WHAT GOES up must come down. That was certainly true of dogecoin, a cryptocurrency enthusiastically promoted by Elon Musk. It surged before the weekend after Mr Musk’s latest tweet about it, but slumped after he jokingly called it “a hustle” on “Saturday Night Live” (SNL), a fixture of American TV comedy, which he hosted on May 9th. More gracefully, on May 5th a prototype version of SpaceX’s massive “Starship” rocket—designed to be the biggest since the Saturn V that took the Apollo astronauts to the moon—flew 10km above Boca Chica in Texas, before flying itself back to its launchpad and settling down gently on the ground. It was not Starship’s first high-altitude test flight. But it was the first that had ended without a fireball.

It was the latest piece of good news for SpaceX, a rocketry firm founded in 2002 by Mr Musk, who is perhaps better known as the founder of Tesla, an electric-car pioneer. Like Tesla, SpaceX has taken an unloved technology and made drastic improvements, shaking up a complacent industry. While Tesla’s mission—“accelerate the world’s transition to sustainable energy”—is grand, SpaceX’s is even grander. Mr Musk used his appearance on SNL to reaffirm his plans to use its cheap rockets to make humanity a “multi-planetary space-faring civilisation” by establishing a colony on Mars. And like Tesla, SpaceX’s valuation has soared. According to Pitchbook, a data-analysis firm, SpaceX’s latest funding round, completed in April, valued it at $74bn, up from $44bn in August 2020. CB Insights, a firm of analysts, ranks SpaceX the third most valuable start-up in the world (see chart).

It may seem odd to describe a 19-year-old firm as a “start-up”. But most of SpaceX’s swelling valuation comes not from the business it already does but, again like Tesla, its investors’ hopes for its future. To pay for its Martian ambitions, SpaceX plans to transform itself into a globe-straddling telecoms giant. It hopes to repeat Mr Musk’s signature trick of making big improvements to existing technologies. Its “Starlink” service, currently open to testers in countries including America, Britain and Germany, is building the biggest satellite network ever, in order to beam fast internet access to every corner of the planet.

Big rocket man

SpaceX’s advances in rocketry provide the launch pad. Its craft are unusual in that they are reusable, rather than disposable. After launch, the first stage of its Falcon 9 can fly itself back to Earth under its own power and after a refurbishment lasting a few weeks it can fly again. Along with a focus on cost-cutting and a willingness to experiment and take risks, that has allowed SpaceX to undercut its competitors drastically.

As with Tesla, complacent incumbents have been trying to respond. United Launch Alliance, a joint venture between Boeing and Lockheed Martin, two aerospace giants, has cut jobs and trimmed costs. In November Tory Bruno, its boss, said prices for its Atlas V rocket were down from $225m per launch to just over $100m. ArianeSpace, a European firm, has also cut prices for its Ariane 5, which is thought to cost around €175m ($213m) per flight. It hopes the Ariane 6, due to make its first flight next year, will be 40% cheaper than its predecessor. SpaceX charges $62m for a fresh rocket, or $50m for a used one.

Low prices, a focus on cost control, and a willingness to take risks and iterate rapidly (another signature Musk trait) have helped SpaceX win contracts with everyone from Iridium and Intelsat, established satellite firms, to startups such as Planet and governments, including those of America, Germany and South Korea. On April 16th NASA awarded SpaceX $2.9bn to develop a lunar lander as part of America’s plan to return astronauts to the moon by 2024 (though the contract was suspended on April 30th; a government agency is reviewing rival firms’ complaints). On September 15th it plans to fly four tourists on a three-day orbital jolly. Morgan Stanley, a bank, describes SpaceX as “mission control” for the fast-growing “emerging space” sector—which, estimates Seraphim Capital, a venture-capital company, attracted $8.7bn of venture investment in the year to March, up by 95% from the year before.

And it is not standing still. Starship has a carrying capacity more than six times that of the Falcon 9. Despite its vast size, it is designed to be fully reusable, and is intended to be far cheaper than SpaceX’s current rockets. Mr Musk hopes Starship could end up costing less than $2m per launch.

But however nifty SpaceX’s technology gets, the launch market, at around $6bn in 2019, is relatively small, says Simon Potter of BryceTech, a firm of analysts and engineers. Many players are shielded from full competition by governments worried about national security. That will limit SpaceX’s market share. Instead, says Adam Jonas, an analyst at Morgan Stanley, SpaceX sees launch as an “enabling technology” for its other plans. The firm’s next target is the telecoms business. “Starlink” aims to provide internet access worldwide, including places where other forms of connectivity are poor or non-existent.

