Can a City Window Become a Power Plant?

Look down on a city and the constraint on solar power becomes geometry. Roofs lie like small postage stamps between roads, cooling equipment and access routes. Windows and walls rise for tens of stories. Sunlight touches that immense vertical area every day, but most of the surfaces that turn it into electricity still lie flat on top.

YKK AP’s answer is not to rebuild the outside of a tower. It is to add a second, generating window inside the glass already there. A thin, light perovskite photovoltaic layer would sit in a frame on the room side, with wiring carrying out direct-current power. Because the solar device also becomes part of the building assembly, the category is called building-integrated photovoltaics, or BIPV.

The important claim is not a cell-efficiency record. It is access. If a system can be installed and serviced from indoors without replacing a curtain wall or erecting extensive exterior scaffolding, a surface previously marked “unavailable” becomes a candidate. A removable frame could suit a leased property. The outer window could shield the cell from rain, hail and wind. A second pane might also be designed to reduce heat loss while it generates.

An inner window is not a generating curtain

An inner window is an independent assembly fitted on the room side of an existing exterior window. In homes, the format is already familiar as a retrofit that limits condensation and noise and improves insulation. The photovoltaic version incorporates cells into the frame or glazed area and takes electricity out through cables. It differs from replacing the entire exterior glazing with solar glass—in construction risk, maintenance, cost and the path sunlight must travel.

Calling it a “transparent solar window” can also hide the central trade-off. Perovskite chemistry, film thickness and electrodes can be adjusted to control color and transmission. But every photon passed to the room is one that cannot also be fully harvested. One 2025 semitransparent-cell study reported 20.59% power-conversion efficiency at just 9.45% average visible transmittance, and 18.73% efficiency at 20.71% transmittance. Those are impressive research results. Whether the remaining view feels like a window depends on the room. A real assembly also loses area and power to frames, wiring, encapsulation, patterning and scale-up.

The window’s job	The new question when it generates	What should be reported
Admit light and a view	How much darkness, color shift, haze or glare can the occupants accept?	Average visible transmittance, color rendering, haze and daylight by orientation
Control heat	Does reduced solar gain save cooling at the cost of winter heat or added electric lighting?	U-value, solar heat-gain coefficient, cavity temperature and annual HVAC effect
Act as safe construction	Can it meet fire, impact, seismic, condensation, opening and egress requirements?	Building and electrical standards, insulation, shutdown and breakage tests
Produce electricity	How much usable power survives the outer glass, year after year?	Active area, DC and AC yield, degradation, uptime and replacement interval

IEC 63092 treats BIPV as both a building product and an electrical product. That dual identity explains why a window company matters here. The job does not end when a high-performing cell is attached to a frame. Weather resistance, pressure, fire behavior, electrical isolation, routing and replaceability have to work as one product.

From a Russian mineral name to 3.8% in Yokohama

Perovskite is not the name of one substance. It describes a family of materials whose atoms form a particular crystal structure. The word reaches back to a mineral identified in the Ural Mountains in 1839. Today’s solar cell is not a slice of that rock. Researchers combine halides such as iodine or bromine with a metal and other ions to make a crystal film that absorbs light and releases electrical charges.

Japan occupies a foundational place in the modern history. In 2009, Akihiro Kojima, Kenjiro Teshima, Yasuo Shirai and Tsutomu Miyasaka reported organolead-halide crystals as light absorbers in a dye-sensitized cell. Their iodide device converted 3.8% of incident solar energy. It also contained a liquid electrolyte that dissolved the light-absorbing material, so durability was poor. When researchers moved to solid-state architectures around 2012, efficiency and stability work accelerated dramatically.

Research cells have since climbed into the high 20% range. The comparison needs its missing nouns: an opaque cell of tiny laboratory area is not a semitransparent cell, a large module or a finished window behind another pane of glass. Printability does not make manufacturing automatic. As film area grows, keeping it uniform, limiting defects, sealing every edge and repeating the same quality over thousands of units becomes much harder.

