Just off the Jimmy Buffett Memorial Highway, the hotel’s rooftop bar is open late. The bartender passes out shots and turns Ozzy up. It’s 11:37 pm on a hot July night in Cape Canaveral, Florida, when our heads all swivel in the same direction. A SpaceX Falcon 9 rocket takes off, its orange plume glowing bright, about 12 miles due north up the Banana River. The “Iron Man” riff starts to blast.

It’s fun for the couple dozen of us there. When we hear the thud of the sonic boom, most everyone lets out some kind of hoot. But for Elon Musk, it’s just another Tuesday. This is SpaceX’s 95th launch of the year, one nearly every other day. That’s more liftoffs than the rest of the world gets into space, combined.

On this particular night, this Falcon 9 took 28 Starlink internet satellites to orbit. Starlink, of course, is another Musk space venture that dominates its competitors. His constellation has more than 8,000 satellites; its closest competitor, Eutelsat’s OneWeb, has about 630 satellites, each supplying less than 1/10th the bandwidth of a Starlink. Amazon is going all in on its own service, called Project Kuiper and led by SpaceX’s former satellite chief. The terms of Kuiper’s license from the feds require it to get 1,600 satellites into orbit by the middle of next year. So far, the Amazon constellation has 102.

It’s hard to quantify, even with those numbers, the geopolitical power that Musk now commands by way of his two space businesses. When Starlink went down for a couple of hours in late July, troops on both sides of the Russia-Ukraine conflict had trouble connecting with their drones—and one another. “Everyone thought it was purely on the front lines, until reports started coming in that he had fallen all over the world,” one officer stationed near the city of Kupiansk, along the Oskil River in eastern Ukraine, texts me. That’s how central Musk is to modern warfare. Two days after the launch I watched from the hotel roof, another Falcon 9 took off from Cape Canaveral, this one carrying four astronauts aboard a Dragon capsule to the International Space Station. SpaceX’s Dragon is currently America’s only way to get humans into space, as Musk reminded his onetime ally Donald Trump when the president threatened Musk’s government contracts.

Now, Musk has a chance to leverage his two dominant positions into a third. For the first time in decades, America is openly working on the weaponization of space, in response to what the Pentagon claims are threats from Russia and China. The Pentagon is investing in spacecraft that can fly up to other countries’ satellites and attack. Separately, the president has pledged $175 billion for a program that could eventually entail hundreds and hundreds of orbiting interceptors and even more communications satellites to allow them to work together.

It’s hard to quantify, even with those numbers, the geopolitical power that Musk now commands by way of his two space businesses.

Musk’s companies are unlikely to build the weapons themselves. But getting them into space, and getting them to talk to one another, that is most certainly in their wheelhouse. So while Musk may not have open access to the Oval Office like he used to, there’s no conceivable way such a buildup won’t benefit SpaceX. The open question is, by how much? When the orbiting rifles are handed out, how many gun lockers will Elon have the keys to?

You might be a little numb at this point to the degree of control that billionaires have over our lives. But you’ve watched Elon Musk stomp and smash and rage his way through politics and policy, even as his companies continue to pull off engineering feats that were once the stuff of sci-fi. So you get what’s at stake if he’s given an outsize role in the weaponization of space. (SpaceX did not respond to requests for comment.)

“The US government depends upon him very heavily,” Victoria Samson, the chief of space security at the Secure World Foundation, tells me. “So even before the election, I had been asking US space officials: ‘You have yoked yourself to a very mercurial personality. Doesn’t that concern you?’”

I. ROCKETS

As recently as the early 2010s, getting to space was expensive and slow. The United States attempted fewer than 20 launches per year. Rockets can cost $10,000 per kilogram or more. Musk and now-legendary rocket engineer Tom Mueller broke through, in part, by being scrappy: They’d swap NASA’s $1,500 latches for ones made for bathroom stalls that cost just $30, and they’d use commercial air conditioners for the Falcon 9’s payload bay rather than buy a cooling system for an estimated $3 million.

While Musk likes to keep up an antiestablishment image, he very much played the Washington game. He drew on his alliances with like-minded people in the government, such as then NASA administrator Michael Griffin, who advocated for cheaper, easier access to space—especially to low Earth orbit, which starts around 100 miles up. When Musk felt others didn’t share that vision, he sued, like the time he alleged that the Air Force had acted illegally when it gave the era’s space monopolist, a joint venture between Boeing and Lockheed Martin called United Launch Alliance, an $11 billion contract for 36 rocket cores.

When the suit didn’t produce instant results, Musk went jingoistic. A few months earlier, in February 2014, Russia had invaded Ukraine, illegally annexing the Crimean Peninsula and triggering a global wave of condemnation against Moscow. Musk rode that wave in his successful push to get Congress and the Obama administration to wind down use of the United Launch Alliance’s signature rocket, the Atlas V, because it relied on Russian RD-180 engines. (The suit was eventually settled out of court.) The combination helped break ULA’s grip on government space launches.

Another big leap came in 2017. SpaceX started reusing its rocket cores, which dramatically brought down the price of getting to orbit. (Eight years later, its Falcon 9 and Falcon Heavy are still the only rockets in their weight classes with reusable cores.) But nothing was more important than Mueller’s continued development of SpaceX’s Merlin engine. It became one of the most durable in aerospace history, even though, as a former employee told me, “performance-wise, it’s terrible.” Its power and efficiency are nothing special. “We didn’t have the resources to do a lot of design and analysis,” he adds. “And so we just tested the ever-loving shit out of the engine. We hot-fired it thousands of times. Now they have an engine that’s super robust.”

Today, thanks in part to its nine reusable Merlin engines, a Falcon 9 can take a kilogram to low Earth orbit for one-third the previous cost; the Falcon Heavy, which uses 27 Merlins, drops the cost nearly in half again. Some 85 percent of Falcon 9 missions go to space with previously used first stages. In 2022, SpaceX jumped from doing around 30 launches per year to more than 60, and last year it hit 138. NASA’s space launch and human exploration efforts are now almost entirely controlled by Musk. A whole new space economy has grown up around him, one that relies on his cheap space access to get networks of small spacecraft into low Earth orbit. Take Planet Labs, the satellite imaging company. Hundreds of its spacecraft were carried by Falcon 9.

Really, no one is even trying to catch up; they’re just trying to find niches in a Musk-dominated ecosystem. ULA is building rockets optimized to reach geostationary orbits, which are farther out, even as many of its customers follow Musk’s lead and keep their satellite constellations closer to Earth. Upstarts like Rocket Lab and Firefly are admired for their ingenuity. But their current operational rockets are tiny by comparison—capable of carrying, at most, a couple thousand pounds, versus 140,000 for the Falcon Heavy.

“SpaceX is a cornerstone in the space industry. And then there’s other cornerstones, like Firefly. We’re very complementary to SpaceX,” says Jason Kim, the CEO of Firefly Aerospace. “It’s kind of like air, land, and sea. There’s no one-size-fits-all kind of transportation method.” (Kim’s not alone in this thinking; Firefly just went public at a valuation of $8.5 billion; Rocket Lab’s market cap is about $21 billion.)

Jeff Bezos has the cash to compete with SpaceX. And he’s certainly been at it long enough—his rocket company, Blue Origin, started a quarter-century ago. But it has had, shall we say, competing priorities. It’s been hard at work on engines; its BE-4 engine is actually powering the first stage of ULA’s new rocket, confusingly enough. You may have seen that Blue Origin has a rocket for near-space tourism, the one that recently carried Bezos’ wife, Lauren Sánchez, and Katy Perry aloft. But the company’s big rocket, the one that’s supposed to compete with SpaceX, has flown exactly once. And when I ask Blue Origin’s rep what makes their rockets any better—or, at least, any different—from Musk’s, he tells me: “I don’t have a solid answer for you on that one.”

China, which once seemed poised to dominate global launch, has had trouble keeping up with Musk’s rising totals, successfully launching between 64 and 68 rockets annually over the past three years. SpaceX is not only launching twice as often, it’s carrying more than 10 times the reported mass to orbit. Stoke Space, founded by Blue Origin engineers, has aerospace geeks in a frenzy, but it has yet to put a rocket on the pad. United Launch Alliance, SpaceX’s OG competitor, has a powerful new rocket—more on that in a bit—but once again, Musk is ahead. He’s working on a truly massive launcher, arguably the biggest ever constructed. Both stages are supposed to be fully reusable (which means, of course, immense cost savings), while neither stage of ULA’s Vulcan will be fully reusable. And that, according to a new report from SpaceNews Intelligence, could relegate the one-time monopolist “to niche roles in government or regional and backup contracts, assuming they survive at all.”

