Welcome to Issue 15, with autonomy on orbit and wildfire model under contract
Last week the capital showed up. This week it started buying hardware and booking customers.
On the Space4AI side, the in-space compute thesis moved from funding to procurement. Sophia Space named a Rotterdam power-electronics firm to design the full power and battery system for its orbital compute demonstrator. The design bet is passively cooled modular slabs that radiate heat straight into vacuum, without active cooling.
On the AI4Space side, spacecraft AI advanced along the whole path from lab to ledger. Shield AI ran its autonomy pilot aboard a satellite in low Earth orbit, and an Indian servicing startup set a 2027 flight to prove out satellite life extension with a paired chaser and target. EarthDaily, meanwhile, put its wildfire-risk model under commercial contract with a large North American insurer.
Both sides are past the pitch deck — one buying parts, the other signing customers.
Specifics below.
Did you know?
We have recently launched the Bits & Orbits Insights
Check out the growing set of independent dashboards, trackers, and maps on the
AI × Space frontier:
Space4AI
Sophia Space selects Spherical as power systems supplier for Athena in-space compute demo
On August 24, Sophia Space named Spherical, a Rotterdam-based aerospace electronics company, to lead the design and delivery of the complete electrical power system and battery architecture for its Athena Project, an in-orbit demonstration of Sophia's TILE computing technology.
Spherical has been selected to lead the design and delivery of the complete Electrical Power System (EPS) and battery architecture for the Athena Project, an in-orbit demonstration of its TILE computing technology, and the selection marks the Dutch company's first major U.S. commercial contract.
The Sophia TILE — Thermal-Integrated LEO Edge — which was covered in Issue 11, is a modular, passively cooled computing and data-storage architecture built for in-orbit processing.
High-performance computing generates significant heat, while the vacuum of space makes the scaling of conventional cooling approaches impractical; TILE addresses that challenge through flat, modular compute slabs that integrate power, processing and passive thermal management, allowing heat to radiate directly into space without relying on active cooling systems.
Spherical develops spacecraft electronics down to the chip level, giving its engineers comparatively greater control over system design, performance and integration.
Spherical will deliver a flight-qualified electrical power system from its modular product line, tailored to Athena's specific mechanical, electrical and interface requirements.
Its engineering team will work with Sophia Space from requirements definition through integration and qualification, with technical work set to begin in the coming days and the mission targeting flight in Q4 2027.
Rob DeMillo, CEO and Co-Founder of Sophia Space, said the company selected Spherical because of the depth of its power systems expertise and its ability to work closely with the team on solutions tailored to the mission.
AI4Space
Shield AI and Sedaro demonstrate on-orbit autonomous AI pilot aboard NOVI satellite
On August 24, Shield AI, Sedaro, and NOVI Space announced the first on-orbit demonstration of Shield AI's Hivemind autonomy pilot, running aboard a NOVI satellite in low Earth orbit.
The flight marks Hivemind's first deployment in the space domain, and both the highest-altitude and longest-duration run of the pilot to date, extending Shield AI's autonomy stack from airborne platforms to an operational satellite. With NOVI's spacecraft acting as leader of a six-satellite constellation alongside five virtual satellites, Hivemind worked with Sedaro's Autonomy Framework for the Edge (SAFE) to perform multi-objective optimization, balancing imaging tasking, spacecraft health, battery state of charge, and pointing across the constellation. Over a 24-hour experiment window, the spacecraft executed 189 Hivemind-generated and SAFE-approved commands. In a second experiment, SAFE and Sedaro's Edge Deployable Simulators (EDS) ran similar optimizations and raised the satellite's Iridium communications contact initiation by 8 percentage points.
The trust mechanism is the validation layer between the pilot and the bus. "The hardest part of autonomous space operations is trust," said Robbie Robertson, CEO and co-founder of Sedaro, describing how SAFE and the Edge-Deployable Simulators provide a predictive onboard model that lets operators validate autonomous decisions before they execute. Todd Wesley, vice president of Hivemind engineering at Shield AI, said the integration moved operators "from manually operating distributed assets to autonomous mission tasking in a seamless workflow." The demonstration also showed that satellites not originally designed for autonomy can orchestrate ISR missions using Hivemind.
The work sits inside a defense-innovation pipeline. Sedaro's SAFE technology and autonomy modes were developed under a Phase II SBIR award from SpaceWERX, the innovation arm of the U.S. Space Force, through an integrated team spanning SpaceWERX, the Air Force Research Laboratory, and Space Systems Command. The partnership traces to late 2025, when Shield AI and Sedaro established Hivemind as Sedaro's preferred on-orbit autonomy software and the Sedaro Platform as the primary environment for developing and validating it. NOVI CEO Dr. Michael Bartholomeusz said the company's satellites are built to advance edge computing and AI on orbit as a platform for on-orbit testing.
OrbitAID unveils 2027 RPO demonstration mission for satellite life extension
On August 25, OrbitAID Aerospace, an Indian in-orbit servicing company, told Payload it plans to fly a satellite life-extension demonstration in low Earth orbit in Q1 2027. The mission will launch two spacecraft — a chaser and a target — to prove out the full servicing sequence in a single flight.
The demo is structured to run each step of the life-extension process end to end. The chaser will use onboard sensors to inspect and characterize the target satellite, approach and dock with it, transfer fuel, and then raise the target to a higher orbit. CEO Sakthi Kumar said the integrated approach is what separates the mission from prior work.
OrbitAID is building the mission on its own hardware and software, including the docking interface and the program that runs autonomous rendezvous and proximity operations. Company officials attributed a design-to-launch-readiness span of 11 months for the two satellites to that in-house technology and control over its supply chain.
OrbitAID expects hardware and lessons from the LEO demo to carry over to GEO, which it views as the larger market for life extension. Kumar said the company plans to move directly into a commercial GEO life-extension mission after the demo, with commercial discussions already underway.
EarthDaily expands AI wildfire-risk model (FLAIM) via commercial deal with North American insurance major
On August 26, EarthDaily announced a commercial agreement with a large North American property and casualty insurance group to deploy its AI-powered wildfire risk model across high-risk markets in the western United States.
The insurer will use EarthDaily's Fire Likelihood AI Model (FLAIM) to support underwriting, portfolio management and growth planning. FLAIM combines regular satellite observations, daily weather data and other geospatial information to produce wildfire probability and fuel-hazard intelligence at scale, assess directional vulnerability showing how fire may move across a landscape, and reflect changes such as vegetation growth, clearing and development instead of relying on historical land-cover assumptions. In one validation, a FLAIM monthly composite indicated very high ignition likelihood across Spokane, Washington for July 2026, ahead of a fire that began in the area on August 1.
The model's technology was built over more than a decade of forest measurement, field validation and wildfire analysis, and was folded into EarthDaily's Earth observation stack following the company's acquisition of SkyForest in early 2025. Phil Green, EarthDaily's Vice President of Forestry, said the probability-based approach lets insurers distinguish between individual properties within broadly defined high-risk areas.
EarthDaily said the intelligence is geographically consistent, regularly updated and structured for direct integration into insurance systems, reducing preprocessing and custom data engineering for operational use. As data from the EarthDaily Constellation becomes available, its high-resolution observations will provide an additional input for refining the wildfire products.
Till next time,
Meta-beat Column of this week
Read also about the AI Pipeline that sits at the core, producing this Newsletter, including its ups and downs of this week: