Elon Musk

On September 28, 2008, a Falcon 1 rocket lifted off from Omelek Island in the Kwajalein Atoll, a speck of coral in the Marshall Islands so remote that the journey from Los Angeles required two commercial flights, a prop plane, and a boat. It was a Sunday. Elon Musk was nearly broke. The first three Falcon 1 launches had failed—a fuel leak and fire in March 2006, an engine shutdown in March 2007, a collision between the first and second stages in August 2008, this last one destroying a payload that included the ashes of James Doohan, the actor who played Scotty on Star Trek, a detail so cosmically on-the-nose that it might have been scripted by the cruelest kind of dramatist. SpaceX had enough money for one more attempt. Tesla, the electric car company Musk had poured $50 million into, was hemorrhaging cash and weeks from insolvency. His first marriage had dissolved. He was borrowing money from friends to pay rent.

The fourth rocket reached orbit.

It is the closest thing to a founding myth that the age of technology billionaires has produced—the entrepreneur who bet everything, lost three times, and on the fourth try bent the arc of an industry toward his will. But founding myths, by their nature, flatten the thing they celebrate. The story of Elon Musk is not really a story about persistence, though persistence is in it. It is not really a story about genius, though that word gets deployed constantly, often by people who have worked alongside him and insist there is no other word adequate to the gap between his mind and theirs. It is, instead, a story about a particular kind of mind—one that reasons from physics rather than analogy, that finds social consensus not just unhelpful but actively dangerous, and that operates with a tolerance for risk and a craving for crisis that most people would classify as pathological. The fact that this mind has, in the span of three decades, helped create the dominant electronic payment system, the world's most valuable car company, the world's most prolific rocket company, a global satellite internet constellation, a brain-computer interface that has been implanted in a human skull, and one of the most chaotic acquisitions in the history of social media—this fact does not resolve the question of whether the pathology is a feature or a bug. It sharpens it.

The biographical sketch is well known, but its internal contradictions are less frequently examined. Elon Reeve Musk was born on June 28, 1971, in Pretoria, South Africa, to Errol Musk—an engineer, rogue, and, in Walter Isaacson's phrase, "charismatic fantasist"—and Maye Musk, a Canadian-born model and dietitian whose own mother had flown a single-engine plane from South Africa to Australia. The family was wealthy. The childhood was brutal.

As a boy, Musk was regularly beaten by bullies. One day a group pushed him down concrete steps and kicked him until his face was, in Isaacson's description, "a swollen ball of flesh." He spent a week in the hospital. But the physical violence, Isaacson argues in Elon Musk, was secondary to the emotional damage inflicted by his father, whose impact on his son's psyche "would linger." Musk developed into what Isaacson calls "a tough yet vulnerable man-child, prone to abrupt Jekyll-and-Hyde mood swings, with an exceedingly high tolerance for risk, a craving for drama, an epic sense of mission, and a maniacal intensity that was callous and at times destructive."

The boy taught himself to program. At twelve he created a video game and sold it to a computer magazine. At fourteen, reading Nietzsche and Schopenhauer and spiraling into existential crisis, he found Douglas Adams's The Hitchhiker's Guide to the Galaxy, which reframed the problem. "A lot of times the question is harder than the answer," Musk later explained. "If you can properly phrase the question, then the answer is the easy part." The book suggested to him that expanding the scope of human consciousness and knowledge was, if not the meaning of life, at least the best available proxy. He was fourteen years old, and he had already identified the organizing principle of his career: the expansion of the possible, pursued through the physics of the actual.

In 1988, at seventeen, he obtained a Canadian passport through his mother's citizenship and left South Africa. He was unwilling to support apartheid through compulsory military service. He was also running—from his father, from the bullies, from a country whose constraints he had outgrown. The pattern would repeat: Musk has spent his life leaving places that are too small for the scale of his ambition, and building new places from scratch when the existing ones refuse to accommodate him.

