> ## Content Index > Fetch the complete content index at: https://bytevyte.com/llms.txt > Use this file to discover other available public pages before exploring further. # Cowboy Space Bets a Kent Factory on Orbital Data Centers That Don't Yet Have Customers - URL: https://bytevyte.com/cowboy-space-bets-a-kent-factory-on-orbital-data-centers-that-dont-yet-have-customers/ - Published: 2026-09-14T08:06:20.000Z - Updated: 2026-09-14T08:06:20.000Z - Description: Cowboy Space leased a 291,035-sq-ft Kent plant to build hardware for orbital data centers, betting on 300 jobs and 2027 capacity before revenue exists. - Author: Bytevyte Editorial - Tags: ai-beats **Cowboy Space** has leased a 291,035-square-foot industrial facility in Kent, Washington, converting a former Costco distribution center into a manufacturing plant for the hardware behind its orbital data centers. The deal, recorded this month, is the largest industrial lease in the Puget Sound region so far in 2026 and is expected to add about 300 jobs. The California company will build satellite hardware and rocket components at the CenterPoint-owned site. The lease is the most concrete commitment yet from a startup that carried a different name a year ago. Cowboy Space was founded less than two years ago as Aetherflux, a space-based solar power venture, and rebranded in May 2026 after raising $275 million in Series B funding at a $2 billion valuation. Its founder, Baiju Bhatt, previously co-founded the brokerage Robinhood. The rebrand points to where the capital moved. Aetherflux was built around collecting solar power in orbit, and the same engineering in large deployable arrays and power management now points at compute. The name change and the funding round landed together in May 2026, which indicates investors were buying the compute story rather than the solar one. Kent has been the Seattle area's industrial belt for decades, home to aerospace suppliers and distribution warehouses. Trading a big-box retail tenant for a rocket and satellite manufacturer puts an orbital compute venture in direct competition for square footage that logistics operators have long dominated. ## What the Kent plant is meant to build Cowboy Space presents its platform as vertically integrated: rockets, solar arrays and space-based data centers developed as one system rather than assembled from outside suppliers. The company puts the data center inside the second stage of its own rocket, so the upper stage becomes the orbital payload instead of simply carrying one. The design echoes Explorer 1, the first U.S. satellite, which was built as a rocket's final stage. Each unit is intended to run at roughly 1 megawatt of compute capacity, drawing power from solar arrays in low Earth orbit. Cowboy Space has filed with the Federal Communications Commission for a constellation of 20,000 satellites, a scale that requires a launch cadence no operator has sustained. If every satellite in that filing carried a 1-megawatt payload, the proposal would imply tens of gigawatts of orbital compute capacity, more than most single terrestrial data center campuses. A 1-megawatt orbital unit is small by terrestrial standards, where hyperscale campuses are measured in hundreds of megawatts. Closing that gap takes manufacturing volume, not a single demonstration satellite. That arithmetic explains the Kent lease: the factory is the product strategy, because the business only works if hardware can be produced in large quantities at a falling cost per unit. The competitive field is filling in around it. Starcloud ordered fifty laser terminals from SpaceX by the end of May 2026, a sign that a rival has moved past concept work. Muon Space has introduced a satellite platform built for orbital compute, and a newer entrant named Orbital raised early funding for a constellation of its own. Cowboy Space's answer is ownership: controlling the rocket as well as the payload gives it a cost lever that rivals must buy from someone else. ## Why orbital data centers are arriving before their customers The tension in the thesis is timing. Factory space, launch hardware and spectrum filings are being locked in now, while the first commercial orbital data centers are not expected to be operational before 2027\. The Kent lease obligates Cowboy Space to years of rent, staffing and equipment spending before customer revenue can be verified. Technical questions remain open. Power in orbit is limited by solar array area and battery mass. Waste heat has to be radiated into vacuum, and the physics pushes toward large, heavy radiator panels. Satellite-to-ground throughput depends on optical links that few operators have demonstrated at scale. None of these problems is unsolvable. Each one adds mass, and mass adds launch cost. The integrated design carries a trade-off of its own. Building the data center into the upper stage saves the mass of a separate payload adapter and dispenser, but it also makes