Space computing’s central engineering problem is not the processor. It is the solar array that powers it — and the radiators required to cool it. The more mass a satellite devotes to generating its own electricity, the less remains for the hardware that actually processes data, and for the heat-rejection panels that keep that hardware from frying itself in the vacuum. Star Catcher Industries and Aethero announced on Monday a Power Purchase Agreement designed to break that constraint, marking the first time a named orbital AI compute company has locked in a dedicated energy supplier before launching its most powerful mission to date.
The deal, announced Monday morning, makes Aethero the 10th major commercial customer to sign a definitive power purchase agreement with Star Catcher, pushing the Jacksonville, Florida company’s contracted backlog past $60 million. Aethero, a San Francisco-based space compute infrastructure provider, plans to deploy its next-generation NxA-ECM compute module aboard its Titan mission — scheduled to launch on a SpaceX rideshare this October.
The power problem that Star Catcher is solving has been well understood for decades: every satellite today must carry enough solar panels and batteries to meet its peak electricity demand on its own. Low Earth orbit makes this particularly punishing. Spacecraft pass through Earth’s shadow for a significant portion of each orbit, and the surface area available for solar panels on a small satellite is inherently limited relative to the power appetite of sensors, propulsion, and — now most pressingly — onboard compute hardware.
Star Catcher’s answer is to outsource peak power delivery the same way a terrestrial company outsources its electricity. The company’s constellation, the Star Catcher Network, will use optical power beaming technology — concentrating sunlight through a Fresnel lens arrangement, conditioning it through a suite of multi-wavelength lasers optimized for the triple-junction photovoltaic cells already installed on client satellites, and transmitting the resulting beam across orbital distances with an autonomous high-precision tracking system. No modification to the receiving satellite is required. The client spacecraft’s existing solar panels absorb the conditioned beam and generate up to 10 times more electricity than they would from ambient sunlight alone.
In November 2025, Star Catcher transmitted 1.1 kilowatts (kW) of optical power to commercial off-the-shelf solar panels along Space Florida’s Launch and Landing Facility at NASA’s Kennedy Space Center, delivering a total of 10 megajoules across the full test campaign — a result that surpassed DARPA’s prior world record of 800 watts, set in May 2025. The company has not yet demonstrated the system in orbit, where the engineering challenge is significantly more demanding: the transmitter and receiver will be moving at orbital velocities in three dimensions, requiring beam-pointing accuracy that atmospheric tests cannot fully replicate. That orbital demonstration is scheduled for later in 2026.
Read more: Star Catcher Industries Closes $65M Series A to Build the First Orbital Power Grid
Aethero has been building toward the Titan mission through a progression of increasingly capable orbital hardware. Its Deimos mission in 2024 established that commercial Nvidia processors could survive and operate in low Earth orbit. Its Phobos satellite, launched in March 2026 on a SpaceX Falcon 9 rideshare, brought that approach to approximately 157 teraflops of on-orbit processing power — sufficient to run workloads for customers including Booz Allen Hamilton and the U.S. Air Force Research Laboratory.
Titan is a different scale of ambition. The mission will deploy the NxA-ECM compute module — the Nvidia-AGX Edge Computing Module — aboard EnduroSat’s FRAME-15 ESPA-class satellite. Each NxA-ECM module is built around Nvidia’s Jetson Blackwell-based Thor silicon, delivering more than 4,000 teraflops per unit. By clustering four such modules through a Kubernetes-based software stack, Titan will reach more than 16,000 teraflops of distributed on-orbit compute — a figure Aethero says is sufficient for real-time inference, autonomous satellite tasking, and distributed data management without ground-station intervention.
The Titan platform draws on 3.4 kW of peak power from the EnduroSat FRAME-15 bus. That is the figure the Star Catcher PPA is designed to augment. Aethero’s reliability architecture addresses the radiation environment through a hybrid approach the company calls radiation hardening by design: radiation-tolerant commercial silicon backed by hardware and software mitigations — including checkpointing, error correction, memory scrubbing, and filesystem resiliency — rather than fully radiation-hardened components, which are far more expensive and constrain performance.