Queen's University in Kingston, Ontario. Then the University of Pennsylvania, where he earned bachelor's degrees in physics and economics in 1997. Then Stanford, for a graduate program in applied physics and materials science. He lasted two days.

"He felt that the Internet had much more potential to change society than work in physics," the Britannica entry notes with characteristic understatement. This was 1995. Netscape had gone public two months earlier. The browser was a revelation, and Musk—who had spent his adolescence in a country that in many ways was still fighting the seventeenth century—recognized immediately that the internet was the largest leverage point available to a twenty-four-year-old with no connections, no money, and no credentials beyond an undergraduate education and an almost frightening capacity for self-instruction.

He and his brother Kimbal founded Zip2, a company that provided maps and business directories to online newspapers. They could not afford an apartment, so they rented a small office, slept on the couch, and showered at the YMCA. They had one computer: the website ran during the day; Elon coded at night. In 1999, Compaq bought Zip2 for $307 million. Musk was twenty-eight years old and had roughly $22 million after taxes.

What he did next is the first clear marker of the difference between Musk and the era's other dot-com millionaires. He did not diversify. He did not cash out. He founded X.com, an online financial services company, and poured almost everything back in. X.com merged with a competing payment startup called Confinity, whose signature product was a service for transferring money via email—PayPal.

The PayPal years were formative but also combative. The company was a proving ground not just for Musk but for an extraordinary cohort that would become known as the PayPal Mafia: Peter Thiel, a contrarian philosopher-investor who cofounded PayPal and would later write Zero to One; Max Levchin, a Ukrainian-born cryptographer who had designed PayPal's fraud-detection systems; Reid Hoffman, who would go on to found LinkedIn; and a half-dozen others who would collectively seed much of Silicon Valley's next decade. Thiel and Musk clashed repeatedly over the company's direction—Musk favored an aggressive financial-services expansion built on X.com's brand; Thiel wanted to focus on the payment product. Musk was ousted as CEO in 2000, replaced by Thiel, but remained the company's largest shareholder.

In 2002, eBay bought PayPal for $1.5 billion. Musk walked away with approximately $170 million after selling his stock. He was thirty-one. And then, in a move that still astonishes people who were present for it, he invested nearly all of it: $100 million into SpaceX, which he founded in 2002. $50 million into Tesla, which he joined as lead investor in 2004. $10 million into SolarCity. Instead of starting another internet company—the rational play for a software engineer with a proven track record—he entered two of the most capital-intensive, failure-prone, lobbying-entangled industries on Earth: aerospace and automotive manufacturing.

The question was not whether this was brave. It was whether it was sane.

The origin of SpaceX is a story about being told no and deciding the problem was with the people saying it. In 2001, Musk approached aerospace engineer Jim Cantrell with a question that sounded like the setup for a science fiction novel: How do we become a multiplanetary species? Cantrell, a veteran of the missile-defense world, was intrigued enough to accompany Musk on a trip to Russia to negotiate the purchase of refurbished intercontinental ballistic missiles. The Russians quoted absurd prices, and one of them reportedly spat on Musk's shoes.

On the flight back from Moscow, Musk opened a spreadsheet and began calculating the cost of building a rocket from scratch. He had read textbooks on orbital mechanics, propulsion, and structural engineering—Cantrell later recalled that Musk had devoured Rocket Propulsion Elements by George Sutton, Fundamentals of Astrodynamics, and the Soviet-era textbook Mechanics and Thermodynamics of Propulsion. "He'd quote passages verbatim from these books," Cantrell told Esquire. The spreadsheet showed that the raw materials for a rocket constituted roughly 2% of the typical launch price. The rest was margin, overhead, and the accumulated inefficiency of a monopolistic industry that had not been seriously disrupted since the Saturn V program.

