A 13-day crossing and a first for Japan
SoftBank Corp. and Sceye, Inc. completed Japan's first trial service from a high-altitude platform station (HAPS) operating in the stratosphere, the companies announced on September 2, 2026. Sceye's lighter-than-air platform departed Roswell, New Mexico at 10:00 p.m. Japan Standard Time on August 9, 2026, flew more than 15,000 km across the Pacific Ocean in roughly 13 days, and reached Japanese airspace at 7:11 a.m. JST on August 23. It then held position off Cape Muroto in Kochi Prefecture, within a radius as small as 5 km, for an extended period while communications tests ran through SoftBank's core network.
The trial is the first successful stratospheric communications test in Japan using a HAPS, according to SoftBank's own research as of August 23, 2026. It is also the first HAPS-to-smartphone and HAPS-to-drone communications test conducted in Japanese airspace. The platform remained over the test area long enough to run voice calls, text messaging, app usage, video streaming, video calls, and an emergency alert messaging system for earthquake and tsunami warnings.
SoftBank is betting that a stratospheric platform can fill the gaps where terrestrial towers fail, and it wants to sell that capability from 2027. The Cape Muroto test area was chosen specifically to simulate a large-scale disaster scenario, such as an earthquake on the Nankai Trough. If the company can keep a platform on station for weeks at a time and serve smartphones directly from 20 km up, it has a credible answer to a problem Japan has faced repeatedly: networks that collapse when the ground infrastructure does.
What the trial actually proved
The headline numbers are the 15,000 km transit and the 13-day flight, but the operational proof matters more. Sceye maintained the platform within a 5 km radius of the target area despite stratospheric wind conditions, confirming both long-distance transit and station-keeping capability under Japanese weather. That is the hard part. Anyone can launch a balloon. Holding it over a disaster zone for days is the engineering problem.
On the communications side, the test used a 4G LTE-equivalent base station installed on the HAPS, connected to SoftBank's core network through a ground gateway. The companies confirmed stable voice calls including emergency calls to the 118 maritime number in cooperation with the Japan Coast Guard, text messaging, social media apps, video calls, and video streaming. Data throughput was equivalent to terrestrial networks while reducing radio interference with ground base stations.
SoftBank also tested HAPS-to-drone communications, with automated drone flight controlled remotely through the HAPS link. Flight control, location transmission, and video all remained stable. The company lists the expected drone applications as disaster damage assessment, supply transport, monitoring of remote islands and offshore areas, logistics in mountainous regions, infrastructure inspection of power lines and roads, and forest monitoring.
Edge computing at 68 milliseconds
The most novel result was the world's first technical verification of HAPS-based edge computing, per SoftBank's claim based on publicly available information as of August 23, 2026. A processing server installed on the platform ran a mobile core network and web server onboard, so smartphone requests could be handled entirely within the HAPS without routing through a terrestrial network. Average round-trip response time was 68 milliseconds, cutting latency by more than 40 percent compared with cloud processing via the internet.
The stated motivation is Physical AI, which SoftBank defines as technology that lets AI analyze sensor and camera data and make decisions so robots can perform flexible and complex movements. Physical AI needs both wide-area coverage and low latency, and a stratospheric platform with onboard compute is one way to get both. Whether that combination finds buyers beyond disaster response is an open question. The 68 ms figure is promising, but it was measured in a controlled trial with limited devices and durations.
It is not clear this will hold at commercial scale. SoftBank's own press release notes the communications tests were conducted with limited durations, areas, and devices. That caveat matters. A trial is not a service, and a single platform over one prefecture is not a network.
What comes next
SoftBank and Sceye plan to commercialize HAPS services in Japan from 2027 onward, using data from this trial to develop operational frameworks and improve communications quality. The platform was still crossing the Pacific back to the United States as of September 2 for extended continuous-flight verification, so the endurance question is not fully settled even now.
Looking toward the 6G era, SoftBank aims to build a three-dimensional communications network integrating satellite, HAPS, and terrestrial systems for connectivity on the ground and to aerial mobility such as drones. Junichi Miyakawa, SoftBank's president and CEO, said the trial marks an important step toward commercialization and that the company has gained confidence to realize a three-dimensional network. Mikkel Vestergaard Frandsen, Sceye's founder and CEO, called the flight a defining moment for the commercial potential of HAPS.
For more on high-altitude platform stations and stratospheric communications, see the Light Reading coverage. The relevant comparison for buyers is not whether the technology works in a trial. It is whether SoftBank can price a service that competes with the satellite constellations already in orbit, and whether regulators treat a 20 km platform as an aircraft, a telecom tower, or something new. That answer will come in 2027, not before.
Why this matters beyond SoftBank
The HAPS idea is not new. Solar-powered and lighter-than-air platforms have been tested for years by Airbus, Google successor efforts, and various defense programs. What makes this trial worth noting is the combination of long-range autonomous transit, station-keeping, direct smartphone connectivity, drone links, and onboard edge computing in a single flight. That is a broader stack than most prior demonstrations, and it is the stack SoftBank would need to sell a commercial service rather than a science experiment.
Japan's geography makes the case obvious. The country has remote islands, mountainous terrain, and a coastline exposed to tsunamis and typhoons. Terrestrial towers are expensive to harden and slow to repair. A platform that can be positioned within hours and serve a 5 km radius from 20 km up is attractive to disaster planners even if the commercial case is still unproven. The 118 maritime emergency call test with the Japan Coast Guard is a signal that SoftBank is pitching this to government buyers as much as to consumers.
The regulatory picture is less settled. A 20 km platform is neither an aircraft in the conventional sense nor a telecom tower, and aviation authorities in Japan and the United States will have to decide how to certify operations, collision avoidance, and end-of-life descent. SoftBank has not disclosed pricing, coverage maps, or service tiers, and the company's own caveat about limited test scope should temper expectations for a 2027 launch.
What comes next
SoftBank and Sceye plan to commercialize HAPS services in Japan from 2027 onward, using data from this trial to develop operational frameworks and improve communications quality. The platform was still crossing the Pacific back to the United States as of September 2 for extended continuous-flight verification, so the endurance question is not fully settled even now.
Looking toward the 6G era, SoftBank aims to build a three-dimensional communications network integrating satellite, HAPS, and terrestrial systems for connectivity on the ground and to aerial mobility such as drones. Junichi Miyakawa, SoftBank's president and CEO, said the trial marks an important step toward commercialization and that the company has gained confidence to realize a three-dimensional network. Mikkel Vestergaard Frandsen, Sceye's founder and CEO, called the flight a defining moment for the commercial potential of HAPS.
For more on high-altitude platform stations and stratospheric communications, see the Light Reading coverage. The relevant comparison for buyers is not whether the technology works in a trial. It is whether SoftBank can price a service that competes with the satellite constellations already in orbit, and whether regulators treat a 20 km platform as an aircraft, a telecom tower, or something new. That answer will come in 2027, not before.
For related coverage on our site, see robotics and drones.