“As we prepare to deploy our Titan mission and scale our vision to our constellation offerings, locking in the critical infrastructure necessary for long-term success is essential,” said Amit Pinnamaneni, CEO and co-founder of Aethero. The PPA will activate progressively as both companies’ orbital networks expand over the coming years.
The clearest way to understand what the Star Catcher model enables is to look at what satellite operators currently spend mass on. A satellite designed around peak power self-sufficiency must carry a solar array large enough to meet maximum demand, plus batteries to bridge the eclipse periods when no sunlight reaches the panels. That mass and surface area are unavailable for anything else.
When Star Catcher’s network delivers on-demand power, an orbital compute operator like Aethero can right-size its on-board solar panels to a baseline load rather than a peak load, reducing both array mass and the satellite’s physical cross-section — which also lowers orbital collision risk. The freed mass does not go directly to compute chips. It goes, first, to radiators.
This is the co-benefit that the orbital data center discussion frequently misses. Heat rejection is the other binding constraint on compute density in space. Unlike a terrestrial server farm, an orbital data center cannot use air cooling or water cooling — heat can only escape into space by radiating from a panel oriented toward the cold void. The more compute hardware a satellite carries, the more radiator surface it needs, and radiators require mass and area that compete directly with solar arrays. Star Catcher’s March 2026 technical analysis found that decoupling power from the solar array allows radiators to be oriented permanently toward cold deep space, since power collection is no longer tied to panel orientation, and allows the mass savings from smaller arrays to be redirected toward larger radiators — which in turn supports higher sustained compute operation.
Star Catcher’s internal analysis estimates that offloading peak power needs to its grid yields a revenue multiplication of 1.9 times to 3.5 times for orbital compute architectures — a figure based on the company’s own modeling and not yet independently audited. “The infrastructure we are building at Star Catcher is foundational to any future we can imagine in space, and this partnership proves that future is arriving now,” said Andrew Rush, co-founder and CEO of Star Catcher. “During my time leading space infrastructure projects at Made In Space and Redwire, the primary wall we kept hitting was power. That wall is now coming down.”
Rush brings a specific institutional background to that claim. Before founding Star Catcher in 2024, he served as Chair of the NASA Advisory Council’s Regulatory and Policy Committee — a role focused on civil space regulation, legislation, and interagency governance — and currently serves on the Council’s Technology, Innovation and Engineering Committee. His co-founder Michael Snyder previously served as chief technology officer at Redwire. Star Catcher’s investors include Shield Capital, a venture firm focused specifically on national security technology, whose involvement signals dual commercial and defense interest in the orbital power grid model.
The Aethero partnership is the most technically specific deal in what has become a crowded category. SpaceX has announced its AI1 orbital data center concept — a satellite with 150 kilowatts (kW) of peak compute and a liquid radiator system as its central engineering challenge — though independent analysis has questioned the economics and thermal design. SoftBank founder Masayoshi Son has argued that chip costs and communication latency make orbital compute thesis irrelevant to the decisive years of the AI race. Google’s Project Suncatcher and Nvidia’s Space-1 Vera Rubin Module have moved the category from concept toward hardware engagement, though no independent cost model yet supports Son’s dismissal as definitively correct or incorrect.
What makes the Star Catcher / Aethero deal notable is that it involves a customer who has already demonstrated two orbital missions and is two months from a third. The Titan mission is not a proposal. It has a launch contract with SpaceX, a confirmed satellite bus from EnduroSat, and a compute architecture whose specifications have been published. The power agreement attached to it is, accordingly, not a speculative commitment — it is infrastructure planning for a mission that was already under construction.
Star Catcher’s commercial momentum extends beyond Aethero. The company has signed power purchase agreements with Loft Orbital, Astro Digital, and Starcloud, among others, and holds a commercial pipeline that it values at more than $3 billion in projected annual recurring revenue — a figure based on its own pipeline assessment and not verified by any independent financial audit. The company employs roughly 40 people and raised $88 million in total capital, with its $65 million Series A in May 2026 co-led by Shield Capital and Cerberus Ventures — the latter represented by retired General Jay Raymond, the first Chief of Space Operations of the U.S. Space Force. “Persistent surveillance, resilient communications, and unhindered maneuverability are all constrained today by power,” Raymond said at the time of the Series A announcement.
