Bits & Orbits Weekly - Issue 6

CEO and founder Sethu Saveda Suvanam framed the programme as a response to rising demand for sovereign, secure satellite systems across Europe. "By combining Direct-to-Device connectivity with orbital AI inference, we are creating a space-based edge computing network that puts real-time intelligence directly in the hands of operators," he said. Kimmo Kanto, Director of Space, Defence and Connectivity at Business Finland, said the agency's role is to share risk with companies and accelerate investment in new technologies.

The programme follows a recent EUR 150 million contract for two small GEO satellites with SLI, against which ReOrbit is expanding its regional production facilities and engineering base. ReOrbit is also collaborating with Google Cloud on secure data movement and processing in space, under an agreement announced earlier in 2026.

On June 23, Pasadena-based Sophia Space announced a pair of moves toward putting compute hardware in orbit: a $7 million SAFE financing round that brings its total funding to $22 million, and a partnership with satellite-bus maker Apex for a 2027 in-orbit demonstration. The round, expected to close that week, was led by EverGreen — the NVIDIA Alumni Investment Network — alongside SparkLabs Group and other undisclosed investors.

The Apex deal centers on flying Sophia's TILE (Thermal Integrated LEO Edge) compute modules aboard one of Apex's Nova bus platforms in 2027, processing sensor data on orbit rather than downlinking raw feeds to ground stations. During that flight, TILE will interact with onboard sensor data so engineers can test its ability to host AI inference models for image processing and similar tasks. A first demo, scheduled for fall 2026, will validate Sophia's orbital operating system, SOOS, aboard Kepler's in-space network, and will inform the TILE hardware flight.

Sophia Space CEO and co-founder Rob DeMillo framed the work as enabling satellites to "make intelligent decisions autonomously," noting that today's satellites collect terabytes or petabytes of data but discard most of it due to a lack of onboard processing and downlink bandwidth. He argued Sophia's approach distributes heat and compute across each TILE rather than relying on larger radiators or hundreds of distributed satellites, which he said makes it "cheaper, it weighs less, and it's more efficient." On competition from SpaceX's planned LEO orbital data centers, DeMillo said he expects those to serve mainly internal needs, leaving room for Sophia's "usable data centers with colocation and managed hosting."

Apex, founded in 2022 by Ian Cinnamon and Max Benassi, holds a record for the fastest clean-sheet satellite design to reach production and operation in space; Cinnamon said Sophia's mission is "exactly the kind of mission our Nova platform is designed to support." CTO and founder Dr. Leon Alkalai said the partnership turns satellites into "autonomous computing hubs." The new capital will fund product development, engineering, and commercial hiring, strategic partnerships, and platform deployment, with EverGreen Founding Partner Jeff Brown and SparkLabs co-founder Bernard Moon both citing the convergence of AI infrastructure and the space economy. In this regard, DeMilloto pointed out the Golden Dome initiative and proliferating constellations as demand drivers.

On June 25, digital security startup SpaceComputer said it will fly a demo mission of its secure compute architecture, dubbed Space Fabric, on a Spacemanic satellite in 2027.

Space Fabric is an open-source architecture that SpaceComputer compares to Red Hat Enterprise Linux, positioned against current systems that company officials describe as bespoke, closed-source creations. It is a mix of hardware and software that sits on top of a satellite's compute module without affecting the on-board computer, and it is application-agnostic across AI inference models, image-processing applications, and comms networks. The demo aims to validate two features: a physical barrier that walls off different payloads sharing the same satellite so their compute operations cannot be tapped or altered, and cryptographic proofs that confirm processes run in orbit were in fact run in orbit. By generating cryptographic keys on board, the system keeps sensitive information from touching terrestrial systems or other payloads.

CEO and cofounder Daniel Bar framed the pitch as a move away from trust-based assurance. "You have a lot of very old school players that are still, in some regard, operating on a 'Trust me, bro' model," Bar told Payload. "Essentially, you have the cryptographic proofs to verify that the compute workload actually took place in-orbit, and there is no way to tamper with it. It's confidential."

CTO and cofounder Filip Rezabek tied the threat to rideshare economics, where many tenants share one bus. "Rideshare options, where you have maybe 20 to 30 different payloads sitting on the same bus…the attacker can be sitting in one of the other payloads on the same bus," Rezabek told Payload. As operators increasingly rely on partners for compute components, SpaceComputer officials argue that security between subsystems becomes the exposure point, not just the communications link.

The European Space Agency convened an expert workshop on space-based data and AI centres at its European Space Operations Centre (ESOC) in Darmstadt, under the banner "Data/AI Centres in Space — Preparing Choices for Europe." The session, titled "Space-based data centres: opportunities, constraints," was described by ESA's Kai-Uwe Schrogl as work toward a substantiated European position on orbiting data infrastructure.

The framing is policy-first rather than a hardware announcement: ESA is assembling expert input to define where Europe should place itself on in-space compute before committing to a programmatic direction. Hosting the discussion at ESOC, ESA's mission-operations hub, ties the question to the agency's existing ground and operations footprint.

