25 Comments
User's avatar
Jillian Berk's avatar

Part count fell because the firm kept the vehicle. When the manufacturer owns the asset after flight, every fitting is a cost it carries forever. Under a cost plus contract the same fitting is revenue. Nobody engineered the earlier versions badly. They were optimized against a different payer. Simplification is what happens when the party doing the design also holds the maintenance bill.

Steve Mudge's avatar

Amazing, what Musk has done with SpaceX is astounding. Thanks for the great article!

Mike Randolph — M Raige's avatar

Here is what Musk built, its the whole team that built the Starship, Claude:

Who Actually Builds SpaceX's Starship?

Elon Musk gets the headlines, but Starship, the largest rocket ever flown, is built by a deep and long-serving engineering organization. Here is who runs it, based on public reporting.

The man at the top. Musk is more than SpaceX's CEO. Fortune describes him as the company's chief designer, lead systems engineer, and singular decision-making authority. The big architectural calls are his.

The engineering leaders under him:

• Mark Juncosa, VP of Vehicle Engineering. Juncosa joined in 2005 and has held his current role since 2015. He also helps Gwynne Shotwell, SpaceX's president, oversee day-to-day operations at Starbase, the Texas site where Starship is built and launched.

• Bill Riley, VP of Starship Engineering. A former Ford and GM engineer, Riley joined in 2010, rose through structures engineering, and has led Starship engineering since March 2020.

• Joe Petrzelka, Booster Engineering. Petrzelka has an MIT PhD and joined in 2012. He moved through Falcon and Dragon work before taking responsibility for Super Heavy, Starship's giant first stage.

• Jacob McKenzie, VP of Raptor. McKenzie joined in 2015 as a propulsion engineer and has led development of Raptor, Starship's engine, since July 2022.

• Charles Kuehmann, VP of Materials. Kuehmann's team develops the advanced alloys and composites that make reusable rockets possible. He holds the same job at Tesla.

• Phil Alden and Derek Turner, production. Alden came from a decade at BMW and Jaguar Land Rover and now runs Starship production. Turner oversees manufacturing across all SpaceX vehicles.

• Bill Gerstenmaier, VP of Build and Flight Reliability. For fourteen years he ran NASA's human spaceflight program. At SpaceX he leads quality engineering and launch readiness.

What stands out. Most of these leaders have been at SpaceX ten to twenty years and were promoted from within. The production side deliberately borrows from the car industry, which suits a goal of building rockets in volume.

The open questions.

• How big is the Starship team? Nobody publicly says. SpaceX as a whole has about 22,000 full-time employees, but no Starship-specific figure is public.

• How concentrated is the decision-making? The leadership below Musk is deep, but he holds the final technical say.

• Is attention being pulled elsewhere? After SpaceX absorbed Musk's AI company xAI in February 2026, several of the executives who built the launch business took on xAI oversight too.

The test ahead. The first flight of the newest Starship version lost its booster after an engine burn failed to start properly. Flight 14, scheduled for September 22, will try to reach orbit for the first time.

Summary

Starship is often pictured as Elon Musk's personal project, but the program runs on a seasoned engineering bench. Below Musk sit vice presidents with long tenure and clearly divided jobs. Mark Juncosa has led vehicle engineering since 2015. Bill Riley has run Starship engineering since 2020. Joe Petrzelka handles the booster, Jacob McKenzie the Raptor engine, and Charles Kuehmann materials. Phil Alden and Derek Turner run manufacturing, and Bill Gerstenmaier, formerly NASA's top human-spaceflight official, oversees reliability.

Most of these leaders joined SpaceX between 2005 and 2015 and worked their way up through structures, propulsion, or production roles. That kind of internal promotion builds people who share a common engineering standard. Recruits from BMW and Jaguar Land Rover bring an automotive approach to building rockets at scale.

Three caveats apply. First, the organization below the vice-president level isn't publicly visible, and SpaceX doesn't report how many people work on Starship. Second, authority is concentrated: Musk is the final technical decision-maker, so the team mostly carries out his design direction. Third, since SpaceX merged with xAI in early 2026, some senior leaders have split their attention between rockets and AI.

The real measure of depth is performance under pressure. The newest Starship's first flight lost its booster, and the next attempt, on September 22, aims for orbit. Whether the program keeps its pace through setbacks and distractions will say more about the team than any org chart.

Steve Mudge's avatar

Of course, all true, but Musk had the fortitude to pull it all together into a cohesive organization. I mean Einstein and Schrodinger made great leaps but it was also on the backs of hundreds of other physicists' work.

