The Engine That Melted 50 Chambers Finally Gets Its Gigafactory.

#SpaceX's Raptor Engine Manufacturing Automation Blitz in Hawthorne Quietly Building the World's First Mass-Produced Full-Flow Staged Combustion Engine Workforce — and the Flight Test 5 Cadence Signals What Happens When Engine Production Becomes the Pacing Item

Engines Now Pace Starship

Each Starship flight burns 39 Raptor engines — 33 on the Super Heavy booster, six on the upper stage. Flight 4 lifted off June 6, 2024. Flight 5 followed October 13. Booster 12 and Ship 30 had stacked by early August, then waited on an FAA license that slipped from mid-September to late November before arriving October 12, one day before launch. Hardware outpaces paperwork.

Starbase integrates vehicles faster than the range clears them. At two flights a year, 78 engines annually — a rate Hawthorne's current line supports. But SpaceX's architecture demands a fleet in the thousands. The math collapses unless engine production shifts from batch builds to continuous flow.

Zero G Talent's board shows SpaceX added 111 roles in the past week. The latest Hawthorne posting: Automation & Controls Engineer for Raptor Manufacturing Systems. The title is deliberate — not "propulsion engineer" but "automation & controls." The bottleneck has moved from test stand to production cell.

Flight 5 proved the vehicle can fly, stage, boost back, and be caught. The next 100 flights need 3,900 engines. The hundred after that need another 3,900. Hawthorne is now the pacing item.

Chamber and Nozzle: Building the Cells

SpaceX posted two chamber-and-nozzle roles the same week: a Manufacturing Automation Engineer and a Manufacturing Engineer, both entry-level, both in Hawthorne, both targeting the same components. The automation role drew 48 applicants in seven days; the CNC-focused role sat under 25. The gap shows where hiring pressure sits.

The automation posting is explicit: "create the machines that build the Raptor engine's combustion chamber and nozzle." Not tend machines. Create them. Responsibilities read like a Gigafactory spec: robotic process automation, smart factory initiatives, in-situ monitoring, data pipelines feeding statistical process control and machine learning models for defect reduction, fixture and material handling solutions holding micron-level tolerances. Preferred skills span Python, PLC logic (Beckhoff, Siemens, Fanuc, ABB, Kuka), machine vision, real-time process control, and direct experience with regeneratively cooled hardware. The CNC counterpart demands multi-axis milling, turning, and precision finishing of cooling channels, plus CAM software, G-code, on-machine metrology (touch probing, laser scanning), adaptive machining, and digital twin integration with version-controlled post-processors.

Both roles require floor time: eight-hour minimum stands, stooping, bending, extended hours and weekends, 10 percent travel to launch and test sites. ITAR restricts eligibility to U.S. citizens, permanent residents, refugees, and asylees.

The cluster signals a cell-by-cell build-out: one team owning chamber machining automation, another nozzle automation, a third stitching them into a digital thread from CAD through CNC telemetry to predictive quality models.

Regenerative cooling channels (intricate, thin-walled, high-temperature alloy) cannot be inspected after the fact. They demand in-process metrology, closed-loop force and vibration monitoring, tool-wear prediction, and fixture systems locating large, flimsy parts to micron repeatability across hundreds of reuse cycles. No prior full-flow staged combustion engine has attempted this at volume. Merlin's open-cycle gas generator was simpler. RD-180 and BE-4 never reached this cadence. SpaceX builds the production system and the engine simultaneously.

The hiring profile reflects that reality. They want automation engineers who have machined metal, welded, heat-treated, and run non-destructive inspection — not just aerospace specialists. The "preferred"preferred" list reads like a semiconductor or automotive controls shop: Python, PLC, robotics, vision, massive data pipelines, statistical process control, digital thread. The chamber and nozzle cells are where Raptor manufacturing starts to look like a Gigafactory rather than a rocket factory.

Cape Canaveral: The Launch-Site Mirror

SpaceX is building a second automation hub at Cape Canaveral Space Force Station that mirrors Hawthorne's Raptor production logic, not for engine assembly yet but for launch-site infrastructure enabling weekly Starship cadence. The job board shows a cluster of automation and controls roles tied to "Starship Launch Hardware" and "Starship Launch Pad": Automation & Controls Engineer, Automation & Controls Specialist, Build Specialist for Automation & Controls, Data & Control Systems Engineer, and Data & Control Systems Technician. All sit in Electrical or Automation & Controls and Maintenance categories, distinct from Starbase vehicle integration roles.