SpaceX was founded in June 2002, headquartered in a warehouse in El Segundo, California, later moving to Hawthorne. Musk was CEO and, crucially, chief designer. This was not a vanity title. He personally involved himself in engine design, avionics, and structural decisions at a level of detail that shocked the experienced aerospace engineers he hired. His method was what he called "first principles reasoning"—breaking problems down to their fundamental physics rather than reasoning by analogy from what had been done before.

The first Falcon 1 launched from the Kwajalein Atoll in March 2006. It failed because of a fuel leak and fire. The second launch, in March 2007, reached space but not orbit. The third, in August 2008, failed when the first and second stages collided after separation. Each failure was specific, diagnosable, and—this mattered—different from the last. SpaceX was not repeating mistakes. It was making new ones, which in engineering is called progress.

Then came September 28, 2008. Flight four. The Falcon 1 reached orbit, making SpaceX the first privately funded company to place a liquid-fueled rocket in Earth orbit. Three months later, NASA awarded SpaceX a $1.6 billion contract to service the International Space Station. The company had come within weeks of extinction and was saved not by luck but by the compounding effect of iterative failure applied at extraordinary speed.

By 2015, SpaceX had achieved something the entire aerospace industry had declared impossible: landing a Falcon 9 first stage on solid ground after orbital insertion. By 2017, it had reflown a recovered booster. By 2020, it had launched astronauts Doug Hurley and Robert Behnken to the ISS in a Dragon capsule—the first crewed orbital spaceflight by a private company. By 2024, more than half of the world's orbital launches were conducted by SpaceX. The company had not merely entered the launch market. It had consumed it.

Tesla's origin story is often told as Musk's, but it isn't—not entirely. The company was founded in 2003 by Martin Eberhard and Marc Tarpenning, two Silicon Valley engineers who believed the lithium-ion battery had reached a density that made a high-performance electric car viable. Eberhard was CEO; Tarpenning was CFO. They named it after Nikola Tesla, the Serbian-American inventor who had been marginalized by Edison and died alone in a New York hotel room. The name was both homage and provocation.

Musk came in as the largest investor in Tesla's 2004 Series A round, contributing $6.5 million of the $7.5 million raised and becoming chairman of the board. He was not a passive investor. He pushed the design of the Roadster toward a more ambitious specification, insisting on a carbon-fiber body and performance metrics that would make it competitive with high-end gasoline sports cars. The Roadster, released in 2008, could travel 245 miles on a single charge and accelerate from 0 to 60 mph in under four seconds. It cost $109,000. It was a luxury item, not a mass-market product, and that was the point: the strategy, which Musk has described explicitly, was to enter at the top of the market, use the margins to fund development of progressively cheaper vehicles, and eventually produce an affordable car that could compete with the Honda Civic.

Eberhard was pushed out as CEO in late 2007 and formally departed in 2008. Tarpenning left the same year. Musk took over as CEO. The transition was acrimonious. But the product roadmap—Roadster, then luxury sedan, then mass market—held.

The Model S arrived in 2012 and was immediately acclaimed as one of the finest sedans ever made, electric or otherwise. The Model X luxury SUV followed in 2015. The Model 3, the car intended to be Tesla's iPhone—the device that moved electric vehicles from niche to mainstream—went into production in 2017 and became the best-selling electric car of all time.

But the Model 3 nearly killed the company. Tesla's Fremont factory descended into what Musk later called "production hell." The problem was not the car's design but the manufacturing process. Musk had attempted to automate every step of assembly, and the robots kept breaking down, jamming, and producing defective parts. He ended up sleeping on the factory floor—literally, on a couch next to the production line—for weeks. He later admitted this was a mistake born of his own algorithm: he had tried to automate (step five) before questioning requirements, deleting unnecessary processes, simplifying, and accelerating (steps one through four). "The big mistake in Nevada and at Fremont was that I began by trying to automate every step," he said. "We should have waited until all the requirements had been questioned, parts and processes deleted, and the bugs were shaken out."

