The Orbital Edge: Satlyt and the Architecture of the Space-Based Cloud

The traditional boundaries of cloud computing are no longer tethered to terrestrial data centers. As the "New Space" economy matures, the focus is shifting from simply getting hardware into orbit to managing the massive amounts of data generated once they are there. At the center of this shift is Rama Afullo, a Kenyan American engineer whose journey through the upper echelons of Silicon Valley’s aerospace and tech giants has led to the creation of Satlyt.

With a freshly secured $8 million seed round and a software suite set to launch on a SpaceX rocket, Satlyt aims to become the "Android of space"—a horizontal software layer that allows satellites to share computing power, process AI models locally, and drastically reduce the costs of operating in the vacuum of orbit.

Main Facts: The Rise of Satlyt and the $8 Million Seed Round

Satlyt, headquartered in Sunnyvale, California, and Nairobi, Kenya, has emerged from stealth with an ambitious mission: to decouple satellite hardware from the software required to run complex, high-compute tasks. The company recently announced an $8 million seed funding round led by Houston-based Non Sibi Ventures. The involvement of Non Sibi is notable not just for the capital, but for the expertise of partner Bernard Harris, a former NASA astronaut with over 20 years of experience, who provided the technical conviction necessary to back a venture focused on orbital edge computing.

The company’s core product is a software platform designed to operate AI models directly on satellite hardware. Unlike competitors such as SpaceX or Google, which are building their own massive orbital data center constellations, Satlyt is positioning itself as a software-first entity. Its platform functions similarly to VMWare or Snowflake, providing an abstraction layer that allows users to run complex workloads without needing to manage the specific underlying satellite infrastructure.

This week marks a pivotal moment for the startup. Satlyt’s software is scheduled to launch on a SpaceX Falcon 9 rocket. This mission will see the software integrated into a spacecraft built by TakeMe2Space, an Indian startup specializing in satellite computing hardware. This launch is particularly symbolic as it occurs alongside the first prototype for Google’s "Project Suncatcher," the search giant’s own foray into orbital data centers.

Chronology: From Corporate Rejection to Orbital Launch

The genesis of Satlyt lies in Rama Afullo’s realization that the major players in the space industry were overlooking the software bottleneck. Afullo’s pedigree includes a tenure in Google’s cloud computing division, followed by a stint at SpaceX’s Starlink in 2024. During his time at both companies, he recognized a recurring pattern: the hardware was evolving rapidly, but the software ecosystems remained closed, proprietary, and inefficient.

• The Internal Pitch (2023-2024): While at Google and SpaceX, Afullo pitched the idea of a horizontal software layer for orbital computing. In both instances, the corporate giants declined, preferring to focus on their proprietary vertical stacks—Google on its cloud infrastructure and SpaceX on its Starlink communication network.
• The Founding of Satlyt (2024): Convinced of the market gap, Afullo left SpaceX to co-found Satlyt. He chose a dual-headquarters strategy, leveraging the engineering talent of Silicon Valley and the burgeoning tech ecosystem in Nairobi, Kenya.
• Initial Proof of Concept (Early 2024): Satlyt successfully flew its software on two demonstration missions. One of the most significant milestones involved a partnership with Google DeepMind to deploy the Gemma AI model on a spacecraft operated by Momentus. This mission proved that large language models (LLMs) and complex AI could function in the constrained environment of a satellite.
• Seed Funding (Mid-2024): The company secured $8 million in seed funding, providing the runway to scale its engineering team and prepare for more complex orbital trials.
• The TakeMe2Space Mission (October 2024): Satlyt’s software launches this week, carrying workloads for NASA, Stellerian, and TakeMe2Space, marking the company’s transition from experimental demos to a revenue-generating service provider.

Supporting Data: The Economics of the Orbital Cloud

The primary driver behind Satlyt’s value proposition is the "downlink bottleneck." Currently, satellites collect vast amounts of data (imagery, sensor readings, signals) and must beam it back to Earth for processing. This process, known as downlink, is fraught with challenges:

1. Cost: Downlinking raw data is prohibitively expensive, often costing thousands of dollars per gigabyte depending on the frequency and ground station availability.
2. Latency: The time it takes for data to reach Earth, be processed, and for a command to be sent back can be minutes or even hours—too slow for real-time applications like collision avoidance or disaster response.
3. Bandwidth: The "pipe" from space to Earth is narrow. As the number of satellites grows, the congestion of radio frequencies becomes a limiting factor.