The Automation & Controls Engineer role for Starship Launch Hardware makes the mandate explicit: "designing and activating the systems that build and launch Starship" across the rocket launch facility. Responsibilities span the full stack: architecting PLC code (preferably Siemens TIA Portal), designing SCADA infrastructure (Ignition software), laying out NFPA79/NEC/UL508A-compliant electrical control panels, managing cabinet builds and field wiring, writing activation procedures for technicians. The Specialist role parallels this with a technician-grade entry point: high school diploma plus three years' controls experience, hazardous-area classification expertise (Class 1, Division 1/2), and the same PLC/SCADA toolchain. Both require Air Force background clearance and up to 15% travel.

This is not pad construction. It is the control-layer build-out for a launch complex that must process a Super Heavy booster and Ship stack, recover them, and turn them around without shipping major hardware back to Hawthorne or Starbase. The same control philosophy (reusable, maintainable PLC code; standardized cabinet design; SCADA visibility across distributed systems) that SpaceX developed for Raptor chamber and nozzle cells in Hawthorne is being ported to the Cape's ground support equipment, propellant generation, and tower/launch mount automation. Mechanical Engineer roles for those systems listed alongside the controls positions confirm the scope: fluid systems, structural mechanisms, and their control systems staffed in parallel.

The end state is local engine capability. A launch site that can swap Raptors on a recovered booster, run acceptance checks, and re-stack without a cross-country logistics tail is the only way to hit flight rates implied by Flight 5's cadence. The Cape's automation hiring is the leading edge — controls engineers who can commission a test cell today and an engine build cell tomorrow.

Full-Flow Staged Combustion at Scale: Manufacturing Problems No One Else Has Faced

Raptor isn't a higher-pressure Merlin. It is the first full-flow staged combustion engine to reach orbit, and the only one ever mass-produced. That distinction creates manufacturing problems no previous program (not RS-25, not RD-180, not BE-4) has faced at volume.

Pressure Ceiling

Raptor 3 operates at 330 bar (4,785 psi) chamber pressure. RD-180 tops out at 267 bar. RS-25 at 206 bar. Merlin at 97 bar. Each step up in cycle complexity enables higher pressure, cascading through the vehicle: smaller chamber for the same thrust, higher exhaust velocity, more specific impulse. But 330 bar means every seal, weld, and bolted joint sees loads with no heritage in production rocketry. During development, SpaceX melted more than 50 chambers and exploded more than 20 engines.

Hot Oxygen Eats Metal

The oxidizer-rich preburner is the central materials challenge. Hot, high-pressure oxygen doesn't just corrode nickel superalloys — above certain temperatures and pressures it makes the metal catch fire. The Soviets solved this for RD-170/180 with specialized alloys that remained state secrets for decades. SpaceX developed its own: SX300 Inconel, then SX500, engineered to contain hot oxygen at up to 830 bar (12,000 psi) in the turbopump discharge. Methane helps (it burns cleaner than kerosene, producing less soot that could serve as ignition sources on hot oxygen-side surfaces), but the materials problem remains the single hardest barrier to FFSC production.

Two Complete Turbopump Systems

A gas-generator engine like Merlin has one turbopump. RS-25 has one fuel-rich preburner driving both pumps. Raptor has two independent preburners and two independent turbopumps (oxidizer-rich and fuel-rich), each producing tens of thousands of horsepower. That means twice the rotating hardware, twice the bearing packages, twice the seal sets, and a gearbox to synchronize them. Every part must survive the same 330 bar main-chamber environment. No heritage supply chain exists; SpaceX machines housings, prints impellers, and heat-treats alloys in-house.

Gas-Gas Injection Demands Impossible Geometry

In gas-generator or conventional staged-combustion engines, at least one propellant enters the main chamber as liquid. In FFSC, both arrive as hot, high-pressure gases. Combustion efficiency exceeds 99 percent because mixing is nearly instantaneous: no droplets to atomize, no liquid to vaporize. But coaxial swirl injectors distributing two gaseous streams uniformly at 330 bar, with internal regenerative cooling channels woven through the injector face, cannot be cast or machined. They must be grown. The 2016 subscale Raptor had 40 percent of its parts by mass 3D-printed; Raptor 3's monolithic injector plate, with graded cellular cooling lattices printed in situ, is a single piece replacing a hand-stacked assembly of hundreds of elements.

