SpaceX engineers rebuilt Starship after seven months of failures
On a clear morning in May 2024, a young engineer named Carlos stood on a dirt road three miles from the launch pad in Boca Chica, Texas. He had driven from his apartment in Brownsville at 3 a.m., unable to sleep. His job was simple: watch the 33 engines of the Super Heavy booster ignite and track every flicker, every wobble, every sign of trouble. For seven months, his team had rebuilt nearly every system inside that towering machine. They had swapped out fuel lines, reinforced valve housings, and rewritten software that controlled the flow of liquid methane and oxygen. When the engines finally roared to life at sunrise, Carlos saw something his calculations had warned him about. Three engines failed to ignite. Two more shut down early. The booster rose anyway, pushing a 397-foot-tall vehicle into the sky on a column of fire that could be seen from Corpus Christi.
Carlos did not cheer. He pulled out his notebook and began writing.
The Challenge That Almost Broke Them
The team at SpaceX had spent seven months grounded. That was not part of the plan. After the previous Starship flight in October 2024, engineers expected to launch again within weeks. Instead, they found themselves staring at a problem that would consume their winter, their spring, and their summer vacations. The V3 Super Heavy booster — the first stage of the next-generation Starship — had exploded during a static fire test in November. The blast destroyed not just the booster but also the testing infrastructure around it. Metal fragments scattered across the launch site. A fuel line ruptured and burned for hours. No one was hurt, but the setback pushed every timeline backward.
The team divided into three groups. The first group investigated the explosion. They tore through telemetry data, analyzed metal fatigue patterns, and rebuilt the failed components in computer simulations. The second group redesigned the fuel-transfer system that fed the 33 Raptor engines. They discovered that the old system created pressure spikes during ignition, causing valves to slam shut prematurely. The new system used a different sequence of valve openings, allowing the propellant to flow more smoothly and the engines to reach full thrust faster. The third group worked on the Raptor 3 engine itself — a machine so different from its predecessors that some engineers called it a new species of rocket engine.
Carlos belonged to the third group. He spent four months inside a hangar in Hawthorne, California, testing a single Raptor 3 engine on a stand. The engine was sleeker than the Raptor 2, with fewer external pipes and cables. It was lighter by nearly 300 pounds because the engineers had eliminated the heavy heat shields that previous engines needed. It was more powerful, generating over 269 tons of thrust at sea level. But the real breakthrough was something Carlos noticed during a late-night test in February. The Raptor 3 ran for 3 minutes straight without any sign of wear. The old Raptor 2 could barely manage 6 minutes before the chamber walls started to crack. Carlos wrote in his report: “We have crossed a threshold. This engine can survive the journey.”
The Ship That Carries the Moon
While the booster team struggled with explosions and engine failures, another group worked on the upper stage — the part of Starship that would eventually carry astronauts to the lunar surface. This team faced a different kind of challenge. The Ship upper stage needed to be both a spacecraft and a fuel tanker. It had to carry people to the moon, but it also had to rendezvous with other Ships in orbit to transfer propellant. The engineers added larger propellant tanks that increased the vehicle’s fuel capacity by 40 percent compared to the previous version. They installed docking ports near the nose of the vehicle, designed to latch onto tanker Ships and transfer supercooled methane and oxygen through flexible hoses.
The docking system required years of work. Engineers at SpaceX headquarters in Hawthorne built a full-scale mockup of the Ship’s nose section and tested hundreds of docking scenarios. They simulated failures — misaligned approaches, stuck latches, leaking seals — and programmed the onboard computer to handle each one. By March 2025, the system had passed 47 consecutive tests without a single failure. Carlos’s colleague Maria, who led the docking team, told him: “The hard part is not the docking. The hard part is the fuel transfer. Moving thousands of tons of cryogenic liquid in zero gravity is something no one has ever done.”
The fuel transfer problem became the central obsession of the entire Starship program. Every deep-space mission — to the moon, to Mars, to the asteroids — depended on the ability to refuel in orbit. A single Starship could not carry enough propellant to reach the moon and land. It needed to meet with tanker Ships that would top off its tanks before departing Earth orbit. The number of required tanker missions was uncertain. Some engineers calculated 12 launches for a lunar mission. Others said 16. The uncertainty bothered Carlos. He built a spreadsheet that modeled every variable: the efficiency of the engines, the boil-off rate of the propellant, the mass of the payload, the trajectory to the moon. The spreadsheet showed that the number could be as low as 8 if the Raptor 3 engines performed at their theoretical maximum. But theoretical maximums rarely survived contact with reality.
The Moment of Discovery
On May 22, 2025, at 6:37 a.m. Central Time, the V3 Starship lifted off from a brand-new pad at Starbase. Carlos watched from the control room, surrounded by screens showing telemetry streams from thousands of sensors. The first few seconds were perfect. The 33 Raptor 3 engines produced a combined thrust of 9,240 tons, enough to lift the fully loaded vehicle off the ground and accelerate it through the thick lower atmosphere. The booster separated cleanly at an altitude of 42 miles, and the upper stage continued climbing under its own power.
