Satlyt's Software Is Already Running AI on Three Satellites

Satellite operators have been paying for bandwidth they don't need for decades. They launch a spacecraft with a sensor, it collects a terabyte of raw imagery, and then they pay between $1 and $20 per gigabyte to beam it all down to Earth, where most of it gets filtered out as noise [The Condia, October 2026]. The procurement question is obvious: what if you could filter it in orbit first?

That's the wedge for Satlyt, a startup founded in 2024 by former Google and SpaceX product manager Rama Afullo. The company doesn't build satellites. It writes software that can be uploaded to the computing hardware already on board a partner's spacecraft, turning it into an edge-processing node [PERPLEXITY SONAR PRO BRIEF]. The initial use case is simple: run AI models in space to analyze imagery or sensor data, and downlink only the useful insights,a change that could save a single satellite operator hundreds of thousands of dollars a year in bandwidth costs alone [HyperAI, Unknown]. The longer-term vision is more ambitious: a distributed network of these software-enabled satellites, forming a virtual "above-cloud" infrastructure for space [PERPLEXITY SONAR PRO BRIEF].

The procurement wedge

The bet rests on a shift in how satellite operators think about their assets. Historically, the spacecraft itself was the product. Satlyt is arguing the real product is the data, and that the spacecraft is just a node in a compute network. Its software is designed to be either pre-installed before launch or uploaded in orbit, a flexibility that lowers the adoption barrier for operators with existing fleets [PERPLEXITY SONAR PRO BRIEF]. The economic pitch is direct. By processing data where it's collected, operators can drastically reduce their most variable and expensive cost: downlink bandwidth. For a constellation of dozens of satellites, the annual savings could run into the millions.

This isn't theoretical. The company has already moved from PowerPoint to orbit. Its software was on a satellite that launched in March 2026 aboard a Momentus spacecraft, where it ran Google's open-source Gemma 3 1B language model [SiliconANGLE, October 2026]. A more recent deployment flew on the TakeMe2Space MOI-1A satellite, which launched on a SpaceX Falcon 9 rocket on October 1, 2026 [India Today, October 2026]. A third mission, involving partners NASA and Stellerian, is testing cloud-computing protocols and image-processing workloads in space [Zamin.uz, October 2026]. For a seed-stage company, having three separate in-space deployments is a tangible traction signal that goes beyond signed LOIs.

The team and the check

Founder Rama Afullo built his career at the intersection of cloud infrastructure and space hardware. He worked in Google's cloud division and was a product manager for SpaceX's Starlink business before founding Satlyt [PERPLEXITY SONAR PRO BRIEF]. His background as a mechanical and space systems engineer, coupled with that commercial product experience, gives him a rare fluency in both the technical constraints of orbital hardware and the enterprise sales motion required to sell software to cautious operators.

That profile convinced a consortium of investors to write an $8 million seed check in October 2026, led by Non Sibi Ventures. The round included participation from TLCOM, Antler, Slauson & Co., and several Africa-focused funds like Launch Africa Ventures and Enza Capital [TechCrunch, October 2026]. The geographic spread of the cap table reflects Afullo's own Kenyan-American background and the global nature of the satellite industry.

Round	Date	Amount	Lead Investor
Pre-seed	2024	Undisclosed	Antler [Antler, October 2026]
Seed	Oct 2026	$8,000,000	Non Sibi Ventures [TechCrunch, October 2026]

The capital is earmarked for expanding the engineering team and scaling customer deployments across additional spacecraft [Via Satellite, October 2026]. The company's total funding is reported to be "well past $10M," suggesting earlier, undisclosed financing beyond the disclosed rounds [Payload Space, October 2026].

The realistic competitive set

Satlyt's most direct competition isn't another startup. It's the inertia of an industry accustomed to its existing workflow and the in-house development teams at large satellite operators. Convincing a risk-averse director of satellite operations to upload third-party software onto a multimillion-dollar asset is a high-stakes sales cycle. The company's answer is to lead with a pure software, no-hardware model and partner with respected entities like NASA to de-risk the technology.

The other competitive pressure comes from the horizon. Larger cloud providers and aerospace primes have taken note of the orbital edge computing trend. Google itself has a related space data center effort, Project Suncatcher, and Satlyt's software recently launched on a SpaceX rocket alongside Google's first prototype for that project [Dealroom News, Unknown]. For now, Satlyt operates as a focused, agnostic software layer. Its challenge will be to build a robust deployment and management platform before its larger partners decide to build,or buy,that capability themselves.

What enterprise buyers are watching

The ideal customer profile here is a satellite operator or data service company with a fleet of imaging or IoT satellites, where data volume is high and the value is in specific insights, not the raw data stream. Think companies in agriculture monitoring, maritime tracking, or border surveillance. For them, Satlyt's software could shift a cost center (bandwidth) into a capability center (real-time, in-orbit analytics).

The next twelve months will be about moving from technical demonstrations to contracted revenue. Key milestones will be securing the first commercial, recurring software license with a satellite operator and expanding the number of active deployments beyond the current three. The company's partnership pipeline, which already includes NASA and Stellerian, provides a credible beachhead, but the ultimate test is a purchase order from a fleet operator whose CFO is motivated by the bandwidth savings.

Sources

1. [TechCrunch, October 2026] Satlyt, founded by a former Google and SpaceX product manager, raises $8M to run AI on satellites | https://techcrunch.com/2026/10/01/satlyt-founded-by-a-former-google-and-spacex-product-manager-raises-8m-to-run-ai-on-satellites/
2. [PERPLEXITY SONAR PRO BRIEF] Satlyt company brief
3. [The Condia, October 2026] Article on satellite downlink costs
4. [HyperAI, Unknown] Article on Satlyt's efficiency gains
5. [SiliconANGLE, October 2026] Article on Satlyt's March 2026 launch with Momentus
6. [India Today, October 2026] Article on TakeMe2Space MOI-1A launch
7. [Zamin.uz, October 2026] Article on NASA and Stellerian partnership
8. [Via Satellite, October 2026] Satlyt Secures $8M in Seed Financing | https://www.satellitetoday.com/finance/2026/10/02/satlyt-secures-8m-in-seed-financing/
9. [Payload Space, October 2026] Article on Satlyt's total funding
10. [Antler, October 2026] Article on Satlyt's pre-seed round
11. [Dealroom News, Unknown] Article on Satlyt launch alongside Google's Project Suncatcher