The Part-Count War

Musk has called Raptor 1 a "crazy spaghetti-mess" of pipes, wires, and sensors. Raptor 2 converted flanges to welds and deleted entire subsystems. Raptor 3 moved most plumbing and sensors inside the engine's main structure, eliminating the external heat shield entirely. The August 2025 redesign cut part count another 30 percent: integrated turbopump housing printed as one geometry (eliminating weld seams and leak paths), nozzle throat with graded cellular lattice, 60 percent fewer brazed joints, 45 percent fewer machining hours. Each generation looks less like a rocket engine and more like a solid block of metal with flow paths grown inside it.

Inspection at Scale

You cannot visually inspect a cooling channel printed inside an injector plate. You cannot dye-penetrant a lattice structure inside a nozzle throat. Non-destructive evaluation requires industrial CT scanning and acoustic tomography, specialized equipment that must keep pace with a production target of one engine per day, then hundreds per year. SpaceX embeds machine-vision CNNs on powder-bed printers, acoustic-emission sensors on sintering furnaces, and near-infrared spectroscopy on debind lines, feeding a digital thread tracing every powder lot to every finished part. No staged combustion program has ever needed this inspection infrastructure at rate.

No Playbook Exists

Aerojet Rocketdyne knows fuel-rich staged combustion. Energomash knows oxidizer-rich. Neither has built a full-flow production line. The IPD program proved FFSC worked with LOX/hydrogen at 275 bar, then was cancelled. The RD-270 ran 27 hot-fire tests in the 1960s but couldn't tame combustion instability. SpaceX inherited the dataset and solved the rest from scratch: proprietary superalloys, generative design tooling, high-speed binder-jet printing at 200 cm³/hour, AI-driven process control pausing a build when a CNN detects a powder-layer anomaly. The workforce hiring for this (automation and controls engineers owning the full stack from PLC logic to robotic end-effectors) looks nothing like traditional aerospace propulsion hiring. It looks like semiconductor fab staffing.

Workforce Shift: From Propulsion Specialists to Automation Generalists

SpaceX's latest Raptor automation posting lists a bachelor's degree and one year automating metallic-component manufacturing as the only hard requirements. Rocket experience is "highly desired" — not required. The preferred-skills column reads like a recruiting brief for a Tesla body shop or Samsung fab: Python, PLCs from Siemens and Beckhoff, robots from Fanuc, ABB, and those automation skills, and data pipelines for predictive analytics.

Those platforms dominate automotive, semiconductor, and consumer-electronics lines. They do not dominate aerospace. A traditional propulsion shop hires engineers who know regenerative cooling channels and injector patterns; this posting hires engineers who know how to make a robot hold ±0.025 mm on a thin-walled Inconel nozzle while a vision system inspects every weld bead. TheLadders categorizes the role under "Manufacturing & Automotive," not "Aerospace & Defense."

The hiring velocity suggests SpaceX is staffing a production system, not an engine program. The workforce now resembles a Gigafactory's: controls engineers writing PLC logic by day and Python models by night, technicians swapping end-effectors on Kuka arms, data engineers building the telemetry backbone for statistical process control.

The shift is deliberate. Raptor 3 consolidates parts into single 3D-printed assemblies. That moves difficulty from assembly fixtures to in-situ monitoring, closed-loop quality, and predictive scrap reduction: problems automotive and semiconductor industries solved years ago. SpaceX imports the solutions, not reinvents them.

Vertical Integration Extended: In-House Automation Equipment Design

SpaceX does not buy automation cells off the shelf. It designs them: fixturing, inspection systems, motion-controlled workstations, the PLC logic sequencing a robotic weld, the end-effector geometry gripping a 3D-printed injector plate without marring its internal cooling channels. The same engineers who write the ladder logic also specify servo drives, tune vision systems, and validate safety zones. That full-stack ownership mirrors how SpaceX builds its own avionics, test stands, and Starlink user terminals. The factory is a product.

The Automation & Controls Engineer posting for Raptor Manufacturing Systems asks for experience designing and commissioning industrial automation and controls systems — not "integrating vendor solutions." Designing and commissioning. The distinction is the point.

Tim Berry, who ran Falcon upper-stage production for a decade before joining JetZero, laid out the sequence at an AIAA forum in July 2024. Step 1: challenge requirements. Step 2: delete parts and process steps. Step 3: simplify. Step 4: go faster. Step 5 — only then — automate. "Most people start with Step 5," Berry said, "and they automate a process that never should have existed in the first place." SpaceX's automation engineers live inside that discipline. They don't receive a frozen process and bolt a robot onto it. They sit in design reviews that cut part count 30 percent, fuse external plumbing into the motor casting, turn a 15-piece turbopump housing into a single print. Then they build the cell that makes the result repeatable.

