Elon Musk, as a businessman, is primarily known for three of his many enterprises.
One is X, formerly Twitter, nowadays judged by investment giant Fidelity to be worth perhaps a quarter of what it cost him. Another is electric car maker Tesla, currently valued at almost five times the price of the next biggest car company. Neither X nor Tesla does anything that other companies can't do.
But then there's SpaceX and its subsidiary, Starlink. Founded 22 years ago, SpaceX has totally transformed the space launch industry. It makes more space launches than any other organisation and operates the 5,000-tonne Starship, the biggest and most powerful flying vehicle ever made by the human race.
SpaceX, however, is not a normal company primarily in business to make a profit. Musk's goal in establishing it was to build spacecraft capable of carrying human beings to Mars, as a necessary step towards human colonisation of the red planet.
"One path is we stay on Earth forever, and then there will be some eventual extinction event," Musk said in 2016. "The alternative is to become a multi-planetary species."
His ambition may have moved a step closer to realisation with the election of Donald Trump, a candidate who Musk helped bankroll and who has now put him in charge of slashing red tape and government spending.
Mars is the choice because it is probably the most hospitable place in the solar system other than Earth, and it is also - in astronomical terms - not far away at certain points in the orbital cycles of the two planets.
Its atmosphere is mostly carbon dioxide and, though it's further from the Sun, it still gets useful amounts of sunlight. Plants would grow there in the local atmosphere, though it would need to be compressed and warmed up.
Musk knew all this back in the early Noughties, and was concerned to find that Nasa had no firm plans at the time for going to Mars. Wealthy from the founding of PayPal, Musk decided to start off by sending a small test greenhouse to the red planet.
But he found that getting a rocket for the mission was difficult. "I went to Russia three times to try to buy a couple of their biggest ICBMs [intercontinental ballistic missiles]," Musk has said. "It was an interesting experience."
At the time, the American launch business was largely sewn up by two monopolistic giants: United Space Alliance, which operated the Space Shuttle for Nasa, and United Launch Alliance (ULA), which handled everything else.
Both companies were alliances between US aerospace mammoths Lockheed and Boeing, and it seemed obvious that nobody could compete against that. If you wanted to put something into space, and you didn't want to go to the French or the Russians, you did business with ULA.
Musk decided instead to set up SpaceX and build his own rockets. At that point most American rockets were fuelled by liquid hydrogen, burned with liquid oxygen. Hydrogen is theoretically the best rocket fuel there is, but it's very difficult to handle safely. Operating hydrogen-fuelled rockets is difficult, dangerous and very expensive.
SpaceX has never used hydrogen fuel, in line with its ambition to bring down the cost of space launch. Its initial Merlin rocket engine, still in service, uses highly refined rocket-grade kerosene, or RP-1. This is cheap to work with as it's a liquid at normal temperatures.
Musk and SpaceX, having worked out what they were doing with the initial Falcon 1 single-engined test rocket, put their new Merlin engine to work in the Falcon 9 - so called as its first stage has nine engines.
Almost straight away, the Falcon 9 was a big success, slashing costs with its easy-to-handle fuel and modern design. Its competitors were all developments of old technology, sometimes very old indeed: ULA's Atlas launcher family is descended from a missile programme of the 1940s.
But there was more to come from SpaceX. Ever since big rockets became a serious thing towards the end of the Second World War, the operating model had been the same. Rockets were single-use, expendable items. They flew once and were destroyed. If you wanted to fly again, you had to build another rocket.
The big exception to this rule, of course, was the Space Shuttle. By the time SpaceX was in business, however, it was clear that the Shuttle was not going to bring down the cost of space launch. Yes, you got the main engines back after using it: but you didn't get the fuel tank or the strap-on boosters. And hydrogen-fuelled operations still cost huge amounts of money.
SpaceX was determined to do better. The Falcon 9 was already impressively affordable, but now Musk and his engineers were keen to make it re-usable.
They started fitting the first stage - the biggest and far the most expensive section of the rocket responsible for generating initial acceleration - with folding landing legs and control systems, which would let it fly back down to land vertically after releasing from the second stage, the part of the vessel that continues into orbit.
SpaceX first accomplished a successful landing on only the Falcon 9's 20th flight, in 2015. Today the Falcon 9's record stands at 397 total launches, making it the most launched American orbital rocket in history. Some 353 successful first-stage landings have been carried out, and 328 of the launches used a first stage that had been up at least once before. Some first stages have been up as many as 23 times.
Falcon 9 does more than just launch satellites. SpaceX built Dragon capsules for it to carry, and soon Musk was in the business of taking astronauts and crew to and from the International Space Station for Nasa, replacing the retired Shuttle.
