The Vision for Orbital Compute
Rama Afullo, an engineer with a pedigree spanning Google Cloud and SpaceX, is betting that the future of computing isn't just on the ground, but in low Earth orbit. After facing internal pushback while attempting to advocate for space-based processing at his previous employers, Afullo co-founded Satlyt. The startup aims to address a critical bottleneck in space exploration: the inefficiency of constant, expensive data transmission to ground stations. By placing AI intelligence directly on board satellites, Satlyt intends to transform these orbital assets from passive sensors into active, data-processing edge computing nodes.
Satlyt has officially announced an $8 million seed round led by Houston-based Non Sibi Ventures. This funding arrives as the startup prepares to deploy its proprietary software stack on an upcoming SpaceX launch. Rather than building hardware or proprietary spacecraft, Satlyt occupies a niche similar to software-defined infrastructure providers like VMware or Snowflake, aiming to create a horizontally integrated, open ecosystem that can operate across a variety of hardware platforms.
The Software-Defined Strategy
The core philosophy of Satlyt is to treat satellite infrastructure as a managed service. Currently, most spacecraft rely on ground controllers to diagnose and rectify even minor operational issues. This model is hindered by latency and the high costs associated with downlinking massive amounts of telemetry data. By deploying AI models, such as Google DeepMind’s Gemma, directly onto spacecraft hardware, Satlyt allows for autonomous decision-making and real-time processing of sensor information. In early demonstrations, this capability resulted in a 60% reduction in transmission sizes related to error diagnostics, translating to significant cost savings per orbital unit.
The company’s upcoming mission, conducted alongside Indian hardware partner TakeMe2Space, serves as a high-stakes proof of concept. This deployment will support three distinct workstreams: testing cloud computing protocols for NASA, executing image processing tasks for the space surveillance firm Stellerian, and demonstrating the platform's ability to host third-party software for TakeMe2Space. This mission marks a pivotal step toward the company's ultimate goal: creating a shared, distributed computing cloud that allows multiple satellites to pool their resources for complex processing tasks.
Why it Matters
- Efficiency: Moving AI inference to the edge (in this case, in orbit) reduces the need for heavy data downlinks, which saves bandwidth and lowers operational costs.
- Standardization: By providing a hardware-agnostic software layer, Satlyt hopes to enable a more flexible "app store" model for satellites, making it easier for space companies to deploy software updates remotely.
- Revenue Potential: Afullo’s model transforms satellites from one-time investments into revenue-generating compute resources, potentially unlocking a new "space-as-a-service" market for data processing.
Outlook and Future Scaling
The path forward for Satlyt is tied to the industry’s increasing adoption of high-performance hardware. While current satellites are often limited by their onboard processing power, the industry is rapidly trending toward integrating GPUs and advanced AI accelerators into standard bus architectures. Satlyt is positioning itself to be the operating system for this new era. As Afullo points out, the inclusion of a GPU is becoming a standard expectation for modern spacecraft; Satlyt intends to ensure that the software driving these processors remains efficient, scalable, and cross-compatible.
The backing of Non Sibi Ventures, which includes the expertise of former NASA astronaut Bernard Harris, signals that the aerospace community sees merit in this pragmatic, software-first approach. Because Satlyt’s business model does not require the creation of massive, high-risk orbital data centers, it remains resilient to the economic and logistical challenges currently facing larger-scale space infrastructure projects. Instead, it rides the wave of the growing number of satellites entering orbit, turning every new launch into a potential opportunity for growth.
