SpaceX's Redmond factory makes 70 satellites a week — and no one else is even close

A benchmark that resets the industry

SpaceX's Redmond, Washington factory builds roughly 70 Starlink satellites per week. The company disclosed that figure in its S-1 filing and confirmed it in a behind-the-scenes video released in August 2025. From a single facility, that works out to about 3,640 spacecraft per year, a rate no peer in the satellite industry has matched.

As recently as 2020, SpaceX was building 120 Starlink satellites per month. The Redmond operation has scaled annual output by roughly 200% in five years. Cornelia Rosu, SpaceX's Senior Director for Starlink Production, said in the video that the company "learned how to sustain a 70-satellite-per-week production rate" through rapid iteration on the factory floor. Akash Badshah, Senior Director of Satellite Engineering, put it more bluntly: "All of those Starlink satellites came from here, right in Redmond."

Amazon's Project Leo, the nearest rival in the Pacific Northwest, is producing "tens of satellites a week" at its Kirkland factory, according to Amazon VP Rajeev Badyal, who spoke at a Tech Alliance event. Leo had a little more than 300 satellites in orbit at that time, versus SpaceX's approximately 9,600. Traditional manufacturers named as competitors in SpaceX's S-1 (Eutelsat OneWeb, Telesat Lightspeed, and AST SpaceMobile) have not disclosed weekly rates that approach Redmond's.

Why does 70 per week matter? Because that throughput makes a 30,000-satellite constellation physically possible on a reasonable timeline. SpaceX currently operates around 12,000 Starlink satellites under existing FCC authorization but has applied to expand to nearly 30,000. At Redmond's demonstrated rate, the company could manufacture that entire expanded constellation in a little over eight years, not counting attrition replacements. No other single facility matches the Redmond site's combination of volume and complexity for complete spacecraft assembly. (SpaceX's own Bastrop, Texas plant produces 15,000 Starlink antenna dishes per day, but that is a different class of hardware.)

The S-1 filing also revealed the financial scale behind the production numbers. Starlink generated $11.4 billion in revenue and $4.4 billion in operating income in 2025, accounting for 61% of SpaceX's $18.7 billion total. It was the only one of the company's three business segments to turn an operating profit. The filing did not break out how many of SpaceX's 22,000 worldwide employees work at the Redmond facility.

What Raptor 3 testing means for Redmond

The Raptor 3 engine that roared through a 152-second burn at SpaceX's McGregor test stand in August 2025 is doing more than power Starship. It is reshaping the skills SpaceX needs at its Redmond satellite factory. The feedback loop between Texas propulsion development and Washington-state manufacturing runs tighter than most of the industry realizes.

McGregor's Raptor South stand has become the world's most active rocket-engine development center. Engineers logged 55 tests since the last major facility update, including multiple relight tests on the Raptor Vertical Stand. The most advanced engine iteration observed carries serial number R3.35, destined for Starship Flight 12 aboard Booster 18 and Ship 39. That engine feeds directly into a launch cadence that demands more Starlink satellites (and faster) than any previous generation required.

On the production side, SpaceX's Boca Chica plant has introduced a re-engineered Raptor variant that reduces part count by nearly 30% through laser powder bed fusion and design consolidation. The integrated turbopump housing, once a multi-piece assembly, is now printed as a single geometry. The injector plate is monolithic. The nozzle throat uses a single-piece nickel-chromium alloy with a graded cellular cooling lattice printed in situ. These changes cut brazed joints by 60% and machining hours by 45%, according to reporting on the simplification effort.

That matters for Redmond because the same additive manufacturing and materials-science expertise transfers directly to satellite propulsion components and structural assemblies. When McGregor validates a new print recipe or a revised cooling-channel geometry, the production implications ripple through every facility building Starlink hardware.

McGregor's testing portfolio extends beyond Raptor. Engineers routinely run "torture tests" on redesigned header tanks for Block 2 Starships and Block 3 boosters, intentionally pushing composite overwrapped pressure vessels to failure. The facility also includes a massive new test rig, visible in aerial photography, that may support Starship Human Landing System testing for NASA's Artemis program. That diversification means McGregor is training engineers across propulsion, structures, and lunar-landing systems simultaneously.

The talent pipeline runs both ways. A propulsion fluids analyst working Raptor systems in Hawthorne develops fluid-handling expertise that applies to satellite propulsion modules. A transport technician at McGregor gains hands-on experience with flight hardware that translates directly to satellite integration work in Redmond. SpaceX's vertical integration model (design, test, and produce in-house) means the engineer watching a Raptor 3 relight test in Texas is often the same one specifying manufacturing tolerances for a Starlink bus in Washington.