But the Shuttle also had another purpose apart from manned space exploration for Nasa. From the first, it had been intended to support the other, bigger US government space programme - the various top secret, "black" spy satellite projects operating through the National Reconnaissance Office. The size of the black space budget is a secret, of course, but it's generally believed to be larger than that of Nasa.
Having managed to grab a big share of the commercial satellite market with the re-usable Falcon 9, and a slice of Nasa's manned low-orbit funding with Dragon, Musk was now ready to make a move on the black budget.
SpaceX opened offices in Chantilly, Virginia - the location of NRO headquarters - the year of the Shuttle's final flight. The US intelligence community and secret space military were evidently receptive, because SpaceX was soon at work on its next rocket.
Musk's engineers bolted together three Falcon 9 first stages in a row beneath a single second stage, creating the Falcon Heavy.
This first flew in 2018, becoming the most powerful rocket then flying in the world. The test flight launched a Tesla Roadster owned by Musk into an orbit round the Sun which will see it further out than Mars at times.
The only rockets that had ever been more powerful were the mighty Saturn Vs which took men to the moon and the Soviet Energia superlifter of the 1980s. Falcon Heavy has now flown 11 times, carrying a mixture of large commercial satellites, science missions and secret spy payloads.
While all this was happening there had been other changes in the US space sector. The end of the Shuttle era had freed up potentially large amounts of budget for manned space exploration. A new, superheavy rocket was going to be required.
In the early 2010s, Falcon 9 was already flying while Falcon Heavy - set to be the most powerful rocket in the world - was on the drawing boards, and it was noticeable that President Obama had been talking to Musk.
In the end, however, the upstart SpaceX didn't manage to take the prize.
Nasa, under firm political direction from Capitol Hill, instead began work on its own Space Launch System superheavy rocket. The SLS is, in essence, a Space Shuttle that you throw away after use, in traditional rocket industry style.
Thirteen years after the project began, it has made one uncrewed test flight. The large industrial base that sustained the Shuttle remains in charge of that large stream of US taxpayer money.
Musk remained determined to get to Mars. Unfortunately for him, his share of commercial space, the black budget and Nasa weren't going to provide enough money for that.
So Musk decided to make another market - and thus Starlink was born. SpaceX went into the business of providing low-orbit satellite communications.
"No one has ever succeeded," in low-orbit satcoms Musk admitted in 2019 as his first satellites went up. And it wasn't because others hadn't tried, especially back in the 90s and Noughties. Iridium, Globalstar and Orbcomm remain operational following bankruptcy. Teledesic, Skybridge, Celestri and Astrolink are gone.
But Musk was undaunted. "If something's important enough, you should try," he recently told an interviewer, "even if the probable outcome is failure."
There was a reason people had made so many attempts to set up low-orbit satellite communications. The standard satcoms method today is a geostationary satellite. The satellite sits still in the sky so that one can point a fixed dish at it and receive high bandwidth transmissions.
That works well for TV, and that's why many businesses have paid the large extra cost of getting satellites all the way up to geostationary orbit 36,000 miles up. Musk has made plenty of money in that market.
Geostationary satellites have their problems, however. Being so far out, there is a serious latency problem, which means they aren't much use for video conferencing or gaming. They're also sharply limited in the number of two-way - as opposed to broadcast - customers they can serve.
This is why, today, almost all internet and two-way data passes through undersea fibre cables, not via satellite.
Low-orbiting satellites could be different. This is Musk's new bet, and not only his. A satellite just a hundred or two miles up doesn't have a latency problem. Yes, it zooms across the sky rather than sitting still, but nowadays you don't need to point a dish at something to achieve a good connection.
There are now thousands of Starlink satellites in the sky and more all the time. Those satellites are up there because Musk has a space launch company with spare capacity. It can easily and cheaply do more launching than anyone else needs it to do.
In Starlink, Musk has created the new, big customer that he didn't have. It's now generally believed among business analysts that Starlink is a major part of SpaceX's revenue today, rather than launching other people's equipment.
Certainly money is coming in from somewhere. It's hard to be sure of the details as SpaceX is private, and Musk - as the world's richest man - can command huge volumes of potential investment backed by his Tesla holdings and various other things.
The huge Starlink revenue needed to sustain SpaceX may or may not be real at this point, but Starlink's potential has clearly been enough to keep SpaceX's finances secure.
What's for sure is that Musk did not stand still following the Nasa decision to build its own SLS super lifter.
SpaceX has moved on to a grander scale of operations. There is a new and larger "Raptor" rocket engine, this time running on methane fuel.
Methane - natural gas, as we know it - isn't liquid at room temperature like RP-1, but liquefying it and handling it is easy and cheap compared to the horrors of dealing with hydrogen. It also burns much more cleanly than RP-1, which is important when you want to use your engines again and again.
