> ## Content Index
> Fetch the complete content index at: https://nexi.fund/llms.txt
> Use this file to discover other available public pages before exploring further.

# SoftBank's Stratospheric Bet: 40 Airships, One 30-Day Ceiling
- URL: https://nexi.fund/sceye-softbank-stratospheric-haps-fleet-2026/
- Published: 2026-11-08T09:00:49.000Z
- Updated: 2026-11-08T09:00:49.000Z
- Description: SoftBank will buy a fleet of at least 40 Sceye high-altitude platforms for Japan after a 30-day trans-Pacific flight. The contract answers the demand question; persistence and turnaround remain unsolved.
- Author: Nexi.fund Labs
- Tags: Space & Expansion, #mode-5, #hook-statistic, #track-B, #layout-asymmetric-long, #brand-heavy

On 9 August, an uncrewed airship left Roswell, New Mexico, and climbed to 20 kilometres. It crossed the Pacific, spent more than a week over Kochi Prefecture, and came home on 4 September. Thirty days. Nearly 30,000 km. No rocket.

On 3 November, SoftBank signed a five-year master services agreement to buy a fleet of at least 40 of them for Japan.

The pitch is narrow and specific. A stratospheric platform can be on station in under 30 minutes, hold a position inside a 5 km circle, and be recovered and replaced. A satellite constellation waits years for a launch slot, then stays where it is until it dies.

The constraint is just as specific. Sceye's platform has flown for 30 days. The company wants six months.

## The 20-kilometre layer

High-altitude platform stations are an old idea riding a new cost curve. A lighter-than-air vehicle uses lifting gas to reach the stratosphere, where winds are manageable and the thin air gives a wide line of sight. Solar panels charge batteries by day; the batteries carry the platform through the night.

Sceye says it was the first to close that power loop, holding altitude over the same area from dusk until the next sunrise. That claim carries more weight than any altitude record. A platform that cannot survive the night is a drone with good marketing; one that can is a candidate for infrastructure.

At roughly 20 km, the platform sits about 1,800 times closer to the ground than a geostationary satellite, according to the company. The practical result is that standard phones connect directly, indoors included, with no special handset and no dish. The link behaves like a cell tower that happens to be 20 km up.

The fleet has two jobs. Most of the 40-odd platforms will carry mobile broadband. A subset will form a free-space optical network, with laser links between platforms. That optical layer is the part that could eventually knit the stratosphere to satellites and ground stations.

---

## What the Japan flight actually proved

SoftBank's own mission log is more useful than the announcement. Service Test 1 departed at 22:00 JST on 9 August and reached Japanese airspace at 07:11 JST on 23 August, after more than 15,000 km and roughly 13 days aloft. It then operated around Muroto City in Kochi Prefecture, holding position within a 5 km radius. Station-keeping is where stratospheric platforms usually fail, so that number matters.

During a week over Japan, the platform carried text messages, voice calls, internet access and video streaming to standard smartphones, including indoors. It also supported emergency communications, remote control of a drone, and edge computing on board. Sceye's chief executive says the onboard compute cut latency by 40% compared with sending the same workload to the cloud.

Two further results support the commercial case. A ground station tracked the platform with laser ranging, a world first according to Sceye and the groundwork for optical links. The platform also returned to New Mexico on 4 September, completing a 30-day, nearly 30,000 km round trip.

| Metric                   | Figure                     | Source        |
| ------------------------ | -------------------------- | ------------- |
| Operating altitude       | \~20 km                    | Sceye         |
| Time to station          | under 30 minutes           | Payload Space |
| Fleet target             | 40+ platforms over 5 years | Payload Space |
| ST1 duration             | 30 days                    | SoftBank      |
| ST1 distance             | \~30,000 km                | Sceye         |
| Station-keeping          | 5 km radius                | SoftBank      |
| Edge-compute latency cut | 40% vs cloud               | Payload Space |
| SoftBank investment      | $15M, Series C 2025        | Payload Space |

Numbers from the 3 November agreement, Sceye's ST1 mission report and SoftBank's trial-service release.

## The persistence problem

The 30-day record is the number to watch. Sceye currently recovers its platforms, breaks them down and tests the parts before reuse. Turnaround is measured in months. Mikkel Vestergaard Frandsen, the founder and chief executive, wants that down to weeks and single vehicles staying aloft for six months or more.

