions. Most notably, they deployed Google DeepMind’s Gemma AI model onboard a Momentus-operated spacecraft. The results were striking: the model successfully reduced the size of transmission data related to onboard software errors by more than 60%.
The Present: Scaling and Deployment
This week marks a major milestone for the company. Satlyt’s software is set to launch on a SpaceX rocket, housed within a spacecraft built by the Indian startup TakeMe2Space. This mission will serve as a high-profile stress test, hosting three distinct customers: NASA, which is testing cloud computing protocols; Stellerian, a space surveillance startup; and TakeMe2Space itself, which intends to demonstrate its satellite’s capacity as a multi-tenant computing hub.
The Architecture of Space-Based Computing
To understand Satlyt’s significance, one must understand the current limitations of space infrastructure. Today, if a satellite detects an anomaly or gathers a massive dataset of imagery, it must transmit that data to a ground station. This process is expensive, latency-heavy, and prone to interruptions.
Afullo’s approach is fundamentally different. Instead of building the "iPhone" (the satellite hardware), Satlyt is building the "Android" (the operating system).
The VMware of the Stratosphere
Afullo draws direct parallels between his company and software titans like VMware or Snowflake. These companies succeeded by abstracting the complexity of hardware away from the end-user. By creating a software layer that manages AI models and compute workloads, Satlyt allows space companies to focus on their mission—be it earth observation, communications, or surveillance—without having to engineer custom, proprietary software for every single satellite they launch.
Overcoming the GPU Shortage
One of the primary challenges in space today is the scarcity of high-powered graphics processing units (GPUs). Launching specialized hardware into orbit is risky and expensive. Satlyt’s software allows for more efficient use of existing, lower-power processors, while preparing the industry for a future where high-performance compute becomes standard. By optimizing AI models to run on limited hardware, Satlyt creates immediate operational efficiencies that translate to significant cost savings—potentially hundreds of thousands of dollars per satellite, per year.
Supporting Data and Strategic Backing
Satlyt’s $8 million seed round is not merely a sign of investor optimism; it is a vote of confidence in a business model that does not depend on the "all-or-nothing" success of massive, dedicated space data centers.
The funding round was led by Houston-based Non Sibi Ventures, a firm with a deep understanding of the aerospace sector. Crucially, the firm counts Bernard Harris, a veteran NASA astronaut with over two decades of service, among its partners.
Why Investors Are Betting on Satlyt
Non Sibi partner Kent Lucas highlighted a key strategic advantage in their investment thesis: scalability. "We don’t need data centers in space for Rama to be wildly successful," Lucas noted. "It can just be driven by the number of satellites going up."
This perspective is vital. The "space data center" vision, while exciting, faces significant economic hurdles and launch bottlenecks. By positioning itself as a universal software layer, Satlyt effectively hedges against these risks. Whether the industry moves toward massive orbital server farms or continues to favor a distributed fleet of smaller satellites, Satlyt’s software remains essential.
Implications for the Future of Aerospace
The implications of Satlyt’s technology extend far beyond mere efficiency. We are witnessing the birth of a genuine "Cloud in Orbit."
The "Managed Service" Paradigm
Afullo’s ultimate goal is to transform satellites into revenue-generating managed services. If a company can rent computing power on a satellite already in orbit, it changes the economics of the entire industry. "The way to think about it is that we turn your satellite into a revenue-generating managed service," Afullo explained.
Next Year’s Frontier: Distributed Computing
The next major project for Satlyt, scheduled for 2025, involves demonstrating a shared computing system that spans two separate satellites. If successful, this will prove that a decentralized "cloud" can exist in orbit, allowing data to be processed and shared between spacecraft without ever needing to touch the ground.
The 20% Goal
Afullo is aiming for a 20% market penetration by the end of the decade. He believes that within a few years, any company launching a satellite without advanced processors—and the software to run them—will be at a severe competitive disadvantage.
"If you’re putting up a satellite without putting up a GPU on it, at the very least, you’re doing yourself a disservice," Afullo remarked.
Conclusion: A New Layer of the Tech Stack
As the space industry matures, the focus is shifting from "how do we get things into orbit" to "what can we do once we are there." Satlyt represents the third wave of the space race: the transition from infrastructure to intelligence.
By decoupling the software from the hardware, Afullo is democratizing access to orbital computing. Whether it is NASA testing new communication protocols or private startups monitoring orbital traffic, the ability to process data at the edge—thousands of miles above the Earth—is no longer a science fiction concept. It is becoming the standard operating procedure for the next generation of space exploration.
For Satlyt, the path forward is clear. By building the middleware for the cosmos, they aren’t just betting on satellites; they are betting on the inevitable digitization of the final frontier. As they prepare to debut their technology at the upcoming TechCrunch Disrupt event, the industry will be watching closely to see if Afullo’s vision of a truly interconnected, cloud-enabled orbital network can become the backbone of modern space operations.