Five experiments on the road to the exhibit

The planned Expo window did not appear without a past. With Kandenko and other partners, YKK AP has changed climate, mounting position, cell format and building use to separate the problems one by one.

The progression matters. Akihabara and the snow festival made the idea visible. Haneda was a comparison site. Aomi added people at work. Osaka added a grid connection. Sapporo City Hall added a harsh winter and a public building. A commercial product would have to answer all of those settings at once.

Sunlight must pass through two windows

Being indoors protects the photovoltaic layer, but it imposes an optical toll. The sun first crosses the existing exterior pane. Glass that looks clear does not transmit every wavelength equally. Heat-reflective and low-emissivity coatings common in larger buildings deliberately reject parts of the infrared or visible spectrum to lower cooling demand. Change the amount and spectral composition of light reaching the cell and its output changes too.

That is why the Telecom Center trial explicitly examined the effect of heat-reflective exterior glass. A calculation based only on outdoor irradiance multiplied by cell efficiency cannot predict an inner window’s annual production. It needs the exact outer glazing, orientation, floor, neighboring shadows, cavity and cell temperatures, wiring and inverter losses.

A vertical surface generally receives less direct solar energy over a year than an optimally tilted, unobstructed roof. It also comes in south, east, west and north orientations. Yet value is not only an annual total. An east façade can contribute in the morning and a west façade later in the day, widening a roof array’s midday peak. Low winter sun can strike a vertical plane more directly. In Sapporo, snow can behave as a reflector. A façade shaded by the next tower all day should simply be excluded.

In a tower, the walls outgrow the roof

For a house or low warehouse with a sunny roof, mature silicon modules will often remain the first choice on output, price and warranty. The inner window is not a universal replacement. Its strongest case is the urban building whose roof is small relative to electricity demand or occupied by chillers, evacuation space, greenery and other equipment.

As a tower rises, floor area and façade grow while the roof remains one plane. That geometry creates the need for vertical generation. National Renewable Energy Laboratory simulations of high-rise buildings similarly found that photovoltaic glazing works best as one part of a package with exterior PV, better window heat performance and east–west surfaces. The result is a modeled possibility under defined assumptions, not a promise that every tower reaches net zero.

An inner assembly can reduce retrofit friction. An owner may test it without altering the exterior appearance, possibly one tenant space at a time. Workers can avoid some high-elevation exterior access and reach a failed unit from the room. But leased buildings create a split incentive: the owner pays for construction, the tenant may receive the energy savings and another party may hold the electricity contract. A product also needs a contractual architecture.

Why Japan is aiming for 20 gigawatts

Japan’s Seventh Strategic Energy Plan and 2025 Energy White Paper describe light, flexible perovskite photovoltaics as a way to expand renewable-energy siting in a country constrained by land and grid capacity. The national strategy targets about 20 gigawatts of deployment by 2040 and gigawatt-scale production capacity by 2030. NEDO’s Green Innovation Fund has allocated a maximum of ¥80.05 billion to next-generation solar-cell development, manufacturing and demonstration.

Iodine is often presented as an industrial advantage. Japan is the world’s second-largest producer, supplying roughly 30% of global output; most Japanese production comes from iodine-rich brine in Chiba Prefecture. For iodide perovskites, that could secure one important input domestically.

One ingredient does not create a domestic supply chain. Lead or a substitute metal, transparent conductors, electrodes, barrier films, glass, solvents, coating equipment and power electronics all matter. Yield in a mass-production line, energy consumption, imported content and intellectual property belong in any serious economic-security claim.

The largest enemy is not rain alone

Halide perovskites can degrade under interacting moisture, oxygen, heat, illumination, voltage and ion migration. A device that survives one accelerated stress may deteriorate unexpectedly when heat and light are applied together. The U.S. Department of Energy therefore identifies more than efficiency as the commercialization problem: operational durability, performance after scale-up, manufacturing reproducibility and bankability—the evidence that allows financiers and insurers to trust the asset. Mainstream generation needs a path to more than 20 years of service.

