"What if every road was a power station, every building a solar panel, and every car ran on the city itself?"
SolarRoad is an open invention proposal for a fully integrated, self-powered road and urban energy ecosystem. Inspired by the simplicity of Scalextric (toy cars powered by an electric circuit track), this project proposes applying that same principle to real cities and highways: roads that power the vehicles travelling on them, harvest solar and thermal energy, guide autonomous transport, and eliminate fossil fuel dependency entirely.
No petrol. No emissions. No traffic jams. No road deaths.
A child's Scalextric set is a perfect proof of concept. The car needs no engine, no fuel tank, no exhaust — it draws power continuously from the track it runs on. Scale that principle to real roads, combine it with solar collection, superconducting transmission, AI traffic management, and next-generation energy storage, and you have an integrated energy and transport system that removes petrol from the equation entirely.
Every road surface becomes a solar collector. Every building window becomes a power generator. Every vehicle draws power from the road itself. Underground superconducting cables carry that energy with near-zero loss. AI manages the entire network in real time.
This is not science fiction. Every component technology exists today in some deployed or demonstrated form. The hard part is integration, materials engineering, honest economics, and the will to deploy. Pieces of the idea have already been tried on real roads, and the lessons from those first attempts shape every design choice here: see Standing on the Shoulders of the Pioneers.
Petrol is finite. That fact alone makes this transition inevitable — the only question is whether we do it proactively or wait until we are forced to. The long-term benefits to the planet, to human health, and to future generations vastly outweigh the initial investment cost. The rollout would be gradual — exactly as 4G was replaced by 5G, as gas lighting gave way to electric — road by road, city by city, country by country.
The initial cost is real. So is the cost of doing nothing.
Transport accounts for 24% of global CO₂ emissions — approximately 8 billion tonnes per year. Australia alone produces ~100 million tonnes of transport CO₂ annually. Globally, 1.35 million people die on roads every year and a further 50 million are injured — nearly all preventable with AI-managed autonomous systems.
Continuing to build petrol infrastructure is not pragmatism. It is short-term thinking.
To illustrate the energy potential, consider Sydney as a case study.
Road surface area: Sydney's metropolitan road network covers approximately 12,000 km of roads. At an average width of 10 metres, that is 120 km² of surface, all of it currently absorbing sunlight and radiating it back as waste heat.
With transparent solar substrate and Sydney's average of 5.5 peak sun hours per day, the theoretical ceiling looks like this:
| Source | Daily generation | Annual generation |
|---|---|---|
| Solar road substrate (conservative, 10% efficiency) | 66,000 MWh | 24,090 GWh |
| Solar road substrate (optimistic, 20% efficiency) | 132,000 MWh | 48,180 GWh |
| Solar building glass (50,000 buildings, 8% efficiency) | 11,000 MWh | 4,015 GWh |
| Thermal harvesting — TEG road modules (5% efficiency) | 26,400 MWh | 9,636 GWh |
| Total (conservative) | 103,400 MWh | 37,741 GWh |
| Total (optimistic) | 169,400 MWh | 61,831 GWh |
The numbers above are a physics ceiling, not a forecast. Roads lie flat, traffic shades them, and a transparent wear layer absorbs some light before it reaches the cells, so real yields start lower. China's 1,080 m Jinan road bears this out: it generated about 1 GWh/year, or roughly 100 kWh/m²/year — about half what the "conservative" row above assumes. Factor in those losses, plus a road-TEG efficiency closer to 1–2% than 5% (see Section 3), and the cautious starting picture looks like this:
| Source | Realistic annual generation | Basis |
|---|---|---|
| Solar road substrate (evidence-grounded) | ~12,000–15,000 GWh | Jinan-measured yield, Sydney's higher sun |
| Solar building glass (8% AVT cells) | ~3,000–4,000 GWh | tilted-facade BIPV, less derating |
| TEG road modules (realistic 1–2%) | ~2,000–3,800 GWh | field-validated road TEG output |
| Realistic total | ~17,000–23,000 GWh |
Sydney's total annual electricity consumption is approximately 45,000 GWh.
