Stratospheric internet could finally start taking off this year

Welcome to MIT Technology Review Narrated. My name is Matt Honan. I'm our Editor-in-Chief. Every week, we'll bring you a fascinating, new, in-depth story from the leading edge of science and technology, covering topics like AI, biotech, climate, energy, robotics, and more. Here's this week's story. I hope you enjoy it. Transcription by CastingWords for internet delivery. Even with nearly 10,000 active Starlink satellites in orbit and the one web constellation of 650 satellites, solid internet coverage is not a given across vast swathes of the planet. One of the most prominent efforts to plug the connectivity gap was Google X's Loon project. Launched in 2011, it aimed to deliver access using high-altitude balloons stationed above predetermined spots on Earth. But the project faced literal headwinds. The loons kept drifting away, and new ones had to be released constantly, making the venture economically unfeasible. Although Google shuttered the high-profile loon in 2021, work on other kinds of high-altitude platform stations, HAPS has continued behind the scenes. Now, several companies claim they have solved loons' problems with different designs, in particular steerable airships and fixed-wing UAVs, unmanned aerial vehicles, and are getting ready to prove the tech's internet beaming potential starting in 2026, in tests above Japan and Indonesia. Regulators, too, seem to be thinking seriously about HAPs. In mid-December 2025, for example, the US Federal Aviation Administration released a 50-page document outlining how large numbers of HAPs could be integrated into American airspace. According to the U.S. Census Bureau's 2024 American Community Survey, ACS data, some 8 million U.S. households, 4.5% of the population, still live completely offline, and HAPs proponents think the technology might get them connected more cheaply than alternatives. Despite the optimism of the companies involved, though, some analysts remain cautious. The HAPS market has been really slow and challenging to develop, says Dallas Kasabowski, a space industry analyst at the consultancy analysis Mason. After all, Kasabowski says, the approach has struggled before. A few companies were very interested in it, very ambitious about it, and then it just didn't happen. Hovering in the thin air at altitudes above 12 miles, HAPS have a unique vantage point to beam down low-latency, high-speed connectivity directly to smartphone users in places too remote and too sparsely populated to justify the cost of laying fibre cables or building ground cellular base stations Mobile network operators have some commitment to provide coverage, but they frequently prefer to pay a fine than cover these remote areas, says Pierre-Antoine Auberg, chief technology officer of Alto HAPS, a spin-off from the European aerospace manufacturer Airbus. With HAPS, we make this remote connectivity case profitable. Alto Haps has built a solar-powered UAV with a 25-metre wingspan that has conducted many long-duration test flights in recent years. In April 2025, the craft, called Zephyr, broke a Haps record by staying afloat for 67 consecutive days. The first months of 2026 will be busy for the company, according to Oboe. Zephyr will do a test run over southern Japan to trial connectivity delivery to residents of some of the country's smallest and most poorly connected inhabited islands. Because of its unique geography, Japan is a perfect testbed for HAPS. Many of the country's roughly 430 inhabited islands are remote, mountainous, and sparsely populated, making them too costly to connect with terrestrial cell towers. Alto HAPS is partnering with Japan's largest mobile network operators, NTT, Docomo, and the telecom satellite operator Space Compass, which want to use Zephyr as part of next generation telecommunication infrastructure. Non-terrestrial networks have the potential to transform Japan's communications ecosystem, addressing access to connectivity in hard-to-reach areas, while supporting our country's response to emergencies, Shugahiro Hori, co-CEO of Space Compass said in a statement. Zephyr, Oberg explains, will function like another cell tower in the NTT-Docomo network, only it will be located well above the planet instead of on its surface. It will beam high-speed 5G connectivity to smartphone users without the need for the specialized terminals that are usually required to receive satellite internet. For the user on the ground, there is no difference when they switch from the terrestrial network to the HAPS network, Oberg says. It's exactly the same frequency and the same network. New Mexico-based Sky, which has developed a solar-powered helium-filled airship, is also eyeing Japan for pre-commercial trials of its stratospheric connectivity service in 2026. The firm, which extensively tested its slick 65-metre-long vehicle in 2025, is working with the Japanese telecommunications giant SoftBank. Just like NTT Docomo, SoftBank is betting on HAPS to take its networks to another level. Mikhail Franson, Sky's founder and CEO, says that his firm succeeded where Loon failed by betting on the advantages offered by the more controllable airship shape, intelligent avionics, and innovative batteries that can power an electric fan to keep the aircraft in place. Google's Loon was groundbreaking, but they used a balloon form factor, and despite advanced algorithms and the ability to change altitude to find desired wind directions and wind speeds, Loon's system relied on favorable winds to stay over a target area, resulting in unpredictable station performance Franson says This required a large amount of balloons in the air to have relative certainty that one would stay over the area of operation which was financially unviable. He adds that Sky's airship can point into the wind and more effectively maintain its position. We have significant surface area, providing enough