In the older movies from the 30s and 40s the screen credits were limited to very brief intros for the main cast and producers/directors. Now we can sit through ten minutes of credits from the makeup assistant to the food caterers to second in command of walking the stars dogs during production. I get it, it takes a village, but even with all those highlighted personnel it's easier to remember the main stars and execs.

Giulio Famiglietti's avatar

What I like about this evolution is that ‘simplification’ here doesn’t mean the underlying problem became simple. Years of learning allowed complexity to be engineered out of the operating surface. I’ve seen the same thing in operations: mature processes often look deceptively simple because dozens of exceptions, handoffs and workarounds have gradually been designed away.

Damian Hadley's avatar

Well written article Brian. Thanks for writing. In particular the way you slowly introduced the complexity was great. The article was orchestrated like a rocket version of Mike Oldfields Tubular Bells! Love it.

Russell G. Johnston's avatar

The fact that version 3 of the Raptor has a 1 on it makes it crystal clear that the thing is beyond my comprehension.

Alan Yang's avatar

Lets not forget the real heroes of this story… the 3-D printing technology and the innumerable engineers and technologists needed to keep the machine precise and capable.

I wonder how long it takes to print the various components and to assemble… who knows, if the printing and assembly is reasonably good, this technology can be scaled to situations where you need a few thousand units printed… ie very high end manufacuring and military equipment.

Cazadores de Valor's avatar

Wonderful engineering explanation. One implication we find striking: Super Heavy uses 33 Raptors, so a 35% thrust improvement per engine is equivalent, in aggregate, to adding roughly 11.5 Raptor 1 engines, before counting the mass removed from shields and external plumbing. Simplification therefore compounds twice: more thrust from every engine and less vehicle mass for those engines to lift.

PhilBuildTheFutureNow's avatar

The metallurgy story. Raptor’s oxygen-rich preburner runs with extremely hot, high-pressure oxygen, which is basically a terrible environment for most metals. That’s one reason full-flow staged-combustion engines were so rare before Raptor.

SpaceX developed its own superalloys, including SX300 and SX500, to survive those conditions. “Delete parts” makes for a great headline, but a fix was developing materials that can sit inside what is like an oxygen torch and keep working.

anton's avatar

I am betting on 3d printing. they literally integrated all the cooling into printed parts. usually 3d printed heat exchangers have a much higher powerdensity than conventional heat exchangers. Also theoretically, detailed controll of hydraulic surfaces opens the possibility for self regulating hydraulics similar to analog computers. Also: Musk NEVER mentions 3D printing with spacex.

Neurology For You's avatar

It’s amazing how each generation gets closer to the ideal “cute robot man wearing a hat” shape that is most efficient for an engine!

Gustavo José Zambrano's avatar

Fantastic breakdown, Brian! Stripping away complexity in a full flow staged combustion cycle sounds impossible until you see Raptor 3 in person.

Removing external sensors and routing channels directly into castings reminds me of how real elegance in engineering is always about subtraction rather than addition.

Loved the side by side comparisons!

James Agada's avatar

SpaceX didn't invent the FFSC engine. The concept has been around for some time. Some chinese company has even gone further on the design of the concept. And an Indian company is also racing. SpaceX appears to be the first to have successfully flown an FFSC.

Steven Kan's avatar

The 3D printed layer lines are very interesting. I'd naively assumed that they would have printed this either bottom-up or top-down to maximize radial symmetry, but apparently not!

Can anyone knowledgeable about DLMSed-Inconel opine on the structural implications for tensile strength/elasticity, and the further effect on gas flow?

Emilio Vazquez's avatar

if they can do this with rocket engines, they can do it with semiconductors manufacturing

Tim's avatar

"These engines are simple and reliable, but limited; there’s only so much thrust you can practically get from a cold gas thruster. So how can we get more thrust?"

Could you elaborate on the nature of thrust limitation with this design?

I understand the limitations on specific impulse and thrust-to-weight ratio with this approach. But thrust? always thought that for this design it's a straightforward function of throat area, expansion ratio and tank pressure.

"version 1 and version 2 of the Raptor required a large, heavy heat shield to protect it from the heat of the rocket exhaust"

That is a very unusual take as wel - please elaborate how does heat shield on the outside of engine helps with heat of rocket exhaust that is inside? Plus it's inside of regeneratively [and possibly film-] cooled nozzle, shouldn't additional heat capacity reduce the performance of cooling system?

Always thought that the purpose of the external shielding is to manage heatflow from reentry

SpC's avatar

Build a working something with fewer parts, there’s less to fail when it’s performing its specific task, right?

Think anyone might apply that to our national Government in D.C anytime soon?