II. SATELLITES

At the end of May, at his factory in Starbase, Texas, Musk was in full Mars evangelist mode. “This is where we’re going to develop the technology necessary to take humanity,” he told his employees, “to another planet for the first time in the four-and-a-half-billion-year history of Earth.”

But as he sketched out his soaring vision of this place cranking out 1,000 enormous Starships per year, Musk repeated a more mundane truth. No, not the part about the Starship’s uneven test record. The one about funding. “Starlink internet is what’s being used to pay for humanity getting to Mars.”

The space launch business is a brutal one, and a space launch business with its eye on Mars is even more impractical. Rockets blow up. Customers are constantly late in their satellite deliveries. But supplying internet service? That’s a way more reliable source of cash. So in 2015, Musk started recruiting engineers to build an internet in orbit.

Big, geostationary spacecraft have been providing internet service from 22,000 miles up since the mid-1990s. The idea of replacing them with a constellation of small, low-flying, low-latency satellites has been around for almost as long. But it wasn’t until Musk brought down the cost of getting to orbit that those plans became not only feasible but massively profitable. Last year, the company earned an estimated $13 billion in revenue, and Starlink accounted for something like 8 of those billions, according to Payload Research. About 70 percent of the company’s launches so far in 2025 have carried Starlink satellites into orbit. Starlink, which started launching in 2019, now claims to have more than 6 million customers, up nearly 50 percent in 12 months. This is why private shares of SpaceX have become highly sought-after investments. (Well, that and the fact that the company seems to pay little or no tax on all that revenue.) As of July, the company was valued at $400 billion, with IPO rumors swirling. SpaceX’s dominance now depends on Starlink.

You could argue Starlink is also supplying an equally outsize share of Musk’s global influence. The deployment of Starlink in 2022 to Ukraine after Russia’s full-scale invasion is by now well known, as are Musk’s decisions to reportedly cut off Starlink over the Kherson region during a crucial Ukrainian counterattack and deny coverage near Russian ships docked in Crimea later that year. Reports of Iranian dissidents getting their hands on Starlink receivers date back to the same period.

On June 13 of this year, Iran’s rulers shut down local internet access after Israel began “Operation Rising Lion,” its air-strike campaign designed to decapitate and destabilize the Tehran government. That same day, right-wing American pundit Mark Levin posted on X that “Elon Musk can put the final nail in the coffin of the Iranian regime by providing Starlink internet to the Iranian people!” To which Musk, a longtime ally of Israeli prime minister Benjamin Netanyahu, responded: “The beams are on.” Iran’s parliament quickly passed a measure imposing prison sentences of up to two years for those caught using Starlink. The number of Starlink users nevertheless spiked to more than 100,000, according to one local report. Musk often lets Starlinks run under the table, sometimes as a way of pressuring governments to license the service legitimately. “But in Iran, they’re just straight-up flouting the rules,” says one knowledgeable observer, “trying to overthrow the government by giving power, giving internet access” to Tehran’s opponents.

It wasn’t the first—or last—time Musk used Starlink in ways that seemed to support the Israeli government’s objectives. As WIRED previously reported, venture capitalists helped the Israel Defense Forces receive Starlink access after the October 7 attacks in 2023. Civilian access in Gaza has reportedly been severely constrained since, with a single hospital being the rare exception. Meanwhile, scam compounds in Myanmar, jihadist networks in the Sahel, and Sudanese rebels have all reportedly had more or less unfettered access to Musk’s satellite network.

Musk isn’t afraid to use his technologies to advance his politics. We’ve seen how his tweaks to X have made it more Nazi-friendly, and how his chatbot will start talking about “white genocide.” We’ve seen Musk enable a war effort, and we’ve seen him deny that aid when it suited him to do so. This has all happened while Starlink is in its relative infancy, before its constellation is complete, before its coverage of the continents is total. The dangers of relying on Starlink have not been lost on geopolitical players: There may be 50,000 Starlink terminals in Ukraine today, but some of the more tech-savvy Ukrainian military units have already weaned themselves off of Musk’s service. “Starlink is not our primary, and sometimes not even the alternative, but rather the contingency,” one officer in the Kharkiv region tells me.