physical space to lift 250 plus kilograms and host solar panels and batteries, he says, allowing sky to maintain power through day-night cycles and therefore staying over an area of operation while maintaining altitude. Satellite internet currently comes at a price tag that can be too high for people in developing countries, says Kasabowski. For example, Starlink subscriptions start at $10 per month in Africa, but millions of people in these regions are surviving on a mere $2 a day. Franson and Oberg both claim that HAPs can connect the worlds unconnected more cheaply. Because satellites in low Earth orbit circle the planet at very high speeds, they quickly disappear from a ground terminal's view, meaning large quantities of those satellites are needed to provide continuous coverage. HAPs can hover, affording a constant view of a region, and more HAPs can be launched to meet higher demand. If you want to deliver connectivity with a low-Earth orbit constellation into one place, you still need a complete constellation, says Oberg. We can deliver connectivity with one aircraft to one location, and then we can tailor much more the size of the fleet according to the market coverage that we need. Starlink gets a lot of attention, but satellite internet has some major drawbacks, says Franson. A big one is that its bandwidth gets diluted once the number of users in an area grows. In a recent interview, Starlink co-founder Elon Musk compared the Starlink beams to a flashlight. Given the distance at which those satellites orbit the planet, the cone is wide, covering a large area. That's okay when users are few and far between, but it can become a problem with higher densities of users. For example, Ukrainian defense technologists have said that Starlink bandwidth can drop on the front line to a mere 10 megabits per second, compared with a peak offering of 220 megabits per second when drones and ground robots are in heavy use. Users in Indonesia, which like Japan is an island nation, also began reporting problems with Starlink shortly after the service was introduced in the country in 2024. Again, bandwidth declined as the number of subscribers grew. In fact, Franson says, Starlink's performance is less than optimal once the number of users exceeds one person per square kilometre. And that can happen almost anywhere. Even relatively isolated island communities can have hundreds of thousands of residents in a small area. There is a relationship between the altitude and the population you can serve, Franson says. You can't bring space closer to the surface of the planet. So the telco companies want to use the stratosphere so that they can get out to more rural populations than they could otherwise serve. Starlink did not respond to our queries about these challenges. Sky and Alto Haps see their stratospheric vehicles as part of integrated telecom networks that include both terrestrial cell towers and satellites. But they far from the only game in town World Mobile a telecommunications company headquartered in London thinks its hydrogen high UAV can compete directly with satellite megaconstellations The company acquired the HAPS developer Stratospheric Platforms in 2025. This year it plans to flight test an innovative phased array antenna, which it claims will be able to deliver bandwidth of 200 megabits per second, enough to enable ultra-HD video streaming to 500,000 users at the same time over an area of 15,000 square kilometres, equivalent to the coverage of more than 500 terrestrial cell towers, the company says. In 2025, World Mobile also signed a partnership with the Indonesian telecom operator Protolindo to build a prototype stratomast aircraft, with tests scheduled to begin in late 2027. Richard Deakin, CEO of World Mobile's Hapstavision World Mobile Stratospheric, says that just nine stratomasts could supply Scotland's 5.5 million residents with high-speed internet connectivity at a cost of £40 million, $54 million per year. That's equivalent to about 60 pence, 80 cents per person per month, he says. Starlink subscriptions in the UK, of which Scotland is a part, come at £75.100 per month. Companies working on HAPS also extol the convenience of prompt deployments in areas struck by war or natural disasters like Hurricane Maria in Puerto Rico, after which Loon played an important role. And they say that HAPS could make it possible for smaller nations to obtain complete control over their celestial internet-beaming infrastructure, rather than relying on mega-constellations controlled by larger nations, a major boon at a time of rising geopolitical tensions and crumbling political alliances. Analysts, however, remain cautious, projecting a haps market totalling a modest $1.9 billion by 2033. The satellite internet industry, on the other hand, is expected to be worth $33.44 billion by 2030, according to some estimates. The use of HAPs for internet delivery to remote locations has been explored since the 1990s, about as long as the concept of low-Earth orbit mega-constellations. The seemingly more cost-effective stratospheric technology, however, lost to the space fleets thanks to the falling cost of space launches and ambitious investment by Musk's SpaceX. Google wasn't the only tech giant to explore the HAPs idea. Facebook also had a project called Aquila that was discontinued after it too faced technical difficulties. Although the current cohort of Hapsmakers claim they have solved the challenges that killed their predecessors, Kasabowski warns that they're playing a different game, catching up with now-established internet-beaming mega-constellations. By the end of 2026, it'll be much clearer whether they stand a good chance of doing so. You are listening to MIT Technology Review, where Teresa Poulterova writes, Stratospheric Internet could finally start taking off this year. This article was published on the 27th of January, 2026, and was read by Martin Buchanan for NOAA.