The 300-Raptor-engine annual production target SpaceX has signaled requires a workforce that understands both the test stand and the factory floor. Every burn at McGregor generates data that feeds back into the print parameters, inspection protocols, and acceptance criteria governing components that end up in satellites rolling off the Redmond line at 70 per week. The Raptor 3 program isn't a separate rocket story. It is the propulsion and manufacturing R&D engine that makes Redmond's satellite throughput possible.

What the job board reveals

SpaceX's Redmond careers page lists over 100 open roles tied to Starlink. That count alone would make the facility a significant satellite employer, but the mix of those roles tells the real story: this is not a procurement-and-integration shop buying subsystems from contractors. It is a vertically integrated factory designing, building, testing, and operating its own satellites, and it needs people who can do all of that under one roof.

The single largest hiring category is software and embedded systems. Dozens of postings target flight software, GNC (guidance, navigation, and control), beam planning, laser mesh routing, and hardware-in-the-loop test automation. The GNC sub-team alone is recruiting for orbit control, ADCS (attitude determination and control), power systems, device navigation, and beam pointing, each a separate role. That spread signals a team simultaneously maintaining the existing Gen2 constellation while building out the next generation of laser-linked, software-defined satellites. When a company hires distinct engineers for orbit determination and for beam pointing, it is not patching gaps; it is scaling parallel development tracks.

The second cluster is manufacturing and production. SpaceX is hiring manufacturing engineers for PCBA, optical fiber, solar arrays, and supplier development across harnessing, batteries, and PCBs. It is also recruiting entry-level and second-shift production technicians, integration specialists, and test technicians in volume. The presence of dedicated "Manufacturing Engineer, Optical Fiber" and "Manufacturing Specialist, Solar" roles, each posted separately, points to production lines specialized enough to need their own engineering attention, not generalists covering multiple processes.

A third, smaller but telling group is silicon and RFIC design. The Redmond board lists RFIC design engineers, MMIC designers, silicon photonics engineers, and high-speed electro-optical IC engineers. These are custom-chip roles. SpaceX is designing its own RF integrated circuits in-house, a move almost unheard of for a constellation operator and one that places the company in direct competition with defense-focused chip firms for a shallow talent pool.

The supply chain and sourcing layer is equally dense. Global supply managers are posted for custom silicon, harnesses, ground infrastructure, and space lasers. Sourcing managers cover connectors, electronics, mechanical commodities, and construction. This is the procurement apparatus of a factory that buys millions of components per month and cannot afford a line-down shortage on any one of them.

Taken together, the job board reads as a factory hiring to produce at least 70 satellites per week while simultaneously building the manufacturing base for whatever comes after, including larger satellites, laser crosslinks at scale, custom silicon, and a ground network that must keep pace. For aerospace engineers in the Pacific Northwest, the implication is concrete: the Redmond corridor is now the place where satellite production is treated as a manufacturing problem, not a science project.

How laser links create an ecosystem

Every Starlink satellite that rolls off the Redmond line carries more than a phased-array antenna. It carries a laser terminal, a device that lets satellites talk to each other in orbit, beaming data between spacecraft without ever touching a ground station. That single component is quietly turning the Redmond factory into a talent pipeline for an entire ecosystem that hasn't fully materialized yet.

Inter-satellite laser links are the connective tissue of the Starlink constellation. Instead of routing user traffic down to an earth station and back up, a laser-equipped satellite can hand data off to a neighbor, then another, then another, keeping the signal in space until it reaches a ground gateway near the destination. The result is lower latency over oceans and polar routes, fewer dead spots, and a network topology that doesn't depend on building ground infrastructure in every country it serves. For a constellation operating at Redmond's scale, those links aren't optional. They're the architecture.

That architecture creates a demand signal extending well beyond SpaceX's own hiring. The optical and laser-manufacturing skills required to build, test, and integrate those terminals are scarce. They sit at the intersection of precision optics, space-qualified hardware, and high-volume manufacturing, a combination that traditional aerospace programs (built around dozens of satellites per year rather than dozens per week) never had to develop at scale. Redmond is where that combination is being forged right now, and the engineers and technicians who learn it there become valuable to anyone else trying to build a networked constellation.

And plenty of companies are trying. Amazon's Project Kuiper has committed to a 3,236-satellite constellation and is ramping its own manufacturing. OneWeb, Telesat, and the European Union's IRIS² program each represent multi-satellite deployments that will need laser-interlink hardware to compete on latency. None of those programs has disclosed production rates approaching Redmond's output, which means none of them has had to solve the manufacturing problems SpaceX's Redmond teams are solving in real time. When their timelines accelerate, and the spectrum and orbit filings suggest they must, the people who already know how to build laser terminals at volume will be the first ones they look for.

