A former product manager at Google and SpaceX, Rama Afullo, recognized the burgeoning potential of in-orbit data processing. Despite his attempts to propose this concept internally at both tech giants, the idea was initially rejected, leading him to establish Satlyt. The company, headquartered in Sunnyvale, California, and Nairobi, focuses on creating open-source software for satellite-based AI, recently closing an $8 million seed funding round.
Satlyt's strategic timing appears impeccable. Both SpaceX and Google have since embraced orbital data centers, with Satlyt's software now scheduled to launch on a SpaceX rocket, accompanying Google's Project Suncatcher prototype. Unlike its larger counterparts, Satlyt is not developing its own satellite hardware. Instead, it concentrates on providing a software layer, akin to VMware or Snowflake, that allows AI models to operate seamlessly across various satellite platforms, aiming to build an 'Android' for orbital computing rather than an 'iPhone'.
The company has already demonstrated its software's capabilities through two successful missions. Notably, it deployed Google DeepMind’s Gemma AI model on a Momentus spacecraft, achieving over 60% efficiency in transmitting software error data. This capability translates into substantial annual savings for satellite operators. Future plans include a collaboration with TakeMe2Space, an Indian startup, for a mission involving NASA and Stellerian, to further test cloud computing protocols and image processing in space. Satlyt's ambitious goal is to establish a shared, multi-satellite computing cloud by next year, effectively transforming satellites into revenue-generating managed services. Backed by Non Sibi Ventures, Satlyt aims for its software to be integrated into 20% of all satellites by the decade's end, anticipating a future where GPUs are standard equipment on every spacecraft.
Satlyt's pioneering work underscores a significant shift towards more autonomous and efficient space operations. By focusing on software and open ecosystems, the company is not only democratizing access to in-orbit computing but also paving the way for a future where satellites are intelligent, self-sufficient, and capable of real-time data processing, ultimately contributing to a more interconnected and technologically advanced world.