Even at these cautious, field-validated yields, SolarRoad covers roughly 40–50% of Sydney's entire electricity demand from surfaces that currently generate nothing — and that is the floor, using technology already demonstrated. Transparent-PV efficiency improves every year. As it climbs toward the optimistic case, the same roads and windows move past half the city's needs and eventually exceed all of it.
The road network stops being infrastructure that costs public money. It becomes infrastructure that generates it.
Solar roads are not a new idea, and that is good news. Several teams have already built the first generation and learned the hard lessons in public. SolarRoad starts where they left off.
The most ambitious was Solar Roadways Inc in Idaho, which tried to make the road do everything at once — generate power, light up with LEDs, heat itself, and carry traffic on a glass surface, all in hexagonal tiles. The Sandpoint pilot peaked at about 80,000 kWh a year against an expected 150,000, most panels never produced power, and the LEDs alone ate more than a quarter of the output. At roughly $39,000 per kW installed, against $4,000 for conventional solar, the cost per watt was the project's undoing.
France's Wattway was simpler: thin PV panels glued onto an existing road near Tourouvre — €5 million for a kilometre, 2,800 panels. But the engineers underestimated leaves, shade, dirt, and the weight of farm vehicles. Panels cracked and delaminated, about 90 metres were destroyed beyond repair, and the road was decommissioned in 2019. Colas, the builder, retreated to small off-road modules.
China's Jinan solar expressway got closer. Its 1,080-metre surface used 90%-transmission concrete, peaked at 817 kW, and fed about 1 GWh a year into the grid — but it still suffered surface damage and theft, and the load-bearing format never scaled, because flat-mounted PV under traffic costs far more per kWh than the same panels at the roadside.
The common thread is the solar driving surface itself. Flat orientation, traffic shading, soiling, and cost per watt all work against a PV cell built into a road you drive on. That tells us where to be careful, not where to give up. SolarRoad makes three choices in response.
- Lead with what already works. Earlier projects lived or died on the solar surface alone. Here the road earns its keep first through dynamic inductive charging (field-proven, 85–88% efficient, about $1.25M/km) and grid power. The transparent PV surface is a layer added on top, not the foundation. No charging stops, smaller batteries, and AI safety all arrive even on a stretch of road that generates no solar at all.
- Put the solar where the sun is. Vertical building glass, noise-barrier and median canopies, and tilted roadside arrays all out-yield a flat, shaded carriageway per dollar and per watt. The driving surface is the last place to harvest sun, not the first, and the Sydney model above is weighted accordingly toward facades and roadside collection.
- Use modular tiles, not monolithic strips. Replaceable cassette modules with embedded sensors (Layer 3) mean a dead cell gets swapped out, not a 90 m strip demolished — exactly the repair Wattway and Solar Roadways couldn't make.
The short version: the parts that sank earlier projects are the parts SolarRoad relies on least, and the part it relies on most, wireless charging, is the part already proven on real roads.
Transparent solar road substrate The road surface is replaced with a load-bearing transparent resin composite embedded with photovoltaic cells. Vehicles drive on it normally; sunlight passes through and is converted to electricity below. Engineered for high compressive strength, maximum solar transmittance, thermal stability, and self-cleaning surface properties.
Solar building glass (metropolitan deployment) In cities, conventional windows are replaced with transparent photovoltaic panels. Buildings retain full visual transparency while every window generates electricity. Combined with thermal conversion of solar heat absorbed by building facades, the entire built environment becomes a distributed power station. In a city like Sydney, where towers receive direct sunlight across enormous glass facades, the energy yield is substantial — see the Sydney Calculation above.
