by Ashlee Vance
The pending launch ignited Musk’s salesman instincts. He wanted to show the public what his tireless workers had accomplished and drum up some excitement around SpaceX. Musk decided to unveil a prototype of Falcon 1 to the public in December 2003. The company would haul the seven-story-high Falcon 1 across the country on a specially built rig and leave it—and the SpaceX mobile launch system—outside of the Federal Aviation Administration’s headquarters in Washington, D.C. An accompanying press conference would make it clear to Washington that a modern, smarter, cheaper rocket maker had arrived.
This marketing song and dance didn’t sound sensible to SpaceX’s engineers. They were working more than one hundred hours per week to make the actual rocket that SpaceX would need to be in business. Musk wanted them to do that and build a slick-looking mock-up. Engineers were called back from Texas and assigned another ulcer-inducing deadline to craft this prop. “In my mind, it was a boondoggle,” Hollman said. “It wasn’t advancing anything. In Elon’s mind, it would get us a lot of backing from important people in the government.”
While making the prototype for the event, Hollman experienced the full spectrum of highs and lows that came with working for Musk. The engineer had lost his regular glasses weeks earlier when they slipped off his face and fell down a flame duct at the Texas test site. Hollman had since made do by wearing an old pair of prescription safety glasses,* but they too were ruined when he scratched the lenses while trying to duck under an engine at the SpaceX factory. Without a spare moment to visit an optometrist, Hollman started to feel his sanity fray. The long hours, the scratch, the publicity stunt—they were all too much.
He vented about this in the factory one night, unaware that Musk stood nearby and could hear everything. Two hours later, Mary Beth Brown appeared with an appointment card to see a Lasik eye surgery specialist. When Hollman visited the doctor, he discovered that Musk had already agreed to pay for the surgery. “Elon can be very demanding, but he’ll make sure the obstacles in your way are removed,” Hollman said. Upon reflection, he also warmed to the long-term thinking behind Musk’s Washington plan. “I think he wanted to add an element of realism to SpaceX, and if you park a rocket in someone’s front yard, it’s hard to deny it,” Hollman said.
The event in Washington ended up being well received, and just a few weeks after it took place, SpaceX made another astonishing announcement. Despite not having even flown a rocket yet, SpaceX revealed plans for a second rocket. Along with the Falcon 1, it would build the Falcon 5. Per the name, this rocket would have five engines and could carry more weight—9,200 pounds—to low orbit around Earth. Crucially, the Falcon 5 could also theoretically reach the International Space Station for resupply missions—a capability that would open up SpaceX for some large NASA contracts. And, in a nod to Musk’s obsession with safety, the rocket was said to be able to complete its missions even if three of the five engines failed, which was a level of added reliability that had not been seen in the market in decades.
The only way to keep up with all of this work was to do what SpaceX had promised from the beginning: operate in the spirit of a Silicon Valley start-up. Musk was always looking for brainy engineers who had not just done well at school but had done something exceptional with their talents. When he found someone good, Musk was relentless in courting him or her to come to SpaceX. Bryan Gardner, for example, first met Musk at a space rave in the hangars at the Mojave airport and a short while later started talking about a job. Gardner was having some of his academic work sponsored by Northrop Grumman. “Elon said, ‘We’ll buy them out,’” Gardner said. “So, I e-mailed him my resume at two thirty A.M., and he replied back in thirty minutes addressing everything I put in there point by point. He said, ‘When you interview make sure you can talk concretely about what you do rather than use buzzwords.’ It floored me that he would take the time to do this.” After being hired, Gardner was tasked with improving the system for testing the valves on the Merlin engine. There were dozens of valves, and it took three to five hours to manually test each one. Six months later, Gardner had built an automated system for testing the valves in minutes. The testing machine tracked the valves individually, so that an engineer in Texas could request what the metrics had been on a specific part. “I had been handed this redheaded stepchild that no one else wanted to deal with and established my engineering credibility,” Gardner said.