But there's something else important about methane, too. Here on Earth we normally get hold of it by refining natural gas. But it can also be made out of carbon dioxide and hydrogen. If you were on Mars and you had a supply of energy - as from a small nuclear reactor or a large solar farm, perhaps - you could make methane and oxygen out of water and the local atmosphere, and refuel a Raptor-engined rocket.
Meanwhile, what Musk and SpaceX have done with the Raptor engine is truly impressive. This is Starship, the biggest flying thing ever made. It comes as a two-part stack like Falcon 9: the Super Heavy booster stage, with 33 Raptor engines, and the Starship spacecraft itself with three in-atmosphere Raptors and three for use in space.
Unlike the Falcon 9 upper stage, Starship is fully re-usable: it can be equipped with heat shielding for atmospheric re-entry. In some applications it might carry as many as a hundred people.
There's another new trick with Starship, too. The booster doesn't need to have heavy landing legs any more: it can fly down to the same gantry it launched from (or another one like it) there to be received by support arms in the so-called "chopstick" manoeuvre. Then it can have another upper stage placed on top of it ready for another launch.
SpaceX has only demonstrated a chopstick gantry recovery once so far, but it is an amazing thing to witness. It certainly impressed Donald Trump when he saw the video.
"I called Elon," said Trump afterwards. "I said, 'Elon, was that you?' He said, 'Yes, it was.' I said, 'Who else can do that? Can Russia do it? 'No.' Can China do it?' 'No.' Can the United States do it, other than you?' 'No, nobody can do that.' I said, 'That's why I love you, Elon, that's great.'"
Trump, as president-elect, travelled to Texas to watch the most recent Starship test alongside Musk: unfortunately the planned booster recovery wasn't repeated. But right back to Falcon 1, SpaceX has seen early test failures, learned from them and gone on to succeed.
"He's a special guy," says Trump of Musk. "He's a super genius. We have to protect our geniuses. We don't have that many of them."
Starships will be able to do a lot of things, but one of the most important missions will be refuelling another Starship in space. Musk has estimated that as few as eight launches might be needed to fully refuel a Starship spacecraft in Earth orbit. And a fully fuelled Starship in Earth orbit, equipped with landing legs, could make a trip to Mars if it had a heat shield as well.
The Starship project was at one time known within SpaceX as the Mars Colonial Transporter, then later as the Interplanetary Transport System - signifying even bolder ambitions.
All this rapid progress has left Nasa and its Space Launch System (SLS) looking, frankly, rather lame. For now SLS lives on, but even Nasa has been compelled to acknowledge the potential of Starship.
The current "Artemis" programme plan for a return to the moon is that Nasa astronauts will lift off atop an SLS, and travel to the moon in the Orion spacecraft.
Meanwhile, SpaceX will launch a Starship configured as a moon lander and get that to lunar orbit, refuelling it above Earth using other Starships. The astronauts will then transfer to the Starship lander and go down to the lunar surface for their visit, before taking off again in the Starship and returning to Earth in the Orion.
But it's getting harder and harder to see the point of having the SLS and Orion involved at all. Why pay huge sums to throw expensive hydrogen rockets into the sea, when you could re-use affordable methane ones and get so much more done, so much faster?
You might almost say that SLS is an example of government wastefulness, something Musk is intensely interested in. It's not exactly clear what the scope of Musk's new government job - as head of the Department Of Government Efficiency - may be, but eliminating government waste and regulatory obstacles to getting things done is certainly meant to be part of it.
At the moment, SpaceX is involved in various massive bureaucratic struggles with the Federal Aviation Administration and environmental authorities. Musk wants to make 25 Starship launches next year: at the moment the regulators want to limit him to five.
Various commentators have suggested that Musk has jumped aboard the Trump bandwagon - as recently as two years ago the two men were trading public insults - so as to get more government contracts for SpaceX. But Musk says all he wants is the regulators off his back.
"There is no quid pro quo," he posted on X in September. "With a Trump administration, we can execute major government reform, remove bureaucratic paperwork that is smothering the country and unlock a new age of prosperity."
Even so, it might just be that despite its powerful support on Capitol Hill, the SLS doesn't have much of a future. If its budget, or a large portion of it, gets diverted to SpaceX, we might see boots on Mars a lot sooner than we think - and perhaps even Musk's dreams of a self-sustaining Martian colony could come true in his lifetime, though that still seems like a big ask.
One can look askance at Musk's handling of Twitter/X as a business and his campaigning for Trump - though many are now reassessing that as part of a coldly rational, cost-effective plan to clear away obstacles ahead of SpaceX. One can doubt that Tesla really is worth far more than any other car company. One can criticise Elon Musk on many grounds for many things.
But any honest observer, looking at SpaceX and the way it is revolutionising access to space after decades of stagnation, would have to admit that Musk is genuinely changing the direction of human history.
It's already an astonishing legacy, and there's probably more to come.