An inner window avoids direct rain and hail, but not aging. A south-facing cavity can become hot. Ultraviolet light still arrives. Edge seals expand and contract. Cleaning chemicals, condensation, earthquakes, furniture impact and repeated cable bending on an operable window are building realities. Because the unit is indoors, electrical isolation and safety after breakage matter wherever occupants can touch it.

If the window frame and generating layer have different useful lives, replaceability will determine the economics. The wooden frame used at Sapporo City Hall was intended to explore a removable approach suitable for leased sites. That detail shows why product design can matter as much as cell chemistry.

“It is only a thin layer” is not a lead policy

Most high-efficiency halide perovskites contain lead. A thin film uses relatively little, but lead dissolved from a broken, wet module remains an environmental and health issue. Lead-free compositions are under study, yet matching efficiency and stability remains difficult. Product responsibility must therefore continue from disclosure to robust encapsulation, containment after damage, collection and material recovery.

Countermeasures are measurable. A 2019 Nature Energy study simulated hail damage and found that a self-healing epoxy encapsulation cut the lead leakage rate by a factor of 375 compared with a glass package sealed only at the edges. Other work has put lead-absorbing material inside the module. These studies do not establish zero risk. They establish that a product can—and should—be tested after deliberate breakage, with the results made public.

In April 2026, Japan’s cabinet approved a bill that includes a recycling framework for solar panels, responding to projections that annual waste could reach as much as 500,000 tonnes in the late 2030s. That is a system-wide PV issue, not a perovskite-only one, and a cabinet-approved bill is not the same as an enacted operating regime. But a new technology should not repeat the old sequence of designing collection only after waste accumulates.

At the Expo, the method matters more than the watt

A sunny-day peak number makes an attractive demonstration. Measurement conditions make a useful one. GREEN×EXPO 2027 is scheduled to run from March 19 to September 26 at the former Kamiseya Communications Facility in Yokohama, spanning spring, summer and almost the autumn equinox. If operated continuously, the exhibit can be more than a static sample.

The unit matters too. Maximum watts from a cell describe capacity under a test condition. Electricity bills and avoided carbon depend on kilowatt-hours through a year. Research DC output is not the same as usable AC electricity in a building. If exhibit lighting, controls or data systems consume power, net output should subtract them.

Do not compare it panel to panel

On price per watt alone, mature rooftop silicon will usually defeat an early-production BIPV inner window. Building-integrated economics need a different baseline: the incremental cost over a non-generating window or over the retrofit already planned. Avoided exterior scaffolding or curtain-wall replacement, heating and cooling effects, occupant comfort and indoor service access may have value.

Benefits must not be counted twice. If a darker window lowers cooling load but requires more electric lighting, the effects offset each other. Insulation from the added pane cannot automatically be credited to the photovoltaic layer. If higher transparency reduces active area, a project with the same number of windows makes less power. The best product may not cover every pane. It may occupy a lower band, a spandrel-like zone or only the sunniest orientations while leaving the primary view clear.

More promising candidates	Cases requiring caution
Roof-limited mid- and high-rise buildings; existing façades that are difficult to replace; lightly shaded south, east and west windows; public buildings needing indoor maintenance access	Buildings with ample space for mature rooftop PV; permanently shaded north façades; windows where color fidelity and an unobstructed view dominate; locations where opening or egress is impaired
Buildings already planning insulation work; leased or temporary sites that value removability; vertical surfaces that can use reflected snow light	Strong solar-control outer glass without spectral data; tenant spaces with no cable route; short projects with no named party responsible for collection

So can a window become a power plant?