Satlyt’s data from its Momentus mission provides a compelling case for edge computing. By running Google’s Gemma AI model on the spacecraft to analyze software errors locally, Satlyt was able to compress and filter the data before transmission. The result was a 60% reduction in the size of the transmission.

According to Afullo, this level of efficiency translates to savings of hundreds of thousands of dollars per satellite per year. When scaled across a constellation of hundreds of satellites, the economic implications are transformative. Furthermore, Satlyt aims to reach a 20% market share of all satellites by 2030, a goal supported by the trend of manufacturers increasingly including GPUs (Graphics Processing Units) and NPUs (Neural Processing Units) in their hardware designs.

Official Responses and Investor Perspectives

The investment community’s interest in Satlyt signals a shift in how "Space Tech" is being evaluated. Rather than focusing solely on launch vehicles or satellite buses, investors are looking for the "connective tissue" of the space economy.

Kent Lucas, Partner at Non Sibi Ventures, emphasized that Satlyt’s success is not contingent on the immediate realization of massive orbital data centers. "We don’t need data centers in space for Rama to be wildly successful," Lucas told TechCrunch. "It can just be driven by the number of satellites going up." This "infrastructure-agnostic" approach reduces the risk for investors, as Satlyt benefits from the overall growth of the space industry regardless of which hardware provider dominates.

Bernard Harris, the former astronaut turned venture capitalist, noted that the operational efficiencies provided by Satlyt solve a pain point he saw firsthand during his time at NASA. The ability for a spacecraft to resolve anomalies autonomously, without waiting for a ground controller’s intervention, increases mission longevity and safety.

Rama Afullo, CEO of Satlyt, frames the company’s mission through the lens of open ecosystems. "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," Afullo stated. This philosophy is reflected in the current SpaceX launch, where Satlyt is hosting software for three distinct entities (NASA, Stellerian, and TakeMe2Space) on a single hardware platform.

Implications: A True Compute Cloud in Orbit

The successful deployment of Satlyt’s software has profound implications for the future of space exploration, national security, and commercial telecommunications.

1. The Decentralization of Space Intelligence

By moving the "brain" of the operation to the satellite itself, we are entering an era of autonomous space assets. For companies like Stellerian, which focuses on space surveillance and tracking orbital debris, the ability to process images in real-time means they can identify potential collisions and adjust orbits in seconds rather than hours.

2. NASA and the Future of Interplanetary Cloud Computing

NASA’s involvement in the current mission to test cloud computing protocols suggests that the agency is looking beyond Earth’s orbit. For long-duration missions to the Moon or Mars, where communication delays with Earth can reach 20 minutes or more, having a localized "cloud" of satellites that can share processing power is a mission-critical requirement.

3. The Democratization of Space Access

Satlyt’s model turns a satellite into a "revenue-generating managed service." Smaller countries or companies that cannot afford to launch their own dedicated constellations can instead rent "compute time" on existing satellites running Satlyt’s software. This could lead to a "Space-as-a-Service" economy, mirroring the way AWS and Azure revolutionized the terrestrial internet.

4. The GPU Gold Rush in Orbit

Afullo’s prediction that every satellite will eventually carry a GPU suggests a hardware revolution is underway. As AI models become more efficient and radiation-hardened chips become more affordable, the "dumb" satellite of the past is being replaced by flying supercomputers. Satlyt is positioning itself as the operating system for this new class of hardware.

Conclusion

As Satlyt’s software hitches a ride on a SpaceX rocket this Thursday, it carries more than just code; it carries a vision of a connected, intelligent, and efficient orbital economy. By focusing on the software layer, Rama Afullo has found a way to bridge the gap between the rigid requirements of space hardware and the flexible, fast-paced world of modern AI. If Satlyt succeeds in its goal of capturing 20% of the market by the end of the decade, the "cloud" will no longer be something we look up at—it will be something that lives among the stars.