The effort to return booster rockets to Earth had been tried and had failed several times; it turns out landing a rocket back on Earth safely is pretty difficult. So Elon Musk was not deluding himself in 2014 when he calculated the odds that his company, SpaceX, would eventually get it right: “not great — perhaps 50 percent, at best.”

And then they did. Just before Christmas 2015, a Falcon 9 booster became the first rocket to deliver a payload to orbit, reorient itself, fly back through the atmosphere, find its landing spot — in that case, a pad on the coast at Cape Canaveral — and touch down softly.

Since then, SpaceX has done it again and again, so many times that Musk has achieved his goal, normalizing a feat once thought impossible. Now SpaceX is on the verge of its 50th landing in a launch now scheduled for Sunday, a milestone celebrated within the company and the larger space industry, which has come to agree that ditching rocket boosters into the ocean — the practice for decades — is an expensive waste of a perfectly good vehicle.

“We got there much faster than I ever thought we would,” said Garrett Reisman, a former NASA astronaut who worked at SpaceX for years and now serves as a consultant. “Just over four years — that is really remarkable, the fact that it has become routine in four years. It’s still not routine to me. I get excited. I still get goose bumps.”

For years, rocket boosters propelled their payloads to space, then separated and fell back to Earth, splashing down into the ocean. To Musk and others who pursued the dream of landing the rockets as a way to make space more accessible, that was like throwing away the airplane after a trip from New York to Los Angeles.

“For us to really open up access to space, we have to have full and rapid reusability,” Musk has said.

Precisely how much money SpaceX saves is hard to say. As a privately held company, it doesn’t release precise dollar amounts. But it sells each Falcon 9 launch for about $62 million, and the overwhelming cost of each launch is in the booster, which houses nine engines. The propellant, Musk has said, is a fraction of the cost, about $200,000 or so.

The space shuttle was reusable as well, but it took an army of workers to get it ready for the next flight, and it never achieved the kind of efficiency initially envisioned. “The key isn’t just to make it reusable,” Reisman said. “If you have to completely rebuild it every time, your economic advantage is going to erode. I can tell you achieving that economically affordable reusability is key, and we’ve done that.”

Jeff Bezos agrees. His space company, Blue Origin, has landed a series of New Shepard boosters as well — but those go up to scratch the very edge of space before falling back down on trips that don’t orbit the Earth. It plans to recover its more powerful New Glenn rocket, however, in much the same way SpaceX does. (Bezos also owns The Washington Post.)

The United Launch Alliance, the joint venture of Boeing and Lockheed Martin, has talked about recovering not the entire booster, but popping out the most valuable part — the engines — and catching them with a grappling hook as they fall back to Earth under a parachute. And Rocket Lab had initially said it wouldn’t try to reuse its small rockets but has since reversed course.

“SpaceX’s technical validation of reusable rockets has opened new horizons for the launch sector while inciting the firm’s competitors to invest in technological innovation as a means to fight market share erosion,” Maxime Puteaux and Alexandre Najjar, consultants at Euroconsult, wrote in a recent SpaceNews op-ed.

SpaceX has also gotten better at refurbishing the rockets faster between flights. While the first booster took a year to relaunch, SpaceX flew one last year after just 82 days, and the rocket it plans to fly on Sunday’s mission had a still faster turnaround — only about 60 days between its last launch and its next one, the company said.

SpaceX doesn’t just land on land. It lands on a robot ship at sea it calls an “autonomous spaceport drone ship.”

“It’s quite a tiny target. It’s like trying to land on a postage stamp there,” Musk said in 2016. “It’s like a carrier landing versus a land landing.”

Bringing the rocket home is “supremely difficult,” Reisman said. The rocket and the drone ship essentially work to meet up in the exact same spot in the ocean, and the booster has to be constantly decelerating and get as close to zero velocity as it can at the precise moment it is touching down. It also needs to be able to survive high temperatures as it screams back through the atmosphere.

For a few years, SpaceX struggled as one booster after the other crashed and burned. It happened so frequently the company came up with a term for it — “rapid unscheduled disassembly.”

When SpaceX finally pulled off a drone ship landing in 2016, Canadian astronaut Chris Hadfield and Buzz Aldrin, the second man to walk on the moon, tweeted their congratulations. “Opens the imagination to what is possible,” Hadfield wrote.

Then-President Obama also weighed in, writing, “It’s because of innovators like you & NASA that America continues to lead in space exploration.”

President Trump has been enamored of the landings as well. Musk “does good at rockets, too, by the way,” he told CNBC recently. “I never saw where the engines come down with no wings, no anything, and they’re landing. I said, ‘I’ve never seen that before.’ ”

SpaceX now has not just one landing pad at Cape Canaveral, but two — one for each of the side boosters that fly on its Falcon Heavy rocket, which come down in tandem.

Having mastered the art of recovering boosters, SpaceX is now going after another part of the rocket — the nose cone, or fairing, which sits atop the rocket and protects the satellite being launched. The company has a couple of ships with giant nets affixed to them that try to position themselves under the fairing as it falls from space under a parachute. So far, it has caught three.

In 2017, Musk said the fairings cost about $6 million each — not a lot in the context of a rocket, but not chump change, either.

“At one point, we’re, like, debating, ‘Should we try to recover it or not?’ ” Musk said. “It’s like, ‘Guys, imagine you had $6 million in cash in a palette flying through the air, and it’s going to smash in the ocean. Would you try to recover that?’ Yes. Yes, you would.”