Japan’s Four Mobile Rivals Tackle the Same 28GHz Problem

Mobile networks are built to compete. Coverage, capacity and reliability are part of what separates one carrier from another. Yet Japan’s four national mobile operators — KDDI, NTT DOCOMO, SoftBank and Rakuten Mobile — have now agreed that one particularly difficult radio problem may be better studied together: how to make 28GHz 5G useful in more places.

The companies signed a memorandum of understanding on September 11 and announced it on September 16. The study will examine the technical feasibility of expanding the usable coverage of millimeter-wave base stations across operator boundaries and the operating arrangements that might make such a system efficient. Base stations, repeaters and related assets are explicitly within scope. At the same time, the release says each company will continue deploying its own millimeter-wave infrastructure. ^[1]

28GHz offers capacity by accepting a harsher law of physics

The attraction of millimeter-wave 5G is bandwidth. Wide channels can carry large amounts of data and support very high peak rates. When DOCOMO launched commercial 28GHz service in September 2020, it advertised a technical maximum downlink of 4.1Gbps. ^[5]

The same high frequency that makes those wide channels practical also makes coverage difficult. The four carriers’ announcement says millimeter-wave signals are highly directional and susceptible to attenuation and blockage. Buildings, walls, foliage and even crowded human environments can change the radio path. Compared with lower-frequency cellular bands, a continuous service area can require many more radio points, repeaters and carefully chosen lines of sight. ^[1]

A simple speed-of-light calculation puts the wavelength at roughly 10.7 millimeters. That short wavelength helps engineers build compact antenna arrays and narrow beams, but the radio link is less forgiving when something gets between transmitter and receiver.

Japan launched 5G in 2020 — but “5G coverage” never meant “millimeter-wave everywhere”

Japan’s commercial 5G era began in the spring of 2020. DOCOMO launched on March 25, KDDI on March 26 and SoftBank on March 27, while Rakuten Mobile was building its own new 5G network that year. ^{[4] [6] [7] [8]}

But 5G is not one frequency. Operators can deliver 5G on mid-band spectrum such as 3.7GHz and 4.5GHz, on repurposed frequencies previously used for LTE, and on much higher millimeter-wave bands. Expanding the area in which a phone displays “5G” is a different engineering task from creating dense 28GHz coverage with very wide channels.

DOCOMO’s own early rollout illustrates the difference. It used allocated 28GHz spectrum in its 2019 pre-service and began commercial millimeter-wave service on September 23, 2020. At launch, DOCOMO said 66 locations were ready, with 164 planned by the end of that month. Millimeter-wave began as a targeted layer, not as a blanket over the country. ^[5]

The new four-carrier study did not appear from nowhere

Japan’s operators have already spent years experimenting with a boundary that would once have sounded contradictory: compete at the service level, cooperate where duplicating physical infrastructure is expensive.

The clearest example is 5G JAPAN, the joint venture created by KDDI and SoftBank in 2020. By the end of fiscal 2023, the two companies said their collaboration had jointly built more than 38,000 5G base stations per company and reduced capital spending by ¥45 billion for each operator. In 2024 they began discussing a broader program extending the joint build from rural areas toward nationwide coverage, with a long-term target of 100,000 stations and ¥120 billion of cost reduction per company by fiscal 2030. ^[3]

Cooperation took another form in April 2026 with JAPAN Roaming, an emergency service that allows users to connect temporarily to another carrier’s 4G LTE network during major disasters or outages. It is not ordinary roaming and it does not solve the 28GHz problem. But it shows that cross-carrier interconnection is no longer treated as exceptional when the public benefit is clear. ^[12]

Then came a development directly relevant to the new study. In May, KDDI and DOCOMO announced a Kyocera-built repeater that can relay both operators’ 28GHz signals in one unit. The device combines filters and amplifier circuits that previously had to be installed separately and can select the best antenna face for the incoming signal from each operator. The companies say the shared unit can reduce installation cost and physical space compared with deploying separate repeaters. ^{[2] [9]}

Why repeaters matter

The brute-force answer to weak millimeter-wave coverage is to install more base stations. In a city, however, the hard part is often not the radio itself. A site needs a location agreement, power, backhaul, mounting hardware, construction access and maintenance. Streetscapes impose limits on size and appearance. Four carriers duplicating all of that at every high-demand location can be technically possible and economically clumsy.