Relay energy stations Distributed battery hubs along road networks act as local storage and distribution nodes. They buffer energy between collection and demand, ensure no single point of failure, and handle the variable nature of solar input. Built with solid-state graphene batteries or next-generation nuclear batteries for maximum energy density and minimal self-discharge. Relay stations also serve as redundancy nodes — if any section fails, adjacent stations cover demand automatically.
Surface: transparent solar resin composite The visible driving surface. Transparent enough to pass usable light to the PV layer below, tough enough to handle heavy vehicle loads, textured for grip. Replaces conventional tarmac. See Section 1 for material science detail.
Inductive power track Continuous inductive coils deliver wireless power to vehicles in motion — the Scalextric principle applied to real roads. Vehicles draw energy continuously while driving. Onboard batteries become emergency backup, not the primary power source. A vehicle never needs to stop to charge. Existing EVs can be retrofitted with receiver coils. Conventional petrol vehicles remain fully functional throughout the transition.
This is the most mature pillar of the system, and it already exists. As of 2025, Electreon runs dynamic wireless charging on real roads in Detroit, on the Swedish island of Gotland (1.6 km, electric trucks and buses), and on France's A10 motorway, where a 1.5 km section delivers 200–300 kW to vehicles at highway speed. Measured transfer efficiency runs above 90% static and 85–88% in motion at 99.9% uptime — as good as plug-in fast charging, without the stop. Installation costs about US$2 million per mile (~$1.25M/km) per electrified lane. This is the technology SolarRoad is anchored to: an inductive road drawing from the existing grid already shrinks onboard batteries, removes charging stops, and cuts vehicle cost and weight, with or without any solar surface above it.
Maglev rail channels For high-speed lanes, magnetic levitation rails embedded in the road bed lift equipped vehicles fractionally from the surface, eliminating tyre-road friction. Higher safe speeds, dramatically reduced tyre wear, lower rolling resistance, improved energy efficiency.
Thermoelectric generator (TEG) modules Road surfaces absorb enormous solar heat. TEG modules convert the temperature differential between hot surface and cooler substrate directly into electricity — energy that would otherwise be radiated as waste heat.
Road de-icing In cold climates, resistive heating elements in the road surface — powered by stored solar energy — keep the surface above freezing. Roads that de-ice themselves. No salt, no grit, no black ice, no weather-related closures.
Inductive power transfer gives a vehicle energy. Maglev gives it freedom from friction. But neither gives the road direct control over the vehicle's motion.
Dynamic Propulsion Matrix adds a third capability: the road itself becomes a programmable linear motor, capable of pushing or pulling any equipped vehicle without mechanical contact.
The roadbed contains a continuous array of embedded electromagnetic coils — a programmable stator running the length of every lane. Vehicles carry a controllable, gyro-stabilised magnet rotor mounted on the underside.
By switching the polarity of road coils ahead of or behind a vehicle, the AI exerts precise push (acceleration) or pull (braking/repulsion) forces directly on the rotor. The gyro stabilisation ensures the rotor maintains correct orientation regardless of vehicle pitch, roll, or road camber.
| Mode | Polarity sequence | Effect |
|---|---|---|
| Acceleration | Coils ahead pull rotor forward | Propels vehicle without using its own motor |
| Braking | Coils ahead push rotor backward | Decelerates vehicle, recovers energy |
| Hold/Spacing | Alternating push-pull | Maintains millimetre-accurate following distance |
| Emergency stop | Full reverse field | Physically halts vehicle faster than friction brakes |
The original blueprint assumes vehicles are responsible for their own motion, with the road only providing energy and guidance. That works — but it leaves a gap. Human reaction time, varying brake performance, and tyre-road friction limits still create risk.
Dynamic Propulsion Matrix closes that gap. The AI doesn't just suggest a speed or route around congestion. It enforces motion directly at the physical layer.
Key benefits:
- Tyre-less acceleration and braking — vehicle tyres become passive rollers. No skidding, no ABS hunting, no brake pad wear. Stopping distance is limited by human comfort, not friction coefficient.