Tesla's IPO in 2010 raised $226 million. By 2020, the company's market capitalization surpassed that of every other automaker on Earth. By 2025, Tesla's annual revenue had reached approximately $94.8 billion, its market cap hovered around $1.5 trillion, and Musk had secured shareholder approval for a compensation package valued at up to $1 trillion—tied to operational targets including an $8.5 trillion market capitalization—that was among the largest in corporate history.

The Cybertruck, with its angular stainless-steel exoskeleton that looked like something designed by a twelve-year-old playing a video game (or, more charitably, like a vehicle from a Ridley Scott film), entered limited production in late 2023 after years of delay. It was polarizing. Everything Musk touches is polarizing. This is not incidental to his method; it is, in some hard-to-articulate way, central to it.

At the beginning of 2022—after a year in which SpaceX had launched thirty-one rockets into orbit, Tesla had sold over a million cars, and Musk had become the richest person on Earth—he spoke ruefully about his compulsion to manufacture crisis. "I need to shift my mindset away from being in crisis mode, which it has been for about fourteen years now, or arguably most of my life," he said. It was a wistful comment. Not a resolution. Even as he said it, he was secretly buying shares of Twitter.

The acquisition of Twitter—later renamed X—is the most confounding chapter of Musk's career, the one that resists the gravitational pull of the "visionary founder" narrative. It does not fit neatly into a story about multiplanetary civilization or the electrification of transport. It fits, instead, into a story about a man whose demons and drives are inseparable, who returns compulsively to the sites of old wounds.

"Over the years, whenever he was in a dark place, his mind went back to being bullied on the playground," Isaacson writes. "Now he had the chance to own the playground."

The timeline is a study in whiplash. In late January 2022, Musk began buying Twitter shares in near-daily installments. By mid-March he held a 9.2% stake, making him the largest shareholder. On April 4, the stake was publicly disclosed; Twitter's stock jumped 27%. On April 5, he was offered a board seat. On April 10, he declined it. On April 14, he offered to buy the entire company for $54.20 a share—roughly $44 billion. (The price, like the $420 Tesla-going-private tweet that had earned him an SEC lawsuit in 2018, was widely interpreted as a marijuana joke: 54.20, 420.)

Twitter's board accepted. Then Musk tried to back out, claiming Twitter had misrepresented the number of bot accounts on the platform. Twitter sued to force the deal. In October 2022, facing trial in Delaware Chancery Court, Musk completed the acquisition.

What followed was the most public demolition of a social media company in history. Within days, Musk fired roughly half of Twitter's 7,500 employees. He sent a company-wide email with the subject line "Fork in the Road"—the same subject line he had used when offering resignation packages to Twitter employees—offering those who remained a choice between "extremely hardcore" work and severance. He disbanded the content-moderation council. He reinstated banned accounts, most notably Donald Trump's, which had been suspended after the January 6, 2021, Capitol attack. He replaced the verification system—blue check marks previously bestowed by Twitter on notable public figures—with an $8-per-month subscription. Advertisers fled. Revenue collapsed.

In July 2023, he renamed the company X.

The conventional interpretation is that Musk overpaid for an asset he then proceeded to destroy, driven by a quixotic commitment to "free speech" that was really an exercise in impulsive power. There is significant evidence for this view. But there is another reading, one that takes Musk's stated intentions more seriously than his critics are willing to: that he saw Twitter as a data asset and a distribution platform that, integrated with his AI ambitions (xAI, Grok) and eventually with SpaceX's Starlink, could form the communication backbone of an empire that spans from payments to rockets. In early 2026, SpaceX acquired xAI—which itself had been training its machine-learning models on X's user data—folding AI, social media, and spaceflight into a single corporate ecosystem.

Whether this is strategic brilliance or post-hoc rationalization for a $44 billion impulse purchase may not be knowable for years.