The timeline for the Star Catcher / Aethero relationship to generate operational value is measured in years, not months. Both parties acknowledge that the PPA will activate progressively as their orbital networks expand. In the near term, two milestones will determine whether the thesis is sound. Aethero’s Titan mission, slated for October 2026, will be the first test of whether 16,000-teraflop distributed compute can function reliably in orbit using commercial Nvidia Blackwell silicon and a hybrid radiation-hardening architecture. Star Catcher’s own first in-space power beaming demonstration, scheduled for later in 2026, will test whether the beam-pointing accuracy that worked along a ground runway can be maintained at orbital velocities between two moving objects. If both demonstrations succeed, the operational PPA becomes a credible commercial contract. If either fails, both companies return to the engineering drawing board before the revenue model activates.
The larger question is whether the orbital power grid thesis scales beyond early adopters. Star Catcher’s commercial pitch rests on the argument that every spacecraft will eventually need its service — for mission extension, maneuvering, power surges, and the kind of computational workloads that current spacecraft cannot sustain alone. The Aethero partnership is the strongest evidence yet that at least one category of orbital customer — AI compute infrastructure — finds the economic argument compelling enough to commit before the grid is operational.
Star Catcher’s system concentrates solar energy and conditions it through multi-wavelength lasers tuned to the specific absorption spectrum of triple-junction photovoltaic cells — the type already installed on most commercial satellites. Because the transmitted wavelengths are optimized for those cells, a client satellite’s existing solar panels absorb the beam and generate electricity from it, just as they would from ambient sunlight. No receiver hardware, antenna, or retrofit is required. The critical engineering requirement is not on the receiving end — it is in the transmitter’s autonomous tracking system, which must maintain precise beam alignment between two objects moving at orbital velocities.
Orbital data centers face two simultaneous mass constraints: power generation and heat rejection. More compute generates more waste heat, and that heat can only escape into space through radiators — physical panels that must be oriented toward the cold void and sized to match the thermal load. When Star Catcher supplies peak power externally, an operator can reduce its solar array mass and reallocate that budget to larger, better-positioned radiators. Counterintuitively, a smaller solar array can also improve thermal performance by allowing the radiator to face cold space without competing with power generation for orientation. The net effect is not just more available power — it is a higher ceiling on sustainable compute density.
A teraflop is one trillion floating-point operations per second. At 16,000 TFLOPS, Titan’s clustered compute modules can run real-time AI inference on sensor data, autonomously task the satellite to collect specific targets of interest, and process high-resolution imagery without waiting for a ground station to issue commands. For defense customers — Aethero’s current Phobos mission already serves Booz Allen Hamilton and the U.S. Air Force Research Laboratory — that autonomous tasking capability is the primary value. For commercial Earth observation and telecommunications customers, it means actionable data products delivered in near-real-time rather than after a ground downlink cycle.
The field sits between validated physics and unproven economics. The physics of solar-powered compute in orbit are sound: Nvidia’s silicon runs in space (Aethero’s Deimos and Phobos missions confirmed this), and heat rejection via radiators is a solved engineering problem at the scales currently planned. The economics are contested. SoftBank’s Masayoshi Son has argued that AI chip costs dominate over electricity costs, making the orbital power advantage irrelevant in the near term. Independent models from researchers at MIT, Arthur D. Little, and the European Space Policy Institute project the market could reach meaningful scale by the late 2020s — but only if launch costs, thermal design, and communication latency challenges are all resolved simultaneously. The Star Catcher / Aethero deal is evidence that at least one compute company believes the economics close at the mission scale Titan represents. Whether the broader market agrees will depend on whether Titan and Star Catcher’s orbital power demo deliver what their engineering plans project.
ⓒ 2026 TECHTIMES.com All rights reserved. Do not reproduce without permission.

source