This is a much bigger market, at least on paper. The International Telecommunications Union, a UN agency, reckons 48% of the world’s population was offline in 2019. Gwynne Shotwell, SpaceX’s chief operating officer, said in 2019 that the worldwide internet-access market was worth perhaps $1trn a year. SpaceX, Mr Musk has said, might aim to capture around 3% of that. Even that sliver would have brought in $30bn two years ago.

Satellite internet is not a new idea. But it is another technology that Mr Musk thinks he can improve. Existing internet satellites fly at high altitude, to maximise coverage. The drawback is that many customers must share a single satellite, limiting capacity. And the time taken for radio signals to travel to high-flying satellites adds unavoidable, and irritating, delays. At the moment satellite internet is usually a last-resort option when nothing better is available—in remote rural areas or ships at sea, for instance.

Tomorrow the stars
Starlink hopes to fix those problems by using its cheap rockets to put thousands of small, cheap satellites in low orbits. In the first quarter of 2021, SpaceX launched more objects, measured by mass, into orbit than every other rocket operator combined, says Mr Potter. Starlink’s 1,500-odd existing satellites already account for around a quarter of all those in orbit. SpaceX has firm plans for over 10,000 more, and has filed paperwork for up to 42,000—more than four times as many satellites as have been launched since the start of the space age.

The prototype service is undergoing testing by thousands of people. Most seem pleased, reporting fast and responsive connections. But the satellite-internet business has a poor record. Iridium went bankrupt in 1999, the year after its launch (it was eventually bailed out by the American government). Intelsat and Speedcast, two established companies, filed for bankruptcy last year, as did OneWeb, a start-up with a similar business model to Starlink’s. Intelsat is currently restructuring and SpeedCast is doing business again under new owners. But the fragility of the business makes assigning a future value to SpaceX tricky. Morgan Stanley’s attempt spans two orders of magnitude, from $5bn to $200bn, with different assumptions about the viability of Starlink accounting for almost all of the difference.

Even with low launch costs, at least two big challenges remain, says Rasmus Flytkjaer of London Economics, a consultancy. One is that most of Starlink’s potential customers are people ill-served by terrestrial internet firms. They tend to live in relatively poor rural areas. Starlink’s price of $99 per month is not cheap even for rich-country users. The other is the cost of the high-tech satellite dishes needed to make the system work: 23-inch antennas that attach to roofs or walls. Since Starlink’s satellites are in low orbits, they zip quickly across the sky. The aerials must be able to track satellites as they move, and switch seamlessly from one to the next as they disappear below the horizon.

Ms Shotwell said in April that the dishes, which SpaceX sells for $499, cost around $1,500 to produce, down from about $3,000 two years ago. SpaceX hopes that economies of scale will eventually drive manufacturing costs down to “a few hundred dollars”. Part of Iridium’s problem, says Mr Flytkjaer, was meeting the capital cost of building up its network before it could attract paying customers. Mr Musk’s deep pockets, he says, should mean SpaceX is less likely to run out of cash than its predecessor two decades ago.

Such challenges may explain Mr Musk’s uncharacteristic lack of bombast when talking about Starlink. Tesla sells cars with features like “Ludicrous Mode” and “Bioweapon Defense Mode”. Starlink, by contrast, calls its public-test programme the “Better Than Nothing Beta Test”. At a space conference last year Mr Musk said Starlink’s goal, for now, was simply not to go bankrupt. He has repeatedly tried to assure existing telecoms firms that Starlink is not a threat, pointing out that the service is ill-suited to serving large numbers of customers in densely populated cities.

Starlink’s test programme is currently available in only a handful of rich countries. Yet the firm said on May 5th that it had collected half a million pre-orders. It has requested regulatory permission for up to 5m users in America alone. In December SpaceX won $886m from America’s government to provide broadband in rural areas; it is reportedly in similar talks in Britain. Not all governments will be as accommodating, since the internet access offered by Starlink could prove tricky for the authorities to censor.

In poorer countries, says Mr Flytkjaer, Starlink’s satellites could be used to connect rural mobile-phone masts to the internet, spreading the connection cost among many users. SpaceX is running tests with America’s armed forces, which are keen on the idea of having internet connectivity on any battlefield. In 2019 the firm demonstrated its ability to provide high-speed, in-flight internet to a military jet.

Despite its unpromising history, Mr Musk is not the only billionaire who thinks satellite internet is an idea whose time has come. After its bankruptcy OneWeb was rescued by the British government and Bharti Enterprises, an Indian conglomerate whose founder, Sunil Mittal, is one of India’s wealthiest men. Jeff Bezos, Amazon’s founder, is every bit as rich as Mr Musk—and just as much of a space cadet, bankrolling Blue Origin, his own private rocket firm. Amazon itself is planning a low-flying satellite-internet similar to Starlink, called Kuiper. The car industry increasingly dances to Mr Musk’s tune. The space industry is going the same way.