As the new hires arrived, SpaceX moved beyond its original building to fill up several buildings in the El Segundo complex. The engineers were running demanding software and rendering large graphics files and needed high-speed connections between all of these offices. But SpaceX had neighbors who were blocking an initiative to connect all of its buildings via fiber optic lines. Instead of taking the time to haggle with the other companies for right of way, the IT chief Branden Spikes, who had worked with Musk at Zip2 and PayPal, came up with a quicker, more devious solution. A friend of his worked for the phone company and drew a diagram that demonstrated a way to squeeze a networking cable safely between the electricity, cable, and phone wires on a telephone pole. At 2 A.M., an off-the-books crew showed up with a cherry picker and ran fiber to the telephone poles and then ran cables straight to the SpaceX buildings. “We did that over a weekend instead of taking months to get permits,” Spikes said. “There was always this feeling that we were facing a sort of insurmountable challenge and that we had to band together to fight the good fight.” SpaceX’s landlord, Alex Lidow, chuckled when thinking back to all of the antics of Musk’s team. “I know they did a lot of hanky stuff at night,” he said. “They were smart, needed to get things done, and didn’t always have time to wait for things like city permits.”
Musk never relented in asking his employees to do more and be better, whether it was at the office or during extracurricular activities. Part of Spikes’s duties included building custom gaming PCs for Musk’s home that pushed their computational power to the limits and needed to be cooled with water running through a series of tubes inside the machines. When one of these gaming rigs kept breaking, Spikes figured out that Musk’s mansion had dirty power lines and had a second, dedicated power circuit built for the gaming room to correct the problem. Doing this favor bought Spikes no special treatment. “SpaceX’s mail server crashed one time, and Elon word for word said, ‘Don’t ever fucking let that happen again,’” Spikes said. “He had a way of looking at you—a glare—and would keep looking at you until you understood him.”
Musk had tried to find contractors that could keep up with SpaceX’s creativity and pace. Instead of always hitting up aerospace guys, for example, he located suppliers with similar experience from different fields. Early on, SpaceX needed someone to build the fuel tanks, essentially the main body of the rocket, and Musk ended up in the Midwest talking to companies that had made large, metal agricultural tanks used in the dairy and food processing businesses. These suppliers also struggled to keep up with SpaceX’s schedule, and Musk found himself flying across the country to pay visits—sometimes surprise ones—on the contractors to check on their progress. One such inspection took place at a company in Wisconsin called Spincraft. Musk and a couple of SpaceX employees flew his jet across the country and arrived late at night expecting to see a shift of workers doing extra duty to get the fuel tanks completed. When Musk discovered that Spincraft was well behind schedule, he turned to a Spincraft employee and informed him, “You’re fucking us up the ass, and it doesn’t feel good.” David Schmitz was a general manager at Spincraft and said Musk earned a reputation as a fearsome negotiator who did indeed follow up on things personally. “If Elon was not happy, you knew it,” Schmitz said. “Things could get nasty.” In the months that followed that meeting, SpaceX increased its internal welding capabilities so that it could make the fuel tanks in El Segundo and ditch Spincraft.
Another salesman flew down to SpaceX to sell the company on some technology infrastructure equipment. He was doing the standard relationship-building exercise practic
ed by salespeople for centuries. Show up. Speak for a while. Feel each other out. Then, start doing business down the road. Musk was having none of it. “The guy comes in, and Elon asks him why they’re meeting,” Spikes said. “He said, ‘To develop a relationship.’ Elon replied, ‘Okay. Nice to meet you,’ which basically meant, ‘Get the fuck out of my office.’ This guy had spent four hours traveling for what ended up as a two-minute meeting. Elon just has no tolerance for that kind of stuff.” Musk could be equally brisk with employees who were not hitting his standards. “He would often say, ‘The longer you wait to fire someone the longer it has been since you should have fired them,’” Spikes said.