That gap is the difference between a fleet and a demonstration programme. Forty platforms that each need a month of refurbishment after a month in the air do not produce continuous coverage. Forty platforms with a six-month duty cycle and a two-week turnaround start to look like a network.

Weather sets the second limit. Stratospheric winds vary by season and latitude, and SoftBank's release notes that the flight plan accounted for wind direction and speed. Station-keeping inside 5 km is impressive; doing it through a Japanese winter is the next test.

SoftBank is not treating HAPS as a satellite replacement. The same operator is adopting Starlink's direct-to-cell service in Japan. Its stated model is a three-layer network: ground, stratosphere, orbit. That is convenient for a telecom that wants resilience without betting on one technology. It also sets Sceye's real benchmark, which is cost per covered user against both a terrestrial tower and a satellite link.

🎯

Sceye sold SoftBank a service, not a satellite: the platform stays owned and operated by the manufacturer.  
  
The first commercial fleet is aimed at nationwide Japanese coverage, with an optical subset layered on top.  
  
The binding constraint is now operations, not altitude. 

## Why Japan is the first market

Japan is a hard test case, which is why it is a useful first market. The country is mountainous and island-dotted, with rural districts that are expensive to cover from towers and increasingly short of workers. It is also earthquake-prone, and a platform that can be repositioned in half an hour fits the disaster-response playbook better than a satellite that cannot move.

The September trial was the first stratospheric communications test in Japanese airspace, according to SoftBank, covering links to smartphones and to drones. The regulatory work that made it possible is part of the asset. Spectrum, flight approval and airspace coordination are slow, national processes; a five-year service agreement gives both companies a reason to keep pushing them.

SoftBank's wider interest is the same one driving its AI and cloud investments. It describes the target as a ubiquitous network that combines ground, sky and space, with 6G, IoT and edge computing as future applications. A stratospheric node is attractive in that architecture because it can host compute close to users without the latency and cost of a data centre or the launch schedule of a satellite.

> This agreement represents the commercialization of HAPS and the realization of our vision for Ubiquitous Transformation.— Masayuki Kamimura, Senior Director, Ubiquitous Network Planning Division, SoftBank Corp.

## What would make this infrastructure

Three numbers decide whether the fleet becomes a layer or stays a pilot programme. The first is turnaround: weeks, not months. The second is persistence: six months aloft, not thirty days. The third is cost per covered user, measured against a tower in the valley and a satellite pass overhead.

None of those numbers is in the 3 November announcement. The contract commits SoftBank to a five-year relationship, and Sceye says it is moving into full-scale production. That is a statement of intent backed by a customer, which is more than most stratospheric ventures have ever had.

The 30-day flight proved the physics. The five-year contract proves SoftBank's patience. The next 18 months have to prove the maintenance model. The stratosphere does not need a better rocket. It needs a schedule that keeps forty platforms in the air.

[ Payload Space: Sceye signs a five-year HAPS fleet deal with SoftBank The clearest account of the commercial structure, fleet size, turnaround targets and the 30-day Service Test 1 results. Payload Space ](https://payloadspace.com/exclusive-sceye-signs-five-year-deal-to-fly-haps-fleet-for-softbank/?ref=nexi.fund) 

The interview with Sceye's chief executive supplies the operating detail the press release leaves out: ownership, service model and the six-month ambition.

[ SoftBank Corp.: Japan's first HAPS trial service The counterparty's mission log, with departure and arrival times, the 15,000 km Pacific crossing and the 5 km station-keeping radius. SoftBank Corp. ](https://www.softbank.jp/en/corp/news/press/sbkk/2026/20260902%5F01?ref=nexi.fund) 

Primary evidence for the flight itself, including the Kochi operating area and the smartphone and drone links tested in Japanese airspace.

[ Sceye: five-year agreement for a stratospheric fleet over Japan The announcement of the master services agreement, the split between broadband and optical platforms, and the SoftBank investment history. Sceye ](https://sceye.com/press-releases/sceye-softbank-sign-agreement-for-fleet-of-stratospheric-platforms/?ref=nexi.fund) 

Useful for the company's own framing of the stratosphere as an infrastructure layer rather than a satellite substitute.