There’s talk in Europe of trying to expand a Starlink competitor. And the Chinese government is working on a pair of low-Earth-orbit constellations—a commercial one for the international market and a government one for the military and intelligence agencies. The plan is to launch 28,000 communication satellites between the two by the 2030s. So far, they’ve launched somewhere around 170, and an alarming number of those have failed in space.

That could leave Amazon as Musk’s most potent rival. The company certainly has the resources. Amazon recently completed a $140 million, 100,000-square-foot facility at the Kennedy Space Center, near Cape Canaveral, to get the satellites ready for launch. It has signed contracts for up to 83 launches from three different rocket companies—several billion dollars’ worth of launch services—to get the Kuiper internet constellation into orbit. Almost half of those launches are with ULA, SpaceX’s original competitor, which is building a new “integration facility” down the road to prep those rockets for Amazon.

Obviously, Amazon has more and better access to consumers than almost any other company on the planet. Kuiper is part of the same Amazon unit that makes Ring cameras and Kindles, and its terminals are designed to be smaller and cheaper than Starlink’s. But its biggest advantage might be Amazon Web Services’ vast network of data centers. For businesses and security-minded government organizations, “AWS means we can basically give these folks private networking capabilities,” a company spokesperson says, meaning they can move their data in secret, “without ever touching the public internet.”

But time isn’t on Kuiper’s side. A public beta test, originally scheduled for early 2024, has slipped to late this year or early next. The original terms of Kuiper’s license with the Federal Communications Commission require the company to launch 1,600 sats by the middle of next year. (Maybe Bezos’ newfound coziness with Team Trump will give Kuiper room to renegotiate.) Either way, Kuiper can’t count on an in-house partnership with a rocket builder like Starlink can. Only about one in eight of the launches by the Bezos-founded ecommerce company are contracted to fly on Bezos’ rockets. But the Blue Origin rocket in question has flown exactly once. SpaceX is apparently so unconcerned by the competition that it just carried Kuiper satellites on a mid-August Falcon 9 mission, and it has plans to do so again on two more.

Today, Starlink’s 8,000-plus satellites have a total bandwidth of 450 terabits per second. That’s comparable to about a third of the bandwidth used across the globe each year, as measured by the consultancy TeleGeography. The next iteration of Starlink satellites, so big they’re only capable of being hauled to space in a Starship, could each add vastly more bandwidth. Musk will have the satellite internet business in an absolute hammerlock. And as if that grip wasn’t tight enough, Musk is now looking for approval from multiple governments to add as many as 30,000 additional sats to his network.

Musk keeps talking about how his megarocket, Starship, is going to take humanity to Mars. But that’s not what the world’s first fully reusable multistage rocket is designed to do. Like all SpaceX rockets, it’s optimized to go to low Earth orbit—its relatively underpowered engines burn through nearly all of their fuel in just a few minutes. (That’s part of why the rockets are reusable. They don’t have too far to fall back to Earth.) On its own, Starship is designed to launch a whole bunch of next-gen Starlinks very quickly and very cheaply. Getting any further out means sending up more Starships—yes, plural—to refuel the first one with liquid methane, in what amounts to “an incredible conveyor belt,” as one aerospace industry executive put it.

The plan is ornate and over-the-top, different in so many ways from the scrappy old SpaceX days. The Starship relies on 39 Raptor engines, which are on their third major design in nine years. And the rocket is such a clunky way of getting to the moon or Mars that analysts like Lucas Pleney—a Musk admirer at Novaspace, a consultancy based outside of Paris—wonder whether Musk’s Red Planet dream has actually been back-burnered or could even be something of a distraction at this point. Starlink is the preeminent player in what appears to be the global communications infrastructure of the near future. If things continue in this direction, Musk won’t just have say-so over who gets connected and how much they’ll pay. He might even be able to access their data, too.

“This is really the elephant in the room. And Musk just points at Mars like, ‘This is my objective. Don’t look at Starlink,’” Pleney tells me. “This is why I’m thinking: Is he really pointing the finger toward Mars and believing in it? Or is he just trying to divert us from the big thing, which is Starlink and how much it will take over?”