Most of the SpaceX employees were thrilled to be part of the company’s adventure and tried not to let Musk’s grueling demands and harsh behavior get to them. But there were some moments where Musk went too far. The engineering corps flew into a collective rage every time they caught Musk in the press claiming to have designed the Falcon rocket more or less by himself. Musk also hired a documentary crew to follow him around for a while. This audacious gesture really grated on the people toiling away in the SpaceX factory. They felt like Musk’s ego had gotten the best of him and that he was presenting SpaceX as the conqueror of the aerospace industry when the company had yet to launch successfully. Employees who made detailed cases around what they saw as flaws in the Falcon 5 design or presented practical suggestions to get the Falcon 1 out the door more quickly were often ignored or worse. “The treatment of staff was not good for long stretches of this era,” said one engineer. “Many good engineers, who everyone beside ‘management’ felt were assets to the company, were forced out or simply fired outright after being blamed for things they hadn’t done. The kiss of death was proving Elon wrong about something.”
Early 2004, when SpaceX had hoped to launch its rocket, came and went. The Merlin engine that Mueller and his team had built appeared to be among the most efficient rocket engines ever made. It was just taking longer than Musk had expected to pass tests needed to clear the engine for a launch. Finally, in the fall of 2004, the engines were burning consistently and meeting all their requirements. This meant that Mueller and his team could breathe easy and that everyone else at SpaceX should prepare to suffer. Mueller had spent SpaceX’s entire existence as the “critical path”—the person holding up the company from achieving its next steps—working under Musk’s scrutiny. “With the engine ready, it was time for mass panic,” Mueller said. “No one else knew what it was like to be on critical path.”
Lots of people soon found out, as major problems abounded. The avionics, which included the electronics for the navigation, communication, and overall management of the rocket, turned into a nightmare. Seemingly trivial things like getting a flash storage drive to talk to the rocket’s main computer failed for undetectable reasons. The software needed to manage the rocket also became a major burden. “It’s like anything else where you find out that the last ten percent is where all the integration happens and things don’t play together,” Mueller said. “This process went on for six months.” Finally, in May 2005, SpaceX transported the rocket 180 miles north to Vandenberg Air Force Base for a test fire and completed a five-second burn on the launchpad.
Launching from Vandenberg would have been very convenient for SpaceX. The site is close to Los Angeles and has several launchpads to pick from. SpaceX, though, became an unwelcome guest. The air force gave the newcomer a cool welcome, and the people assigned to manage the launch sites did not go out of their way help SpaceX. Lockheed and Boeing, which fly $1 billion spy satellites for the military from Vandenberg, didn’t care for SpaceX’s presence, either—in part because SpaceX represented a threat to their business and in part because this startup was mucking around near their precious cargo. As SpaceX started to move from the testing phase to the launch, it was told to get in line. They would have to wait months to launch. “Even though they said we could fly, it was clear that we would not,” said Gwynne Shotwell.
Searching for a new site, Shotwell and Hans Koenigsmann put a Mercator projection of the world up on the wall and looked for a name they recognized along the equator, where the planet spins faster and gives rockets an added boost. The first name that jumped out was Kwajalein Island—or Kwaj—the largest island in an atoll between Guam and Hawaii in the Pacific Ocean and part of the Republic of the Marshall Islands. This spot registered with Shotwell because the U.S. Army had used it for decades as a missile test site. Shotwell looked up the name of a colonel at the test site and sent him an e-mail, and three weeks later got a call back with the army saying they would love to have SpaceX fly from the islands. In June 2005, SpaceX’s engineers began to fill containers with their equipment to ship them to Kwaj.