The answer is not that one window will run a tower. It is that roofs, exterior BIPV, efficient glazing, batteries and demand controls can be joined by a surface that previously did nothing. Thousands of modest generating areas can expand the urban renewable-energy boundary.

YKK AP’s advantage is not that it invented the perovskite. It is attempting to turn a light-harvesting film into a window that can be mounted, wired, replaced and lived beside. Since 2024 its demonstrations have moved through snow, heat-reflective glass, an occupied office, grid connection and a city hall. GREEN×EXPO can bring that progression before the public.

But a future window cannot be judged by the beauty of a glowing prototype. How many kilowatt-hours does it produce? How much light does it remove? How quickly does it age? How does it contain lead after breakage? Who takes it back? Is its total cost lower than the value of the electricity, construction and comfort it delivers? When those numbers are public, a generating exhibit becomes generating architecture.

A city cannot add more roof. Its windows are already there. That is why this restrained, interior intervention could matter.

Sources and reporting basis

1. YKK AP, “BIPV inner-window perovskite test begins at Sapporo City Hall” (March 2, 2026; Japanese)
2. YKK AP, “Haneda Building-Integrated Photovoltaics Demonstration Laboratory” (April 25, 2025; Japanese)
3. YKK AP, “Film-type perovskite BIPV inner-window test at Telecom Center Building” (Aug. 5, 2025; Japanese)
4. YKK AP, “Grid-connected BIPV inner-window test at Tanimachi YF Building” (Oct. 28, 2025; Japanese)
5. YKK AP and Panasonic Holdings, “Glass-type perovskite inner-window implementation test” (Nov. 20, 2025; Japanese)
6. YKK AP, “Glass-type perovskite solar-cell demonstration in Shizuoka Prefecture” (June 18, 2025; Japanese)
7. YKK AP and Sapporo, “Cooperation agreement for a next-generation BIPV demonstration” (Jan. 20, 2025; Japanese)
8. GREEN×EXPO 2027, official event overview
9. A. Kojima et al., “Organometal Halide Perovskites as Visible-Light Sensitizers for Photovoltaic Cells”, JACS (2009)
10. National Renewable Energy Laboratory, Best Research-Cell Efficiency Chart
11. U.S. Department of Energy, “Perovskite Research Directions”
12. NREL, “Combination of Stressors Key to Testing Perovskite Solar Cells” (2023)
13. Agency for Natural Resources and Energy, Energy White Paper 2025, next-generation energy section (Japanese)
14. NEDO Green Innovation Fund, Development of Next-Generation Solar Cells
15. NEDO, design and installation guidelines for film-type solar cells (March 18, 2026; Japanese)
16. Agency for Natural Resources and Energy, “Perovskite Solar Cells: Japan’s Strategy”
17. Chiba Prefecture, “Iodine, a Chiba resource of global importance” (Japanese)
18. IEC 63092-1:2020, requirements for BIPV modules
19. IEC 63092-2:2020, requirements for BIPV systems
20. NREL, “PV Windows Unlock Goal of Increased Energy Efficiency of Skyscrapers” (2022)
21. J. Li et al., “Fabrication of Bifacial-Modified Perovskites for Efficient Semitransparent Solar Cells”, Molecules (2025)
22. S. Jiang et al., “Reduction of lead leakage from damaged lead halide perovskite solar modules”, Nature Energy (2019)
23. X. Li et al., “On-device lead sequestration for perovskite solar cells”, Nature Sustainability (2021)
24. Ministry of Economy, Trade and Industry, cabinet decision on a recycling-law amendment bill (April 3, 2026; Japanese)

Editor’s note: Specifications unique to the GREEN×EXPO exhibit have not been invented beyond the news material for this edition. Figures for previous YKK AP demonstrations include plans current at the time of each release and do not represent commercial performance or warranty. Efficiency figures cannot be compared directly across cell area, transparency and test conditions. This report is based on material available by 3:14 a.m. Japan Standard Time on August 15, 2026.