III. SPACEWAR

Maybe you’re ok with the idea of Musk controlling the internet above the sky or deciding what can or cannot get off the planet. The Pentagon is under a different set of orders. “It is the policy of the United States,” reads Section 2273 of Title 10 of the US Code, to maintain “at least two space launch vehicles” that can take “national security payloads into space whenever such payloads are needed.”

In other words, it’s against the spirit of US law and policy for anyone to have a monopoly on military spaceflights. And so the Pentagon has purposely split up its gigantic “national security space launch” program. Blue Origin was awarded over $2 billion in these contracts, even though its big rocket has flown only that single test flight. Rocket Lab says it gets at least half of its revenue from defense and security agencies.

For now, Musk’s biggest competitor for military and intelligence missions is his next-door neighbor at Cape Canaveral. Just a mile from SpaceX’s launchpad is a 50,000-square-foot warehouse holding 21-story rockets, broken into their component stages. These rockets are rather different from Musk’s. The first stage is optimized to go about 100 miles up, two times higher than the Falcon 9’s core. The second stage is specifically designed to take a satellite another 20,000 miles. That’s the domain of the military’s most sensitive communications and spy satellites. It’s where the United Launch Alliance, the old monopolist, hopes to rule again. ULA’s idea is not just to give the government those God’s-eye views but also to help the Pentagon wage war in space.

“Defense of satellites in orbit, that was off the table through all of the Biden administration. You could not have what we call counterforce. You weren’t allowed to put a weapon in space, even to defend your satellite yourself,” Tory Bruno, ULA’s chief executive officer, tells me. “Under this administration,” he adds, “we are now allowed.”

The Pentagon is candid in its ambitions to try to destroy and disable Chinese and Russian spacecraft—and protect American ones from what they say are similar attacks. In March, the US military released a report titled “Space Warfighting.” It includes high-level plans for “orbital strike,” or “actions taken to destroy, disrupt, or degrade adversary space platforms.”

While the world’s militaries are increasingly eyeing low Earth orbit, ULA is openly pitching its high-flying second stage as a space warfare platform—hiding satellites in orbits where the Chinese can’t find them, maybe using the second stage itself to attack enemy spacecraft.

I spoke with several aerospace and defense industry insiders who tell me this might be feasible—but first, ULA has to start getting its rockets up consistently. That hasn’t been easy. Depending on whom you ask, the first operational launch of ULA’s new Vulcan rocket in mid-August was somewhere between two and five years behind schedule.

So while the military theoretically wants and needs ULA, Blue Origin, and these other SpaceX challengers to succeed, the reality looks quite different. Musk’s competitors say he has been receiving defense contracts that should’ve been theirs. SpaceX got seven of the nine national security space launches awarded in April, for example, totaling $846 million. Additional national security contracts that were supposed to be earmarked for emerging players wound up being sent SpaceX’s way instead.

Musk was poised to have even more on top of that, after spending immense capital to elect the president and then join his government.

In the first weeks and months of 2025, his friends at the Trump State Department tried to strong-arm nations like Gambia into buying Starlink. The Trump White House hit tiny Lesotho with crushing 50 percent tariffs; the country then quickly licensed Starlink as a way of demonstrating “goodwill and intent to welcome US businesses,” according to a State Department memo obtained by The Washington Post. Vietnam, Bangladesh, and India all quickly made similar deals after stalled negotiations. American diplomats promote American businesses all the time, but this was something different. “If this was done by another country, we absolutely would call this corruption,” Kristofer Harrison, a former State and Defense Department official in the George W. Bush administration, told ProPublica. “Because it is corruption.”

Jared Isaacman—who reportedly paid SpaceX $200 million for a private space trip—was initially lined up to run NASA. The president, in his inaugural address, seemed to send a love note to Musk by promising to “pursue our manifest destiny into the stars, launching American astronauts to plant the Stars and Stripes on the planet Mars.” In March, the Trump Commerce Department rewrote the rules for a $42 billion broadband program that make it harder for wired networks to get grants—and easier for satellite providers like Starlink to do so. (Musk is already pushing the states of Louisiana and Virginia to give him more money.) The Trump Pentagon, already tilting in Musk’s direction, leaned even heavier that way. It reportedly floated the idea of yanking funding for one of its more sensitive satellite communications networks and handing billions of dollars to a Musk-built constellation instead. That’s on top of the billions the Defense Department has committed to “Star-shield,” its private, military-grade version of Starlink.