About one hundred islands make up the Kwajalein Atoll. Many of them stretch for just a few hundred yards and are much longer than they are wide. “From the air, the place looks like these beautiful beads on a string,” said Pete Worden, who visited the site in his capacity as a Defense Department consultant. Most of the people in the area live on an island called Ebeye, while the U.S. military has taken over Kwajalein, the southernmost island, and turned it into part tropical paradise and part Dr. Evil’s secret lair. The United States spent years lobbing its ICBMs from California at Kwaj and used the island to run experiments on its space weapons during the “Star Wars” period. Laser beams would be aimed at Kwaj from space in a bid to see if they were accurate and responsive enough to take out an ICBM hurtling toward the islands. The military presence resulted in a weird array of buildings including hulking, windowless trapezoidal concrete structures clearly conceived by someone who deals with death for a living.
To get to Kwaj, the SpaceX employees either flew on Musk’s jet or took commercial flights through Hawaii. The main accommodations were two-bedroom affairs on Kwajalein Island that looked more like dormitories than hotel rooms, with their military-issued dressers and desks. Any materials that the engineers needed had to be flown in on Musk’s plane or were more often brought by boat from Hawaii or the mainland United States. Each day, the SpaceX crew gathered their gear and took a forty-five-minute boat ride to Omelek, a seven-acre, palm-tree-and vegetation-covered island that would be transformed into their launchpad. Over the course of several months, a small team of people cleared the brush, poured concrete to support the launchpad, and converted a double-wide trailer into offices. The work was grueling and took place in soul-sapping humidity under a sun powerful enough to burn the skin through a T-shirt. Eventually, some of the workers preferred to spend the night on Omelek rather than make the journey through rough waters back to the main island. “Some of the offices were turned into bedrooms with mattresses and cots,” Hollman said. “Then we shipped over a very nice refrigerator and a good grill and plumbed in a shower. We tried to make it less like camping and more like living.”
The sun rose at 7 A.M. each day, and that’s when the SpaceX team got to work. A series of meetings would take place with people listing what needed to get done, and debating solutions to lingering problems. As the large structures arrived, the workers placed the body of the rocket horizontally in a makeshift hangar and spent hours melding together all of its parts. “There was always something to do,” Hollman said. “If the engine wasn’t a problem, then there was an avionics problem or a software problem.” By 7 P.M., the engineers wound down their work. “One or two people would decide it was their night to cook, and they would make steak and potatoes and pasta,” Hollman said. “We had a bunch of movies and a DVD player, and some of us did a lot of fishing off the docks.” For many of the engineers, this was both a torturous and magical experience. “At Boeing you could be comfortable, but that wasn’t going to happen at SpaceX,” said Walter Sims, a SpaceX tech expert who found time to get certified to dive while on Kwaj. “Every person on that island was a fucking star, and they were always holding seminars on radios or the engine. It was such an invigorating place.”
The engineers were constantly b
affled by what Musk would fund and what he wouldn’t. Back at headquarters, someone would ask to buy a $200,000 machine or a pricey part that they deemed essential to Falcon 1’s success, and Musk would deny the request. And yet he was totally comfortable paying a similar amount to put a shiny surface on the factory floor to make it look nice. On Omelek, the workers wanted to pave a two-hundred-yard pathway between the hangar and the launchpad to make it easier to transport the rocket. Musk refused. This left the engineers moving the rocket and its wheeled support structure in the fashion of the ancient Egyptians. They laid down a series of wooden planks and rolled the rocket across them, grabbing the last piece of wood from the back and running it forward in a continuous cycle.
The whole situation was ludicrous. A start-up rocket company had ended up in the middle of nowhere trying to pull off one of the most difficult feats known to man, and, truth be told, only a handful of the SpaceX team had any idea how to make a launch happen. Time and again, the rocket would get marched out to the launchpad and hoisted vertical for a couple of days, while technical and safety checks would reveal a litany of new problems. The engineers worked on the rocket for as long as they could before laying it horizontal and marching it back to the hangar to avoid damage from the salty air. Teams that had worked separately for months back at the SpaceX factory—propulsion, avionics, software—were thrust together on the island and forced to become an interdisciplinary whole. The sum total was an extreme learning and bonding exercise that played like a comedy of errors. “It was like Gilligan’s Island except with rockets,” Hollman said.
