Ex-Google & SpaceX Product Manager Fuels Satlyt with $8M for AI on Satellites

Satlyt, a pioneering company co-founded by Kenyan American engineer Rama Afullo, is making significant strides in the emerging field of orbital data centers. Afullo, whose professional journey included stints at Google’s cloud computing division and SpaceX’s Starlink, recognized an immense, untapped potential for advanced computing capabilities in space. Despite initial rejections when pitching his vision internally at both tech giants, his conviction led him to depart from SpaceX in 2024 to establish Satlyt, headquartered in Sunnyvale, California, and Nairobi.

Satlyt's core mission is to develop sophisticated software that enables multiple satellites to collaboratively undertake large computing tasks while in orbit. The company recently announced an $8 million seed round, an exclusive detail shared by Afullo with TechCrunch, signaling strong investor confidence. The timing for Satlyt’s venture appears opportune, as evidenced by SpaceX's subsequent full commitment to orbital data centers and Google’s ongoing Project Suncatcher, an initiative focused on space data centers. Notably, Satlyt differentiates itself by not planning to construct its own spacecraft, a path taken by competitors like SpaceX, Google, Starcloud, and Cowboy Space Company.

Instead, Afullo’s team is dedicated to building the foundational software necessary to operate complex AI models directly on satellites. He likens Satlyt’s strategic approach to that of VMware and Snowflake, companies renowned for providing platforms that allow users to run intricate software without the burden of managing underlying infrastructure. Afullo articulates this vision clearly: "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." This analogy underscores Satlyt's ambition to offer versatile software compatible across a wide array of satellite platforms, fostering an open and collaborative space computing environment.

The efficacy of Satlyt's technology has already been demonstrated through two successful missions. The immediate focus of their software is to enhance operational efficiencies for spacecraft, which traditionally rely heavily on ground-based flight controllers for problem resolution. The process of transmitting data back to Earth, known as downlink, is both costly and often slow. By leveraging onboard AI to make real-time decisions, Satlyt's software promises substantial savings, whether in autonomously resolving anomalies or by processing critical sensor readings directly on the spacecraft, thereby reducing the need for extensive data transfer. An impressive example of this capability was earlier this year when Satlyt deployed Google DeepMind’s Gemma AI model on a Momentus-operated spacecraft. This deployment successfully reduced the transmission size for onboard software errors by over 60%, an efficiency that Afullo estimates could save hundreds of thousands of dollars per satellite annually.

Looking ahead, Satlyt’s software is poised for another significant milestone with an upcoming launch scheduled for Thursday on a SpaceX rocket. This mission is particularly noteworthy as it will also carry the first prototype for Google’s Project Suncatcher. Satlyt’s software will be deployed on a spacecraft developed by TakeMe2Space, an Indian startup specializing in satellite computing hardware. This mission will cater to three distinct customers: NASA, which is funding tests for cloud computing protocols in space; Stellerian, a startup focused on space surveillance and eager to test image processing workloads; and TakeMe2Space itself, demonstrating the satellite's capacity to host third-party software.

While high-powered GPUs are currently a rarity in orbit, the existing opportunities for Satlyt primarily revolve around the types of workloads they are currently addressing. However, the company harbors grander ambitions. Their next major project, anticipated for next year, involves creating a shared computing system – effectively, a cloud – that spans two different satellites. Should this endeavor prove successful, it will lay the groundwork for Satlyt to establish a genuine compute cloud in orbit, operating as a third-party provider for spacecraft manufacturers. As Afullo succinctly puts it, "The way to think about it is that we turn your satellite into a revenue-generating managed service."

The $8 million seed round was spearheaded by Houston-based Non Sibi Ventures, with partner Bernard Harris, a former NASA astronaut with over two decades of experience, playing a crucial role in validating the firm's belief in this innovative space-based business model. A key factor in Non Sibi’s investment decision was Satlyt’s pragmatic approach; its success isn't contingent upon a "highest-octane vision" of massive space data centers, which currently face economic uncertainties and launch bottlenecks. Non Sibi partner Kent Lucas emphasized this point, stating, "We don’t need data centers in space for Rama to be wildly successful… It can just be driven by the number of satellites going up."

Afullo’s long-term vision is to see Satlyt’s software integrated into 20% of all satellites by the end of the decade. He anticipates that by then, nearly every spacecraft builder will be equipping their satellites with GPUs and other advanced processors. He strongly advocates for this technological integration, asserting, "If you’re putting up a satellite without putting up a GPU on it, at the very least, you’re doing yourself a disservice." Satlyt is also gaining recognition by participating in TechCrunch Disrupt's broader Battlefield competition, scheduled for October 13-15 in San Francisco.