But all of that is peanuts, potentially, compared to Trump’s pursuit of a “Golden Dome” over America, which would allegedly protect it from ballistic, hypersonic, and cruise missiles all at once. Trump picked the top general from the Space Force to lead it and promised that such a defense would include “space-based sensors and interceptors.” That would make it a reboot of the Reagan-era “Star Wars” missile defense boondoggle—but with better tech to detect targets, and AI to coordinate the interceptors. Some experts think that gives it a better chance of working this time around—as long as the US deploys thousands of weapons in space.

Here’s why: If an orbiting interceptor is going to shoot down a missile as it’s taking off, the interceptor needs to be relatively close to the ground. But that means any particular interceptor is only over a particular target for a few minutes. According to analysts like Todd Harrison of the American Enterprise Institute, “it takes about 950 interceptors spread out in orbit around the Earth to ensure that at least one is always in range to intercept a missile during its boost phase.” Every additional missile you want to stop means another 950 interceptors. China and Russia both have somewhere between 300 and 500 land-based ballistic missiles in their arsenals.

Even if Harrison’s off by a factor of 10, it would still be a monumentally expensive proposition, with a whole other fleet of satellites to coordinate the thing. Trump’s allies rammed $25 billion into this year’s budget as a down payment, even though there’s no formal concept for how such a thing might work. Trump says he’s willing to spend $175 billion in three years to get it done. Other defense leaders think the real cost could top half a trillion. And for a minute, it looked like Musk had much of it locked up. “I think that much of the commercial space industry and much of the defense space industry is worried that this could be an inside deal,” one executive at a top military contractor told me in May. “He’s got unlimited, deep pockets. He’s already proven he can put 7,000 satellites up. I think the administration thinks he’s the smartest man in the world. Our only hope is that the Department of Defense stands by their principles of fair and open competition.”

Musk denied he was interested in Golden Dome, posting on X that he’d rather focus on the Mars mission. But Reuters reported that Musk was not only part of a front-running team to build the anti-missile system but that “in an unusual twist, SpaceX has proposed setting up its role in Golden Dome as a ‘subscription service’ in which the government would pay for access to the technology rather than own the system outright.”

Of course, a subscription is something that can be turned off. Musk would have the kill switch to America’s orbiting weapons system. The man behind “MechaHilter.” The guy who gave the thumbs up to using his satellites as part of a campaign to overthrow a government and appeared to do very little as hundreds of thousands starved. That dude, with an outsize role in the making and maintaining of an arsenal in space—one in which the guns are pointed down at Earth.

The Trump-Musk alliance, of course, has since blown up, and along with it some of the most grandiose dreams and Bond-villain fears of Musk’s concentration of power in space. In early June, Musk briefly threatened to abandon the American astronauts aboard the ISS—SpaceX’s Dragon capsule was the US’s only way to bring them home—before thinking better of it. The president, for his part, threatened to pull Musk’s government contracts, posting that “the easiest way to save money in our Budget, Billions and Billions of Dollars, is to terminate Elon’s Governmental Subsidies and Contracts. I was always surprised that Biden didn’t do it!”

The Trump administration, according to The Wall Street Journal, did conduct a review of NASA and Department of Defense deals. But finding ways to remove Musk was functionally impossible. “Everybody in DOD was like, ‘No, no, no, no, no, no, let’s not do that.’ Because they’re so dependent on SpaceX now, and frankly, becoming increasingly dependent on Starlink,” one deeply connected industry observer says. In the end, Musk didn’t lose a dime in his existing federal contracts. SpaceX and Starlink were found to be too vital to US interests, too entrenched in our infrastructure even to appease Donald Trump.

Which returns us to the status quo. The space industry—domestic and international, commercial and military—has been thoroughly remade in Musk’s image. The US government and the Musk empire can no longer live without one another. They’re symbiotic. And the only one who can threaten that is Musk himself. “He’s a dickhead,” this source adds, “but his tech works, like, way better than anybody else’s.”