There is a passage in Isaacson's biography that functions almost as a Rosetta Stone for understanding how Musk operates. "I became a broken record on the algorithm," Musk told Isaacson. "But I think it's helpful to say it to an annoying degree."

The algorithm has five steps:

The algorithm came with corollaries: all technical managers must have hands-on experience. Comradery is dangerous because it prevents people from challenging each other's work. "It's OK to be wrong. Just don't be confident and wrong." Never ask your troops to do something you're not willing to do. "A maniacal sense of urgency is our operating principle." And, most characteristically: "The only rules are the ones dictated by the laws of physics. Everything else is a recommendation."

This is not, it should be said, the advice you will find in the Harvard Business Review. It is not designed to create a pleasant workplace. It is designed to create the kind of workplace that lands rockets on drone ships and ramps automotive production from zero to a million units in a decade. The human cost is significant and well-documented: burnout, turnover, lawsuits, estranged colleagues, a trail of people who describe Musk as both the most brilliant and the most difficult person they have ever worked with. The question the algorithm raises is not whether it works—the rockets fly, the cars sell—but whether it is replicable by anyone who does not have Musk's specific combination of technical depth, risk tolerance, and apparent indifference to social friction.

While the public argued about Twitter's blue check marks, Musk's less visible ventures were achieving milestones that, in a less noisy era, would have dominated front pages for weeks.

Starlink, the satellite internet constellation SpaceX began launching in 2019, had grown to approximately 7,000 active satellites by 2025—more than half of all active satellites in orbit. The Federal Communications Commission had authorized SpaceX to deploy up to 15,000 second-generation satellites, and the company's long-term plans called for a constellation numbering in the tens of thousands. Starlink had over five million subscribers. It provided internet service to remote communities, disaster zones, and—controversially—combat zones. The Ukrainian military relied on Starlink for battlefield communications, a dependency that gave Musk a geopolitical leverage no private citizen had held since perhaps the era of the East India Company.

Neuralink, the brain-computer interface company Musk cofounded in 2016, reached its own threshold in early 2024 when it successfully implanted a device called Telepathy in a 29-year-old patient with quadriplegia. The recipient could control a mouse cursor on a screen, browse the web, and play chess—all using thought alone. A second patient, who had sustained a spinal injury, received an improved implant later that year. The company's next announced product, Blindsight, aimed to restore vision in patients with partial or complete blindness.

The Boring Company, born of Musk's frustration with Los Angeles traffic in late 2016, had a more modest trajectory. It built a pair of tunnels beneath the Las Vegas Convention Center for approximately $47 million—a fraction of conventional tunneling costs—and opened a shuttle system in 2021. But the original vision of autonomous electric sleds traveling at 125 to 150 mph had been quietly replaced by human-driven Tesla cars moving at 40 mph. Proposed projects in Baltimore, Chicago, Los Angeles, and Washington, D.C. were "later discarded or indefinitely shelved."

xAI, founded nine months after the Twitter acquisition, was developing Grok, an AI assistant positioned as a rival to OpenAI's ChatGPT and Google's Gemini. The company trained its models on data from X's user base and publicly available sources. It was, in structural terms, the connective tissue between the social media platform and the broader AI race—and, after xAI's acquisition by SpaceX, between all of those things and the space business.

The pattern across these ventures is consistent: Musk identifies a domain where existing solutions are orders of magnitude more expensive or less capable than physics allows, applies first-principles reasoning to redesign the core technology, and then scales relentlessly while absorbing criticism, litigation, and the departure of people who cannot sustain the pace. The pattern's results are undeniable. Its sustainability—for the organizations, for the people inside them, for Musk himself—is an open question.

In July 2024, following an assassination attempt on Donald Trump, Musk publicly endorsed the Republican presidential candidate. Through his America Political Action Committee, he became the country's largest political donor, contributing $288 million to Trump and other Republican candidates. He appeared at campaign rallies. He set up a sweepstakes paying $1 million a day to registered voters in swing states, a move that prompted lawsuits alleging it was illegal.