The additive line proves the model. SpaceX installed nearly 30 Velo3D laser powder-bed machines under an $8 million agreement ($5 million for technology licensing, $3 million for engineering support) and secured the right to modify the platform for internal use. The machines on the floor are not stock. SpaceX runs "highly-customized variants of commercially available additive manufacturing and post-processing technologies, as well as in-house developed solutions," according to a 2025 technical deep-dive. The same control engineers who tune the laser scan strategy also write the in-situ monitoring: convolutional neural networks scanning each powder layer at micron resolution, recurrent neural networks listening to acoustic emissions during sintering, near-infrared spectroscopy tracking binder decomposition in real time. When a CNN flags a binder smear, the printer pauses, refreshes powder locally, and reprints the layer. Scrap drops 15 percent. That loop (sensor, model, actuator) is designed, integrated, and maintained by SpaceX controls engineers.

The digital thread extends the same ownership upstream. Every powder lot carries a QR code. Hopper feed rates, binder-jet logs, debind furnace cycles, HIP parameters: all append to a distributed ledger. If a nozzle throat shows anomalous crack growth, the trace runs back to a specific powder shipment, a humidity reading in the storage silo, a thermal ramp deviation in the sinter furnace. The automation engineer owns that traceability architecture, not a vendor's MES module.

At Kennedy Space Center, the new Gigabay applies the same logic at facility scale: 24 specialized work cells, 400-ton cranes, standardized interfaces so a cell proven in Hawthorne replicates in Florida without re-engineering. The Automation & Controls Specialists hired at the Cape are building that mirror: electrical and control systems across the launch site, from pad umbilicals to post-flight inspection robots.

Vertical integration usually stops at the product. SpaceX extended it to the equipment that makes the product. The component, its graded cellular cooling lattice, its integrated turbopump housing: none of those geometries exist without the printers SpaceX modified, the inspection cells SpaceX programmed, the fixturing SpaceX machined in-house. The engine is the output. The automation stack is the lever.

What Flight Test 5 Reveals About the Production Rate Target

Flight Test 5 flew October 13, 2024, with Booster 12 and Ship 30, totaling 39 Raptor engines (33 sea-level on the booster, three sea-level and three vacuum on the ship). Serial numbers alone tell a story: B12 and S30 imply at least 12 boosters and 30 ships built or in the pipeline, yet only five integrated flights had launched. Gaps (Booster 8 never flew, Ships 26, 27, and 32 absent from the flight record) suggest hardware produced faster than the launch cadence could absorb.

Cadence accelerated sharply after Flight 5. Flight 4 launched that date; the next flight followed 130 days later. Flight 6 launched November 19, 2024, 37 days after Flight 5. Flight 7 January 16, 2025 (58 days), Flight 8 March 6, 2025 (49 days), Flight 9 May 27, 2025 (82 days). The trend compresses: from roughly one flight per quarter to one every 6–8 weeks. At that rate, 39 engines per flight implies annualized consumption of roughly 250–300 engines (before reuse enters the picture).

Reuse changes the math. Elon Musk said after Flight 5 the booster "looked great" with only "a few outer engine nozzles warped from heating & some other minor issues, but these are easily addressed." He added the system is "designed to achieve reflight of its rocket booster ultimately within an hour after liftoff." Flight 7 marked the first reflown Raptor; Flight 9 flew 29 reflown engines on Booster 14.2. Each successful catch and reflight reduces net-new engine demand per flight, but the production rate must still support the initial fleet build and attrition of engines that don't meet reuse criteria.

Hiring velocity in Hawthorne confirms the factory is scaling for that demand. Roles cluster around chamber and nozzle automation cells, safety logic, and motion-controlled systems: the exact disciplines needed to turn Raptor's complex full-flow staged combustion hardware (3D-printed oxidizer turbopumps, oxygen-rich preburners, chamber pressures exceeding 300 bar) into a product built on a line, not a bench.

McGregor test data corroborates the throughput push. NASASpaceflight reported 24 Raptor firings in a single week in late 2024, and Raptor 3 (lighter, higher-thrust, with integrated manifolds) entered "scaled production by early 2025." Publicly stated goals target "over 1,000 engines annually." Flight Test 5's serial numbers, the post-Flight 5 cadence, and the Hawthorne automation hiring spike all point to the same conclusion: SpaceX is building the factory to hit that 1,000-engine rate in 2025–2026, not just the flight line to consume it. The pacing item has shifted from "can the vehicle fly?" to "can the engine line deliver?" — and the automation cells coming online in Hawthorne and the Cape are the answer.

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