A repeater takes a usable signal from one direction and relays it toward another area. KDDI and DOCOMO’s 2026 shared device measures 216 by 216 millimeters and 246 millimeters high, weighs 4.9 kilograms and is designed for installations such as streetlights. It can also choose alternative relay paths when buildings, vegetation or other environmental changes degrade a route. ^[2]

That does not mean the four-carrier study will simply adopt that same product. The new memorandum gives no architecture. But the repeater demonstrates the underlying economics: one physical installation can potentially carry value for more than one carrier without forcing those carriers to merge their networks.

2026 is showing where millimeter-wave is most useful

Japan’s recent deployments increasingly concentrate on places where traffic density makes ordinary capacity scarce.

At Comic Market 108 in August, KDDI used a vehicle-mounted millimeter-wave repeater as part of its event network. In September, at ROCK IN JAPAN FESTIVAL 2026, KDDI and Wi2 went a step further: 28GHz was used as the backhaul for carrier-neutral Wi-Fi. Visitors did not need a millimeter-wave handset or even an au subscription to benefit indirectly from the high-capacity radio link. ^[10]

Another experiment attacked the problem of metal railway cars. KDDI and JR East used a millimeter-wave-compatible glass antenna to bring an outdoor signal into a Yamanote Line carriage and rebroadcast it inside. In a stationary train, the companies said the portion of the car capable of 1Gbps communication rose from about 40% to about 97%. ^[13]

DOCOMO, meanwhile, announced a smart-city program with Mitsubishi Estate in the Otemachi-Marunouchi-Yurakucho district. The plan combines existing 28GHz holdings with newly acquired 26GHz spectrum and advanced antenna systems, while explicitly treating urban aesthetics and line-of-sight placement as engineering constraints. ^[11]

Taken together, these projects point toward a practical role for millimeter-wave: not necessarily one continuous nationwide layer, but extremely high-capacity islands placed where people, devices and traffic concentrate.

“Sharing infrastructure” can mean several very different things

A cellular site is a stack of systems. Operators can share a tower or rooftop while keeping separate radios. They can share antennas or radio equipment while keeping separate spectrum. They can share transport and power. More extensive RAN-sharing architectures can go further still. Those are not interchangeable arrangements.

The September memorandum deliberately leaves the answer open. It says the study will cover technologies and approaches involving millimeter-wave base stations, repeaters and related assets, and that deployed facilities should be used more effectively. It does not say which network layer will be shared or whether a customer of one carrier will attach directly to another carrier’s radio. ^[1]

Sharing equipment does not mean identical networks

Infrastructure sharing creates an obvious economic opportunity, but it also touches the core of competition. Japanese operators compete on network quality as well as price. Site selection, capacity planning, tuning and operational discipline all contribute to the experience customers perceive.

The four-carrier release therefore includes an important qualifier: each operator will continue actively deploying its own millimeter-wave equipment. The collaboration is framed as a way to use deployed facilities more effectively, not as a replacement for independent investment. ^[1]

Nor would a shared installation automatically make service identical. A handset must support the relevant bands and radio features; authentication and network configuration still matter; and performance still depends on congestion and propagation conditions. Sharing can make more radio available in more places without erasing every technical difference between carriers.

Six years into 5G, the hard questions are less glamorous

The first 5G announcements were dominated by peak speed: gigabits per second, ultra-low latency and new immersive services. By 2026, the more consequential questions are often physical and operational. Where can an antenna be mounted? Who pays for the site? How many boxes can fit on a pole? Can one repeater serve several networks? How do operators add capacity to a packed festival, station or business district without building four parallel versions of the same street furniture?

Millimeter-wave makes those questions unavoidable. Its wide bandwidth is valuable precisely where data demand is intense, but its propagation characteristics demand dense, careful infrastructure. The technology’s greatest strength and its deployment weakness are two sides of the same frequency choice.

The four-carrier memorandum is not a solution yet. It is a sign that Japan’s operators now see the 28GHz coverage problem as partly a shared infrastructure problem rather than four completely separate engineering problems.

After 5G JAPAN, the KDDI-DOCOMO shared repeater and nationwide emergency roaming, the next experiment is broader: four competitors asking what they can share without surrendering the things on which they still intend to compete.