- Regenerative braking at grid scale — braking energy flows directly back into the superconducting backbone (Layer 4), not into a vehicle battery that may be full. Round-trip efficiency exceeds 90%.
- Millimetre-accurate platooning — AI can space vehicles 0.5 metres apart at highway speeds. No human could react that fast. The road handles it continuously.
- Emergency response override — when an ambulance approaches, the AI can gently push surrounding vehicles apart and hold them in a moving corridor. No horn, no erratic swerving.
- Debris rejection — the gyro rotor is only energised when needed. When idle, it has no net magnetic field, so it doesn't attract ferrous road debris. A brief reverse-polarity pulse can actively eject anything that does stick.
The gutter becomes a multi-purpose infrastructure conduit running the full length of every road. A fibre-optic backbone carries real-time data between vehicles, sensors, relay stations, and the central AI. Embedded sensor arrays monitor traffic, structural integrity, temperature, weather, and emergencies, doubling as system redundancy. The same channel harvests rainwater for urban supply and protects the utility runs (power, communications, and thermal management) in a single conduit. Along the kerb, pavement-side inductive strips charge e-bikes, e-scooters, and mobility devices, while safety sensors at junctions watch for pedestrians and cyclists.
Superconducting power cables in vacuum chambers Conventional cables lose energy as heat through resistance. Superconducting cables at low temperatures have near-zero resistance — energy travels from collection to storage to end-use with minimal loss. Housed in insulated vacuum chambers 2–4 metres beneath the road, forming the zero-loss energy backbone of the network.
Direct energy storage nodes Underground storage at distributed intervals. Design principle: minimise conversion steps — each conversion loses energy. Solar electricity goes directly to electrical storage and is delivered back as electricity. Graphene supercapacitors for fast-response buffering. Solid-state batteries for day-scale storage.
AI central traffic management A city-wide AI system manages all vehicle routing in real time. Traffic flow becomes a mathematical optimisation problem — not chaotic emergent behaviour. Traffic lights become unnecessary. Congestion is eliminated by design. In major cities, the efficiency gain alone represents a significant reduction in wasted energy, time, and pollution from idling.
Emergency vehicle priority Ambulances, fire engines, and police receive absolute routing priority. A clear corridor opens ahead of an emergency vehicle in real time across the entire network. Response times improve. Lives are saved directly.
Zero road deaths Human error causes over 90% of road accidents. Remove human error through AI routing, maglev guidance, and vehicle-to-vehicle communication, and road deaths approach zero. Globally, 1.35 million lives per year. It may be the system's single largest humanitarian benefit.
Autonomous and driver-assisted vehicles Fully autonomous on SolarRoad infrastructure, or driver-assisted with human override at all times. Vehicles can leave the circuit and travel on conventional roads normally. The system is opt-in by design throughout the transition.
Vehicle-to-grid energy balancing Parked vehicles feed surplus battery energy back into the road network and into homes. The entire vehicle fleet becomes a distributed storage grid — a geographically dispersed battery stabilising supply and demand around the clock.
Noise reduction Maglev guidance and smooth composite surfaces eliminate the majority of tyre-road and mechanical noise. Cities become significantly quieter.
The transparent road surface is the most novel and critical component. Everything else — maglev, inductive charging, AI routing, superconductors — exists in deployed forms today. The solar substrate is what makes this genuinely new.
The core trade-off Transparency and solar efficiency pull against each other, and closing that gap is the central challenge. The 2025 figures set the realistic bar: a perovskite–organic tandem transparent cell now reaches 12.3% efficiency at 30% transparency, and hybrid silicon solar windows reach about 8.3% efficiency at 75% visible transmittance, against 20–22% for conventional opaque panels. So the "15–20% at >40% transmittance" the substrate stack below assumes is a target for the next few years, not a product you can buy today. A road surface built right now would sit closer to 8–12% — which is why the Sydney figures above derate the road-solar contribution.