On June 23, Dublin-based Ubotica Technologies announced an $11M funding round to commercialize its Live Maritime Intelligence platform, which runs AI inference onboard satellites to detect maritime threats in near real time. Ireland's Act Venture Capital and Finland's Greencode Ventures co-led the round, with participation from existing investor Atlantic Bridge, a global growth equity firm.

Ubotica's pitch rests on running AI at the orbital edge rather than downlinking raw imagery for ground processing. CEO and cofounder Fintan Buckley said the company spent eight years proving out orbital AI across more than ten missions; the company reports running hundreds of thousands of inferences in orbit, deploying more than 30 Earth observation models on board satellites, and a 100% mission success rate. Working with NASA's Jet Propulsion Laboratory and the European Space Agency, it says it delivered the first spacecraft to autonomously identify a target ahead and reorient itself to capture it.

Live Maritime Intelligence continuously flags areas of elevated maritime risk, then uses autonomous ground-to-orbit tasking to dynamically point available sensors—optical, SAR, or RF—at emerging activity instead of relying on fixed collection schedules. The platform is hardware-agnostic, integrating on any satellite with edge-processing capability rather than locking software to a dedicated chip built in before launch. "Live Maritime Intelligence predicts where risk is emerging, tasks the right satellites and sensors, and delivers decision-grade intelligence in minutes," Buckley said to Oceannews.

Buckley framed the timing around European security concerns, as governments grapple with threats to undersea cables, offshore energy assets, pipelines, and shipping routes, alongside shadow fleets, dark vessels, and sanctions evasion. Ubotica says demand over the prior 12 months has been driven by the platform's ability to surveil exclusive economic zones that are often far larger than a nation's landmass.

On June 25, Firefly Aerospace (Nasdaq: FLY) announced it had acquired Space-ng Inc., a Littleton, Colorado, provider of AI-powered vision navigation and autonomous guidance systems. Financial terms were not disclosed; J.P. Morgan Securities LLC served as Firefly's exclusive financial advisor.

Space-ng's vision navigation software flew on Firefly's Blue Ghost Mission 1, where it determined position and attitude, detected hazardous lunar terrain, and autonomously executed two hazard-avoidance maneuvers to redirect the lander in real time before its March 2025 touchdown in Mare Crisium. "Space-ng's vision navigation technology proved itself in the most critical moments of our descent," said Firefly CEO Jason Kim. The deal also brings high-resolution spacecraft cameras and AI compute hardware that enable optical navigation, rendezvous and proximity operations, and docking without GPS or GNSS.

Space-ng was founded in 2024 as a spinoff of Farm-ng, a company founded in 2020 that builds AI vision, navigation, and robotics for farming. Co-founder and CEO Ethan Rublee is joining Firefly as Chief Engineer of Software, overseeing the company's spacecraft software suite, and the team is being fully integrated under the Firefly brand. The acquisition is Firefly's second software purchase in roughly a year, following its October 2025 acquisition of AI defense software company SciTec.

Firefly plans to deploy Space-ng's technology across its Blue Ghost landers and Elytra orbiters for an existing mission manifest. That includes three more lunar landings under NASA's Commercial Lunar Payload Services program (Blue Ghost Missions 2, 3, and 4), the NASA MoonFall mission in which an Elytra OTV delivers four drones to the lunar south pole, and a space domain awareness mission in LEO under the Defense Innovation Unit's Sinequone Project. "Not only can Space-ng's capabilities help us with landing on the Moon, but they can also help us with all the other missions that need that kind of real-time navigation and hazard avoidance," Kim told Payload.

On June 23, Loft Orbital announced an agreement with NASA's Jet Propulsion Laboratory to host and fly demonstrations of JPL artificial intelligence software on Loft's on-orbit infrastructure for Earth science. The demonstrations fall under the Federated Autonomous MEasurement (FAME) project, funded by NASA's Earth Science Technology Office (ESTO).

Flight demonstrations begin in June 2026, with additional deployments on future AI-enabled satellites scheduled across 2027 and 2028. The JPL software is aimed at reducing data latency by removing humans from the processing loop, delivering near real-time insights on wildfires, flooding, and other natural disasters. Loft's spacecraft fly a high-performance processing architecture that supports edge computing and lightweight AI applications. The FAME spacecraft also carry intersatellite links, letting them communicate with one another or with Earth rapidly.

The software will demonstrate autonomous tip-and-queue tasking without ground intervention, which aims to shorten the time between a satellite observation and actionable data reaching scientists, emergency managers, and first responders. "This collaboration with JPL represents a significant step forward in applying artificial intelligence where it matters most in orbit, processing data in near-real time to support urgent decisions on Earth," said Paul Lasserre, General Manager, AI for Space at Loft.

NASA has deployed AI and autonomy software before, but the FAME collaboration is structured to scale that software onto commercial platforms and pave the way for wider adoption of government AI payloads on commercial satellites.

Till next time,