Trump, who had snubbed Musk at an earlier White House electric vehicle summit (the Biden administration had pointedly excluded Tesla in favor of unionized automakers), now promised Musk a role leading the Department of Government Efficiency—DOGE, a reference to the Dogecoin cryptocurrency meme that Musk had long championed. Musk vowed to "balance the budget immediately" by cutting $2 trillion in government spending, which would "involve some temporary hardship."

What followed was, depending on one's political orientation, either a necessary demolition of federal bloat or a constitutional crisis in real time. DOGE teams took over the Office of Personnel Management on the afternoon of Trump's inauguration, January 20, 2025. More than two million government employees received a deferred-resignation email with the subject line "Fork in the Road"—an echo, again, of the Twitter playbook. DOGE gained access to Treasury Department databases. Musk announced on X that he and Trump would shut down USAID; its website was taken down, most of its 10,000 employees placed on administrative leave, programs worldwide suspended. The National Institutes of Health saw $4 billion in indirect-cost funding cut. The Institute of Education Sciences had most of its contracts canceled.

Lawsuits proliferated. Critics charged that the executive branch was violating the separation of powers by refusing to spend congressionally appropriated funds. Republicans who controlled Congress countered that there was nothing unusual in a new administration reviewing spending.

On May 28, 2025, Musk stepped down from DOGE. He said he was "disappointed" with Trump's proposed "one big beautiful bill" for potentially increasing the deficit and undermining DOGE's cost-cutting measures. The Congressional Budget Office estimated the bill could increase the deficit by $2.4 trillion. In the days that followed, Musk and Trump traded public barbs. Trump threatened to revoke government contracts. Tesla's market capitalization briefly shed more than $150 billion.

The arc from rocket designer to government efficiency czar to public feuding with the president who gave him the job is, at minimum, difficult to square with a coherent strategic vision. It is easier to read as the behavior of a man who, having bent two industries to his will, found a bigger system to disrupt—and discovered, perhaps for the first time, that the federal government is not a startup, physics is not its operating system, and the algorithm does not apply when the requirements are set by 535 members of Congress with their own survival incentives.

Someone once asked Musk on Reddit how he learns so much so fast. His answer has become one of the most frequently cited passages in the self-improvement internet, and it deserves scrutiny not because it is wrong but because it reveals something about the architecture of his mind:

"I think most people can learn a lot more than they think they can. They sell themselves short without trying. One bit of advice: it is important to view knowledge as sort of a semantic tree—make sure you understand the fundamental principles, i.e., the trunk and big branches, before you get into the leaves/details or there is nothing for them to hang on to."

The metaphor is botanical but the impulse is structural: learn the load-bearing elements first, and everything else becomes a decoration you can hang or remove at will. This is how Musk taught himself orbital mechanics, propulsion, and materials science well enough to co-design rockets. It is how he moved from software to finance to automotive to aerospace without the usual apprenticeship period in any of them. It is also, potentially, why he underestimates the complexity of domains—politics, media, organizational psychology—where the trunk and branches are not governed by physics but by human irrationality, and where first-principles reasoning is less useful than pattern recognition, empathy, and the ability to read a room.

Jim Cantrell, the aerospace engineer who accompanied Musk to Russia, was asked on Quora how Musk learned enough about rockets to run SpaceX. Cantrell didn't really answer the question as stated. Instead he answered a more interesting one: what allows Musk to attempt things everyone else considers impossible? The answer, Cantrell suggested, was not a single trait but a cluster of them—an unusual combination of intellectual voracity, first-principles reasoning, monomaniacal focus, and a genuine inability to internalize the word "no" as anything other than evidence that the person saying it has not thought hard enough.

