A new, high-flying kind of drone recently flew for 25 days straight over Arizona starting July 11, setting a world record for aircraft endurance while soaring as high as 70,000 feet over the desert.

But the solar-powered Zephyr, under development by European plane-maker Airbus, isn’t really an aircraft in the conventional sense.

Instead, the flimsy-looking vehicle with the 82-foot wingspan and two tiny propellers is what the aerospace industry calls a “high-altitude pseudo-satellite”—a plane that flies so high for so long that in many ways it’s more like an orbital spacecraft.

The European Space Agency called drones such as Zephyr “a missing link between drones and satellites.”

Zephyr and similar vehicles also in development could do satellite jobs at drone prices. And that’s a really big deal. Inexpensive drones flying nonstop for weeks, months, or even years at a time could transform science, communications, and military surveillance.

But Zephyr and its aerial kin face stiff competition from tiny “microsatellites” that are getting cheaper and easier to deploy. A three-pound “cubesat” satellite costs a few hundred thousand dollars to build and launch—and can last a year or longer in low orbit a few hundred miles above Earth.

Clutches of cubesats, deployed by the dozen by a single rocket, can complement or even replace traditional large satellites that can cost millions or even billions of dollars to build and launch.

By contrast, building a single Zephyr costs roughly $6 million, but the drone can fly back-to-back missions for many years at a very low day-to-day cost.

“The future of [high-altitude pseudo-satellites] will be driven mainly by the evolution of technologies of potential competitors, such as microsatellites constellations,” Flavio Araripe d’Oliveira, Francisco Cristovão Lourenço de Melo, and Tessaleno Campos Devezas wrote in a 2016 study.

Researchers had been talking about pseudo-satellites since the 1970s, but it wasn’t until the 1990s that the technology—smaller electronics, lighter airframes, and more reliable control systems—caught up with the concept.

Zephyr began as a British government research project back in 2001. Soon thereafter, the U.S. wars in Afghanistan and Iraq opened a floodgate of interest in, and funding for, high-flying drones that could surveil for insurgents for longer than the roughly 24-hour endurance of a comparatively low-flying Predator drone.

Pseudo-satellite prototypes cropped up all over the world in the form of winged drones and enormous airships. Several crashed, including Google’s Solara, NASA’s Helios, and a U.S. Army spy blimp. With the partial withdrawal of U.S. troops from Iraq and Afghanistan, funding dried up and many of the pseudo-satellite programs ended.

But Zephyr endured and, in 2010, set a endurance record of 14 days during testing over Arizona.

In 2013, Airbus bought the Zephyr design from the British government. The government returned the favor three years later, purchasing three copies of the high-flying drone for trials.  “Zephyr is a cutting-edge, record-breaking piece of kit that will be capable of gathering constant, reliable information over vast geographical areas at a much greater level of detail than ever before,” then-U.K. Defense Secretary Michael Fallon said.

Airbus continued refining the drone, pushing its altitude and endurance higher and higher. The 25-day flight in Arizona was a triumph for Airbus’ engineers, who had spent years balancing Zephyr’s payload and power-consumption to boost its efficiency in the air. “You have to find the right equation between flying altitude plus battery life, maintaining this or that power,” Alain Dupiech, an Airbus spokesperson, told The Daily Beast.

Hand-launched by a team of three during daytime when its solar panels are most effective, the 165-pound Zephyr soars to a maximum altitude of around 70,000 feet, where wind resistance is low and conditions are perfect for a pseudo-satellite’s leisurely flight profile and 20-miles-per-hour top speed.

During the day, Zephyr flies high, drawing power from the sun and charging its lithium-ion batteries. After dusk, the Zephyr begins draining its batteries to keep its propellers spinning and slowly loses altitude.

In its record-setting July-to-August flight, Zephyr bounced from 70,000 feet during daytime down to 50,000 feet at night and then back up. Eventually its battery wore out, much like an iPhone battery ultimately dies after being recharged too many times.

It’s Zephyr’s battle with the sun that really defines its development. A pseudo-satellite tasked with gathering weather data, relaying communications signals, or spying on enemy forces needs to fly high to be effective. “The real question is not endurance in terms of days, it’s endurance in terms of altitude,” Dupiech explained.

Zephyr’s record-setting, 25-day flight makes for good headlines, but endurance records aren’t important to Airbus, Dupiech said. “This is not interesting.”

“We’re trying to provide the world with something it doesn’t have: pertinent persistence,” Dupiech added. In other words, a drone that can fly high for as long as the user needs it to do in order to perform a particular mission, and at reasonable cost. Say, monitoring an unfolding natural disaster, temporarily filling in gaps in communications coverage, or keeping watch over a battlefield.

That could mean weeks or months of nonstop flying. “At this stage, most of those missions are not calling for a year and half up there,” Dupiech said. That means Zephyr might not need to directly compete with a cubesat constellation that lasts a year or longer.

At present the British military is Zephyr’s only paying customer. But with every successful test flight, the drone’s capabilities expand and its prospects likely improve. One prominent potential buyer is Facebook. The social-media company wants to use high-altitude pseudo-satellites to extend internet coverage to regions of the world that currently have poor connectivity.

But Facebook’s own Aquila communications drone performed poorly in tests. In June the company shut down the Aquila program, creating an opening for Zephyr to take its place. “Going forward, we’ll continue to work with partners like Airbus,” wrote Yael Maguire, Facebook’s director of engineering.

Zephyr’s next test flight is scheduled for October over western Australia.