A solar-powered airship floating 16.5 kilometres above the Pacific coast of Japan has handled phone calls, video streams and drone control for SoftBank, and done part of the network’s processing on board. SoftBank said on 2 September that it had completed Japan’s first HAPS trial service with US developer Sceye, off Cape Muroto in Kochi Prefecture.
HAPS stands for high-altitude platform station: an aircraft or airship that sits in the stratosphere, far above airliners but far below satellites, and acts as a cell tower in the sky. SoftBank and Sceye now aim to offer commercial service from 2027 onward.
Others have beamed mobile signals down from the stratosphere before, but what sets this trial apart is that part of the network’s brain went up with the radio.
What The Trial Actually Showed
Sceye installed a mobile core network, the software that normally sits in an operator’s data centre and manages calls and data sessions, together with a web server directly on the airship. According to Sceye’s account of the mission, requests processed on board came back in 68 milliseconds on average, more than 40 percent faster than sending them to cloud servers on the ground.
The rest of the test list reads like a disaster-preparedness checklist. Ordinary phones made voice calls, sent texts, streamed video and received earthquake and tsunami alerts, while the team also tested emergency calls to 118, Japan’s coast guard number, and remote control and video from drones. SoftBank says performance was comparable to ground networks, with little interference to existing cell sites.
Getting there was a test in itself: the airship, called ST1, left New Mexico on 9 August and spent about 13 days crossing more than 15,000 kilometres of the Pacific, then held station within a five-kilometre radius over its target area for more than a week before heading home.
Not everything in the headlines was a finished capability, and a laser test on 23 and 24 August tracked the airship from the ground and gathered data on how the atmosphere affects light, but, as one report noted, it did not carry traffic between the platform and the ground. It is groundwork for a future optical link, not the link itself.
Why Put The Network’s Brain In The Sky
The onboard core matters most when things go wrong below, because a normal stratospheric relay still needs a connection back to the operator’s ground network, so if an earthquake cuts the fibre and power that connection may fail too. An airship that can handle calls and data itself can keep people in touch even when disaster damages the ground side. That is a very Japanese concern, given how often earthquakes and tsunamis knock out cell sites there.
The other target is the sky itself, and SoftBank chief executive Junichi Miyakawa has argued that HAPS will be key infrastructure once “3D telecommunications networks will be needed to support sky-based mobility”, meaning drones and other aircraft that networks designed for people on the ground were never built to serve. Processing data close to those drones cuts delay for control and video.
For IoT builders, the edge-computing result is the part to watch. Drones, ships and remote sensors are exactly the devices that live beyond the reach of ground networks, and a platform that can both connect and process their data nearby is closer to a flying data centre than a flying antenna.
A Long Road To This Point
SoftBank has been at this for a while, having started developing HAPS in 2017 and run what it called the first stratospheric 5G connectivity test in 2023. In June 2025 it changed pace, investing in Sceye and taking exclusive rights to offer HAPS services in Japan using Sceye’s lighter-than-air platform, while continuing its aircraft work in parallel.
Sceye, founded in 2014 and based in Moriarty, New Mexico, raised a Series C round in September 2024 at a $525 million pre-money valuation and flew to Brazil in April 2026 before the Japan mission.
“Flying from the US to Japan demonstrates the performance required to make the stratosphere a viable layer of infrastructure,” said chief executive Mikkel Vestergaard Frandsen.
Japan’s other big operator is running a rival programme: NTT Docomo and Space Compass committed $100 million to AALTO, the Airbus unit behind the fixed-wing Zephyr, and in 2025 they connected ordinary 4G phones from about 20 kilometres above Kenya, with a service launch targeted for 2026.
History offers a caution, though, since Alphabet’s balloon venture Loon shut down in 2021 after, in the words of its chief executive Alastair Westgarth, failing to get the costs low enough to build a sustainable business.
Where HAPS Fits Between Towers And Satellites
The case for HAPS rests on its middle position: at 16.5 kilometres, the airship sits far closer to the ground than even low-orbit satellites, so signals are stronger and delays shorter, and Sceye claims one platform can cover an area equivalent to about 500 ground towers.
Regulators have opened the door, too, after the 2023 World Radiocommunication Conference identified mobile bands for HAPS acting as mobile base stations, including frequencies between 694 and 960 megahertz and at 1.7, 2 and 2.6 gigahertz, which is what lets them talk to ordinary phones and IoT devices rather than special terminals.
That puts HAPS in competition, and sometimes partnership, with satellites that connect directly to phones. We looked at that race in AST SpaceMobile’s bet on cell towers in orbit and in SpaceX’s Starlink Mobile plans. Satellites cover the whole planet; an airship covers one region intensely, and an operator can move it wherever coverage is missing.
What Still Has To Be Proven
SoftBank itself says commercial service depends on setting up an operating framework and further improving communication quality. Beyond that, the questions are the ones that sank Loon: how long each airship can stay aloft, how reliably it holds position through Pacific weather, and what each hour of coverage costs compared with towers and satellites.
Everyday Infrastructure Or Emergency Backup
A flight across the Pacific and a week on station prove that the platform can reach Japan and do useful work there. They do not yet prove that SoftBank can keep several airships aloft for months, through changing weather, at a cost that competes with towers on the ground and satellites overhead.
SoftBank’s answer will decide what HAPS becomes: if the numbers work, stratospheric platforms could become a standing layer of coverage for drones, ships and remote sensors along Japan’s coast; if they don’t, HAPS may end up as a premium emergency service that goes up when an earthquake takes the ground network down. The commercial launch planned from 2027 will show which way it leans.