the hardware expendable and unreachable. A failed component cannot be serviced, and a generation of chips cannot be swapped once it is obsolete. Terrestrial operators upgrade racks in place; Cowboy Space cannot. Workload fit is another constraint. Orbital capacity suits jobs that tolerate distance from users, such as batch training runs, synthetic data generation and archival processing. Real-time inference for interactive applications is a poor match given round-trip latency through ground stations. The addressable market is narrower than the phrase AI compute demand implies. The strategic choice facing every player in this market is whether to buy components or build them. Starcloud's order for laser terminals is the asset-light route: purchase proven hardware, integrate it, launch sooner. Cowboy Space chose the opposite. Integration promises lower unit costs at volume and control over the launch schedule. It also concentrates every technical risk on a single balance sheet, so a rocket failure or a thermal problem delays the compute business too. Capital structure sharpens the risk. The company raised $275 million against a $2 billion valuation before flying any of its own hardware, and every rocket and satellite it builds adds a physical unit cost that software companies never face. A 291,000-square-foot factory raises fixed costs further, because industrial space must be paid for whether or not the production line runs. | Milestone | Detail | Status | | ------------------------------------ | ----------------------------- | --------------- | | Kent, Wash. facility | 291,035 sq ft, about 300 jobs | Leased | | Series B funding | $275M at $2B valuation | Closed May 2026 | | FCC constellation filing | 20,000 satellites | Filed | | Payload power target | About 1 megawatt per unit | Design target | | First commercial orbital data center | Not expected before 2027 | Unproven | Investors funding the real estate are underwriting industrial leases against a business model that has not yet produced revenue. That risk profile differs from paying a software company's cloud bill. If launch schedules slip or on-orbit reliability disappoints, the rent is still due. The counter-argument has merit. Terrestrial data centers face interconnection queues, land prices, permitting timelines and local opposition that have stretched buildouts by years. Cowboy Space positions orbital capacity as a way around those bottlenecks. If a grid connection in a major market takes four years and a factory lease in Kent can be signed today, the orbital option looks cheaper on paper long before a single satellite flies. A signed lease proves intent and access to capital. It does not prove that a 20,000-satellite constellation can be licensed, launched or paid for. Industrial landlords get paid first and stay insulated from those outcomes, which is why the real estate signal arrives so much earlier than the technical one. Between now and first revenue, several gates have to clear. Cowboy Space must show that its rocket reaches orbit reliably, that a data center payload survives launch and thermal cycling, and that optical links can move data at commercially useful rates. Each gate consumes capital and depends on the one before it. The roughly 300 jobs tied to the facility also put the company in competition with established aerospace employers for machinists, technicians and production engineers. The region's supplier base and skilled workforce make Kent a practical place to build hardware. Retaining those workers through a multi-year development program is a separate problem. ## Why this matters The Kent lease shows how quickly orbital data centers have moved from conference slides into industrial real estate, and how far ahead of their own economics that buildout has run. Companies in this race are now signing multi-year property commitments years before the launch cadence, flight hardware and on-orbit reliability that would justify them exist. For anyone weighing orbital capacity as a future compute source, the signals worth tracking are flight demonstrations and named customers, not additional square footage. ## Related Articles - [SpaceX Orbital Compute: Selling Capacity Before the Satellites Fly](https://bytevyte.com/spacex-orbital-compute-selling-capacity-before-the-satellites-fly/) - [Reflex Aerospace Wins Exclusive Orbital Satellite Platform Contract for Orbital-1](https://bytevyte.com/reflex-aerospace-wins-exclusive-orbital-satellite-platform-contract-for-orbital-1/) - [Starcloud's $250M round doubles its valuation, but orbital data center scale hinges on Starship](https://bytevyte.com/starclouds-250m-round-doubles-its-valuation-but-orbital-data-center-scale-hinges-on-starship/) ✔Human Verified --- *Researched and cross-referenced against primary sources by the Bytevyte editorial team. This article was generated with the assistance of artificial intelligence and reviewed by the Bytevyte editorial team.*