# What Happened
Satlyt, a startup co‑founded by Rama Afullo, announced that it has closed an $8 million seed round led by Houston‑based Non Sibi Ventures. The company’s software, designed to run AI models on satellites, is scheduled to launch on a SpaceX rocket on Thursday, sharing the launch with the first prototype of Google’s orbital data‑center effort, Project Suncatcher.
# The Key Details
- Founder background: Rama Afullo is a Kenyan‑American engineer who previously worked in Google’s cloud computing division and had a brief stint at SpaceX’s Starlink communications network in 2024.
- Idea origin: While at both Google and SpaceX, Afullo pitched the concept of orbital data centers internally; both companies declined, prompting him to leave SpaceX and co‑found Satlyt.
- Company footprint: Satlyt is headquartered in Sunnyvale, California, and Nairobi, Kenya.
- Funding: The $8 million seed round was disclosed exclusively to TechCrunch.
- Launch details: Satlyt’s software will fly aboard a spacecraft built by Indian startup TakeMe2Space. The mission includes three customers: NASA (testing cloud‑computing protocols), Stellerian (testing image‑processing workloads for space surveillance), and TakeMe2Space (demonstrating that its satellite can host third‑party software).
- Technical approach: Satlyt does not plan to build its own spacecraft. Instead, it provides software that lets users run AI models on satellites, drawing a comparison to VMware and Snowflake in the cloud‑computing world.
- Early results: Earlier this year, Satlyt ran Google DeepMind’s Gemma AI model on a Momentus‑operated spacecraft, reducing the size of a transmission about onboard software errors by more than 60 %—a saving Afullo estimates could be hundreds of thousands of dollars per satellite each year.
- Future goals: The company plans to demonstrate a shared computing system spanning two different satellites by next year, aiming to create an orbital "cloud" that third‑party satellite builders could use.
- Market vision: Afullo hopes Satlyt’s software will be present on 20 % of satellites by the end of the decade, arguing that omitting a GPU from a satellite is a disservice to the operator.
# Why It Matters
The announcement highlights a growing trend toward moving computing workloads closer to the data source—in this case, to satellites themselves. By enabling AI inference on orbit, Satlyt’s approach could reduce the need to downlink large volumes of raw data, cutting both transmission costs and latency. This is particularly relevant for applications such as Earth‑observation, space surveillance, and in‑orbit servicing, where timely insights are valuable.
Satlyt’s open‑software model contrasts with the vertically integrated, proprietary systems being pursued by SpaceX and Google. If successful, it could lower the barrier for satellite operators to adopt advanced onboard processing without having to develop custom software stacks, much like how Android allowed diverse hardware manufacturers to run a common operating system.
# What It Means for Readers
- Technology users and businesses: More efficient onboard processing could lead to cheaper and faster data products derived from satellite imagery or sensor feeds, benefiting industries that rely on geospatial analytics.
- Developers: Satlyt’s platform promises a familiar environment for running AI models, potentially allowing developers to reuse terrestrial machine‑learning workflows in space with minimal adaptation.
- African tech ecosystem: With a headquarters in Nairobi, Satlyt represents a home‑grown African‑linked startup participating in high‑tech space innovation, which may inspire similar ventures across the continent.
- Investors and industry watchers: The involvement of a former NASA astronaut as a venture partner signals confidence in the viability of space‑based computing services that do not require massive orbital data‑center infrastructure.
# What Happens Next
Satlyt’s immediate next step is the upcoming launch on the SpaceX rocket, where its software will be tested in a real‑world orbital environment alongside NASA, Stellerian, and TakeMe2Space payloads. If the demonstration meets its objectives, the company plans to attempt a multi‑satellite shared‑computing experiment next year, a milestone toward building an orbital cloud service.
# VANTIX Takeaway
Satlyt’s effort to create an open, hardware‑agnostic software layer for satellites could reshape how computing is done in orbit, making advanced AI capabilities more accessible to a broader range of satellite operators and potentially lowering the cost of space‑derived data for users worldwide.