The job postings tell the story. Zero G Talent's board currently lists a Mechanical Engineer, Optical & Space Lasers Manufacturing role at SpaceX's Redmond facility, a position laser-focused on the production side of exactly those terminals. That single listing is a signal: the factory isn't just assembling satellites, it's building institutional knowledge around optical interconnects that the rest of the industry will need to draw on.

Every laser terminal that ships from Redmond makes the Starlink network more capable. But it also makes the people who built it more capable, and more recruitable, for the next constellation that needs the same thing. The Redmond factory is producing satellites at a rate no other manufacturer has matched. Less visibly, it's producing the workforce that will build whatever comes after them.

The S-1 that cracked the talent market open

Before SpaceX filed its S-1, the Redmond factory's output was a black box. Analysts estimated Starlink production rates from launch cadence and orbital telemetry, educated guesses rather than hard numbers. The filing cracked that window open. For the first time, a publicly visible document tied SpaceX's name to a specific weekly production figure for a specific facility, giving competitors, suppliers, and the engineers they're trying to recruit an actual benchmark to react to.

That number landed in a talent market that had been operating on opacity. Traditional satellite manufacturers (Lockheed Martin, Northrop Grumman, Airbus Defence and Space) build bespoke spacecraft on multi-year timelines. A production rate of even five or ten satellites per quarter counts as high volume in that world. When SpaceX's filing made clear that Redmond was pushing out hardware at a rate an order of magnitude beyond that, it reframed what "aerospace manufacturing" means as a career. Engineers reading that filing didn't just see a production stat; they saw a signal about where the work is heading.

The disclosure also forced a recruiting recalibration. Competing employers in the Pacific Northwest, including Blue Origin in Kent, Amazon's Project Kuiper, and the various defense contractors with offices within commuting distance of Redmond, suddenly had to answer a question they'd never faced: "Why shouldn't I go work where they're building seventy a week?" That question doesn't have a comfortable answer when your own program is still in low-rate initial production or, in Kuiper's case, hasn't yet launched operational satellites at all.

For hiring managers at those companies, the S-1 data point became a retention risk as much as a competitive-intelligence asset. Engineers with Starlink-relevant skills, including optical interlink testing, flat-panel phased-array integration, and high-rate RF assembly, now had a transparent, publicly citable reason to leave a slower program for SpaceX's Redmond line. The talent market didn't just get more competitive; it got more legible, and legibility favors the employer whose numbers look like a production line, not a science project.

The downstream effect is visible on Zero G Talent's own board. SpaceX added 115 roles in the past week alone, including positions like mechanical engineer optical space lasers manufacturing at the Redmond facility. That hiring velocity is the S-1's logic playing out in real time: once you tell the market your production rate, you also tell it you need people to sustain it, and every engineer watching knows exactly what they'd be walking into.

Starfall and the orbital-servicing workforce

On June 23, SpaceX launched something almost nobody saw coming. A Falcon 9 lifted off from Cape Canaveral Space Force Station at 6:52 a.m. EDT carrying Starfall, a 10.2-foot disk-shaped reentry capsule designed to return manufactured goods from orbit. SpaceX had revealed almost nothing about the vehicle before launch. The demo mission's goal was controlled atmospheric reentry and recovery, a capability aimed squarely at the emerging in-space manufacturing market. No commercial reentry vehicle has successfully cracked that niche at scale.

The connection to Redmond isn't direct (Starfall launched from Florida, not Washington), but the engineering skills the mission demands overlap heavily with what the Redmond Starlink factory is building. Controlled reentry requires thermal protection systems, autonomous guidance and navigation, and optical communication links capable of handing off tracking data during plasma blackout. Those are the same competency clusters listed in SpaceX's open Redmond roles, including a position in that same optical and laser manufacturing discipline.

Starfall signals something larger about where SpaceX is headed. The company already operates the largest commercial satellite constellation in orbit. A return capsule closes the loop: hardware goes up, product comes down. That requires a workforce fluent in orbital servicing, debris mitigation, and end-of-life satellite management, not just manufacturing. SpaceX's own approach to debris mitigation spans the full satellite life cycle: pre-launch design for impact resistance, remote tracking and monitoring, onboard detection and avoidance, and de-orbiting procedures at retirement. Each of those layers needs engineers who understand both the orbital environment and the hardware operating in it.

The 70-satellites-per-week production line trains engineers at scale. Some of those engineers will stay on the manufacturing side. Others will migrate toward the newer problems Starfall represents, including reentry dynamics, payload recovery, and orbital logistics. The Pacific Northwest aerospace labor market is already tight. SpaceX's Redmond facility is producing not just satellites but a generation of engineers whose skills transfer to the broader orbital-infrastructure economy that companies across the industry are racing to build.

Working in space? Zero G Talent tracks the openings: browse space jobs, openings at SpaceX, and the people building the field.