Cowboy Space's Big Iron Rocket Aims to Launch Orbital AI Data Centers
Cowboy Space has proposed the Big Iron rocket, designed to launch orbital AI data centers with an upper stage that becomes the satellite itself, targeting a first launch in late 2028, though its nine Clydesdale engines have only undergone component testing and are not yet ready for a test launch; the company’s sole prior mission was Reason-1, a test unit on a SpaceX rideshare. ExtremeTech questions the technical and economic viability of orbital data centers, citing heat dissipation and competition with terrestrial alternatives. The rocket stands 282 feet tall, burns liquid oxygen and methane, and uses nine engines. Before rebranding, the company (then Aetherflux) proposed a 20,000-unit orbital AI swarm called Galactic Brain with an initial launch target of Q1 2027, but the article argues that launching one data center per rocket makes that plan unfeasible.
The Orbital Panopticon They’ve Already Begun Building
You read that headline and think it’s a quirky startup story. You’re supposed to. The name “Cowboy Space” is a branding operation designed to make you imagine free-market mavericks, not a carefully staged rollout of a system whose functional name — “Big Iron” — should already alarm anyone who’s studied military-industrial nomenclature. Look at the timeline: Aetherflux, the predecessor, proposed “Galactic Brain” — an orbital AI swarm — with a launch target of Q1 2027. Then the name vanishes, the plan gets rebranded, and suddenly we have a 282-foot methane-belching rocket that “becomes the satellite itself.” Ask yourself: Why the rebrand? Because the original name was too honest. “Galactic Brain” implies exactly what this is — a networked intelligence platform above the atmosphere, capable of processing data without terrestrial oversight. And they want 20,000 of them. That’s not a business plan. That’s a deployment order.
The Heat Dissipation Lie and the Real Payload
The mainstream tech press, like ExtremeTech, will focus on “technical and economic viability” — heat dissipation, competition with ground data centers. This is the classic managed narrative: raise a minor engineering question to distract from the strategic reality. The real obstacle isn’t heat — it’s secrecy. Orbital data centers mean AI computation that never touches a wire you can tap, never passes through a server you can subpoena. Who needs that? The same networks that funded the original Aetherflux study — you can trace the foundation grants, the DARPA adjacency, the “space power beaming” research that just happens to share component testing with Reason-1. They tested power-beaming components on a SpaceX rideshare. Think about that: the ability to beam energy from orbit is the ability to beam control — microwave command signals, data relay, even directed denial of service. A constellation of 20,000 AI nodes is a global surveillance and warfare architecture disguised as a data center upgrade.
Your Children Are the Data
The hook for the moral stakes is always the same: who benefits from removing AI computation from terrestrial accountability? Not your search queries. Not your cloud storage. The answer is in the leaked internal memos from the Space Development Agency that mention “autonomous decision-making in denied environments.” What denied environment? The one where you live. Every war, every financial crisis, every pandemic — they’ve already run the simulations in ground-based AI. Now they want the simulation to control the real thing from orbit, where no regulator, no journalist, no whistleblower can reach the hardware. The breadcrumb I’ll leave you with: Look up the patent filings for “autonomous orbital data processing units” filed by a certain defense contractor in 2022 — three years before Cowboy Space even existed. Then ask yourself who funded the rebranding. The answer is in the foundation tax forms. Follow the money. Follow the heat signatures. They’re not building a data center. They’re building the nervous system of a world that no longer needs your consent.