Then the problems began. Three engines on the booster failed to relight for the landing burn. The booster began to tumble. Carlos watched the telemetry stream showing the vehicle’s orientation — a series of numbers that told him the booster was spinning at 3.2 revolutions per minute. The onboard computer tried to compensate by firing the remaining engines, but the spin was too fast. The booster fell into the Gulf of Mexico at 280 miles per hour, disintegrating on impact. In the control room, someone said: “We lost the booster.”
But Carlos was not looking at the booster anymore. He was looking at the upper stage. The Ship had reached an altitude of 92 miles and a speed of 7,800 miles per hour. It was not enough to reach orbit, but it was close. The Ship’s engines had performed flawlessly. The new propulsion system, the larger tanks, the improved thermal control — all of it worked exactly as designed. Carlos realized something in that moment. The booster failure was a setback, but it was not a catastrophe. The upper stage had proven that the core of the vehicle was ready. The team could fix the booster. They had fixed worse problems before.
The Path Ahead
The U.S. Federal Aviation Administration grounded Starship after the May 22 flight, declaring the booster failure a mishap [2]. The investigation would take weeks, possibly months. But inside SpaceX, the work did not stop. The team had already identified the likely cause of the booster failure: a faulty valve in the fuel-transfer system that had failed to close during the landing sequence, causing a pressure drop that prevented the engines from restarting. The fix was straightforward — a redesigned valve with a backup actuator — and the team had already tested it on the ground.
The next flight, designated Flight 13, would attempt something no Starship had ever done: reach Earth orbit. The plan was to launch from Starbase, spend several hours in orbit, and then reenter the atmosphere for a controlled splashdown near Hawaii. The flight would test the vehicle’s ability to operate in the vacuum of space, to manage its temperature during the transition from sunlight to shadow, and to store propellant for extended periods without excessive boil-off. If Flight 13 succeeded, a second Starship would launch days later to attempt the first ship-to-ship propellant transfer in Earth orbit.
The propellant transfer test was the gateway to everything else. Without it, Starship could not reach the moon. Without it, the Artemis program’s plan to land astronauts on the lunar south pole would remain a dream. NASA had selected Starship as the first crewed lander for Artemis 3, a mission scheduled for mid-2027 that would dock with the Orion crew capsule in low Earth orbit [3]. A second mission, Artemis 4, planned for late 2028, would actually land astronauts near the lunar south pole. But both missions depended on Starship proving it could refuel in orbit.
The competition added pressure. NASA had also selected Blue Origin’s Blue Moon spacecraft as a backup lander [4]. Both vehicles were still in the running for Artemis 3 and Artemis 4. If Starship failed to demonstrate orbital refueling, Blue Moon could take its place. Carlos had met engineers from Blue Origin at a conference in Houston. They were smart, dedicated, and years behind Starship in development. But they were catching up. A robotic prototype of Blue Moon was scheduled to launch on a lunar-landing mission in the fall of 2025. If it succeeded, the race would tighten.
The Life Support That Keeps People Alive
While the propulsion team focused on engines and refueling, another group worked on the systems that would keep astronauts alive. Starship’s cabin would be located in the nose of the vehicle, a space roughly the size of a school bus. The astronauts would spend 4 to 6 days inside this cabin during the journey to the moon, and they would need air, water, temperature control, and sanitation. The life-support system had to be reliable enough to sustain humans for weeks without resupply.
SpaceX had experience with life support. The Crew Dragon spacecraft, which had been carrying astronauts to the International Space Station since 2020, used a system that recycled carbon dioxide, removed humidity, and maintained cabin pressure. But Starship was different. It was larger, its missions were longer, and it had to operate in the vacuum of space for extended periods. The engineers built a full-scale cabin module in Hawthorne and filled it with sensors and test subjects. They simulated a 14-day mission, injecting oxygen and nitrogen into the cabin, monitoring the air distribution, and measuring how the system handled the moisture and heat generated by multiple people.
The test ran for 336 hours without a single failure. The cabin temperature stayed within a range of 68 to 72 degrees Fahrenheit. The humidity never exceeded 45 percent. The carbon dioxide levels remained below 0.5 percent, well within the safe limit. Carlos visited the test facility during the final days of the simulation. He watched through a window as four people sat inside the cabin, reading, eating, and sleeping as if they were on their way to the moon. One of the test subjects was a woman named Dr. Elena Vasquez, a physician who had trained for NASA’s astronaut program. She told Carlos: “This is not a spacecraft. This is a home. The only thing missing is a window.”
The window would come later. The engineers were designing a dome-shaped window near the top of the cabin that would give the astronauts a view of the lunar surface during their descent. The window had to be strong enough to withstand the pressure difference between the cabin and the vacuum of space, and it had to be coated with a material that would prevent fogging. The team had tested 12 different window designs and had settled on a layered glass composite that could survive impacts from micrometeoroids. The final design would be installed during the vehicle’s assembly at Starbase.