In the spring of 2025, voters in Cameron County, Texas—near the U.S.-Mexico border, approximately twenty miles from Brownsville—approved the incorporation of a new city called Starbase. The vote was 212 to 6, a margin that would have been suspicious in any normal election but was simply a function of the electorate: roughly 280 people, most of whom either worked for SpaceX or had a relative who did. Two-thirds were men. The average age was about twenty-seven. Many had never voted in any election before.

The city snaked around parcels of land owned by SpaceX, incorporating the company's launch pad, rocket manufacturing facility, and employee housing—Airstream trailers, modular homes, and some larger houses. A building bore a huge mural depicting a colony on Mars. The planned school was called Ad Astra—"to the stars." Bobby Peden, a SpaceX vice president, was elected mayor.

Starbase is, in one reading, a company town of the kind that would have been familiar to Andrew Carnegie or Henry Ford—a place where the employer provides not just work but the social infrastructure of daily life. In another reading, it is something stranger: the first physical foothold of a man who intends to build cities on other planets, practicing the art of municipal creation on the only planet currently available. The New York Times described it as "one of the nation's more unusual elections." No campaigning took place. There were no yard signs.

It was not clear whether Musk voted.

By 2025, Elon Musk's net worth was estimated at $393 billion, making him the richest person on Earth by a margin that itself exceeded the GDP of most countries. He ran—or had recently run—six major companies: Tesla, SpaceX, X (formerly Twitter), xAI, Neuralink, and The Boring Company. He had fourteen children. He had been married three times (twice to the same woman, the English actress Talulah Riley). His public persona oscillated between that of a visionary engineer and an erratic provocateur.

He had cofounded OpenAI in 2015 as a nonprofit intended to be a socially responsible counterbalance to Google's lead in AI, then departed in 2018 citing a conflict of interest with Tesla's own AI efforts. OpenAI's management later said Musk had sought to run a for-profit arm or merge the organization with Tesla. In 2024, Musk sued OpenAI and its major investor Microsoft, accusing them of straying from the venture's nonprofit foundations. In February 2025, Musk and a consortium of investors offered to acquire OpenAI's nonprofit arm for $97.4 billion. Sam Altman, OpenAI's CEO, posted his response on Musk's own platform: "no thank you but we will buy twitter for $9.74 billion if you want"—a reference to X's diminished valuation from the $44 billion Musk had paid.

The paradox is not that Musk is complicated—all people are complicated—but that the same traits that produce SpaceX's Falcon 9 also produce the chaos of the Twitter acquisition. The same first-principles mind that redesigned rocket economics also decided to fire half of a social media company's staff in a week. The same craving for crisis that kept him sleeping on the Tesla factory floor also drew him into a political alliance that cost his flagship company $150 billion in market capitalization in a matter of days.

In an interview with Ars Technica in May 2025, just days after stepping down from DOGE, Musk spoke almost entirely about Starship. He was focused on the thermal protection system for reentry—the tiles, the coatings, the gap filler, the attachment techniques. "I feel like we've got about an 80 percent chance of having solved those issues," he said of the engine bay problems that had caused explosions on the previous two flights. "To really have a 100 percent chance, it requires the design iteration on the engine." He described the process of discovering that bolts attaching the thrust chamber to the injector head needed to be retightened after firing—that the seal would gap slightly, allowing fuel and oxidizer to combine in the wrong place, and "it only takes a tiny amount of fuel and oxygen combining in a bad spot to explode the engine."

This is where Musk is most legible: in the granular details of combustion physics, in the iterative debugging of systems that obey the laws of thermodynamics, in the convergence of theory and metal. It is the domain where his semantic tree has the deepest roots and the sturdiest branches. Everything else—the tweets, the politics, the feuds, the company towns, the $44 billion impulse purchases—hangs from those branches at varying distances, in varying states of stability.

On a remote stretch of the Texas Gulf Coast, near a Border Patrol checkpoint on a two-lane highway that closes during launches, a mural on the side of a building depicts a colony on Mars. Inside, engineers are tightening bolts.