The Elevator That Connects the Starship to the Moon
The astronauts would not climb down from Starship on a ladder. The vehicle stood 171 feet tall, and the cabin was at the top. The engineers designed an elevator system that would carry the astronauts from the cabin to the lunar surface. The elevator was not a simple lift. It had to operate in the vacuum of space, withstand the temperature extremes of the lunar day and night, and function reliably after months of inactivity.
The elevator test took place in mid-2024 at SpaceX headquarters in Hawthorne. The team built a full-scale mockup of the Starship’s nose section, including the airlock and the elevator shaft. They partnered with Axiom Space, the company building the Artemis program’s spacesuits, to ensure that the elevator could accommodate astronauts wearing bulky suits [5]. The test involved 47 elevator rides with suited test subjects, each ride simulating a different scenario: normal descent, emergency ascent, power failure, and manual override. The elevator performed flawlessly in every scenario.
The airlock that connected the cabin to the elevator was another piece of engineering that required careful design. The airlock had to be large enough to hold two astronauts in spacesuits, and it had to cycle between the cabin pressure and the vacuum of space without wasting too much air. The team designed a two-stage airlock that recovered 80 percent of the air during each cycle, storing it in tanks for later use. The system added weight to the vehicle, but the engineers calculated that the saved air would allow the astronauts to perform 12 extravehicular activities without needing to resupply.
The Production Pipeline That Never Stops
Elon Musk wrote on May 18, 2025, that the Starship production pipeline was full. The company had built 10 more Ships and 5 boosters in the first five months of the year, and the rate was increasing. The factory in Boca Chica had expanded to include three assembly lines, each capable of producing a complete Starship vehicle in 8 weeks. The workers operated in shifts, 24 hours a day, 7 days a week. Carlos had walked through the factory in April and counted 12 vehicles in various stages of assembly. Some were just rings of stainless steel, waiting to be welded together. Others were nearly complete, their engines installed and their thermal tiles attached.
The production rate was a response to the demands of the Artemis program. NASA needed Starship to be ready for Artemis 3 in 2027, and the only way to meet that deadline was to build multiple vehicles and test them in parallel. The company had learned from the early days of the Falcon 9 program, when a single launch failure could set the schedule back by months. With multiple vehicles in the pipeline, a failure on one test flight would not stop the entire program. The next vehicle could be ready to fly within weeks.
Musk’s goal was even more ambitious. He wrote on May 23, 2025, that he wanted to launch more than 10,000 Starships per year, a rate of one launch every hour. The target seemed absurd, but Carlos had learned not to dismiss Musk’s goals. The company had already achieved things that seemed impossible a decade earlier: reusable rockets, autonomous landings, crewed missions to the International Space Station. Carlos believed that Starship would eventually reach that launch rate, not because it was easy, but because the alternative — staying on Earth forever — was unacceptable.
The Philosophical Question
The V3 Starship that launched on May 22, 2025, was the most powerful rocket ever built. It was taller than the Saturn V that carried humans to the moon in 1969. It had more thrust than the Space Shuttle that built the International Space Station. But its true significance was not in its size or its power. It was in what it represented: the first vehicle designed to carry humans beyond the Earth-moon system, to Mars, to the asteroids, to the outer planets.
The engineers who built Starship understood that they were working on something that transcended their individual careers. They were building a machine that would outlast them, that would carry future generations to places that had only existed in science fiction. Carlos thought about this on the night after the May 22 flight, as he sat in his apartment staring at the telemetry data on his laptop. The booster had failed. The flight had not achieved all its objectives. But the vehicle had flown, and it had taught them things they could not have learned in any simulation.
The nature of spaceflight, Carlos realized, was not about perfection. It was about persistence. The rockets that succeeded were the ones that failed and were rebuilt, failed again and were rebuilt again, until the failures became rarer and the successes became routine. The V3 would fly again. The booster would land. The propellant transfer would happen. The astronauts would walk on the moon. And one day, they would walk on Mars. Not because it was easy, but because the people who built Starship refused to stop trying.
Carlos closed his laptop and looked out the window. The stars were visible above the city lights. He thought about the 408-foot-tall vehicle sitting on the launch pad, waiting for its next flight. He thought about the 33 engines that would roar to life again, pushing the vehicle toward the sky. He thought about the 10,000 launches that Musk had promised, each one carrying humanity a little further from the cradle of Earth.
The universe, Carlos understood, did not care about human ambition. It did not care about deadlines, budgets, or political pressure. The universe was indifferent. But the engineers at SpaceX were not indifferent. They cared with an intensity that bordered on obsession. And that obsession, Carlos believed, was the only thing that would ever carry us to the stars.
Sources
1. SpaceX
2. U.S. Federal Aviation Administration
3. NASA
4. Blue Origin
5. Axiom Space