Road-specific requirements
| Property | Requirement |
|---|---|
| Compressive strength | ≥ 10 MPa |
| Skid resistance (wet) | Friction coefficient ≥ 0.45 |
| Solar transmittance | ≥ 40% to PV layer |
| Temperature range | −40°C to +85°C without delamination |
| Service life | ≥ 20 years under traffic loading |
| Surface hardness | Vickers hardness ≥ 600 HV |
Leading PV approaches
Perovskite solar cells — high efficiency potential, rapidly improving stability, manufacturable as thin films. Lab efficiency records now exceed 25%.
Organic photovoltaics (OPV) — flexible, lightweight, printable onto substrates. Lower efficiency but highly compatible with resin composite embedding.
Quantum dot PV — tuneable absorption spectrum, high efficiency potential at high transparency, early-stage but promising for the specific balance required.
Design principle: minimise conversion steps, maximise round-trip efficiency, maximise energy density, maximise cycle life.
| Property | Value |
|---|---|
| Round-trip efficiency | ~95–98% |
| Energy density | ~10–30 Wh/kg (rapidly improving) |
| Cycle life | >1,000,000 cycles |
| Charge/discharge speed | Seconds to minutes |
| Self-discharge | Moderate (days to weeks) |
Best for: short-term buffer storage at relay stations, vehicle-to-grid peak balancing, high-frequency charge/discharge applications.
| Property | Value |
|---|---|
| Round-trip efficiency | ~90–95% |
| Energy density | ~400–500 Wh/kg (projected near-term) |
| Cycle life | ~5,000–10,000 cycles |
| Charge/discharge speed | Minutes to hours |
| Self-discharge | Very low (months to years) |
Best for: long-term strategic storage, relay stations requiring days of stored capacity, vehicle onboard batteries. No liquid electrolyte — no fire risk.
| Property | Value |
|---|---|
| Output type | Low, continuous, ultra-reliable |
| Service life | 10–50 years depending on isotope |
| Self-discharge | None |
Best for: always-on sensor arrays, superconducting vacuum chamber maintenance, AI traffic node backup power.
| Layer | Technology | Purpose |
|---|---|---|
| Relay stations — fast response | Graphene supercapacitors | Immediate demand buffering |
| Relay stations — day storage | Solid-state batteries | 24-hour supply coverage |
| Infrastructure nodes | Nuclear batteries | Always-on sensor and AI power |
| Grid-scale bulk storage | SMES underground | Large-scale balancing, zero loss |
| Vehicle onboard | Solid-state batteries | Range extension and V2G |
The substrate is where solar collection, thermal harvesting, structural strength, and electrical generation all have to coexist in a single tile. Getting the material stack right is the engineering heart of the entire system.
Perovskite PV cells capture energy from visible light. Directly beneath them, thermoelectric generator (TEG) modules capture the heat that the PV cells themselves generate. Combined theoretical yield from a single surface: 20–28% — significantly better than either technology deployed alone.
| Layer | Material | Function |
|---|---|---|
| 1 — Wear surface | Alumina-reinforced polymer (8–12mm) | Load-bearing, transparent, textured for grip |
| 2 — PV layer | Perovskite thin-film cells | Visible light → electricity (15–20%) |
| 3 — Encapsulant | Transparent polymer resin | Protects PV, passes IR heat downward |
| 4 — TEG modules | Bismuth telluride (Bi₂Te₃) | Heat differential → electricity (5–8%) |
| 5 — Heat spreader | Graphene sheet | Even heat distribution + electrical collection |
| 6 — Insulator | Aerogel composite | Maintains TEG temperature differential |
| 7 — Bond layer | Flexible graphene adhesive | Retrofit bond to existing tarmac |
| Mechanism | Source | Lab/target efficiency | Realistic field efficiency |
|---|---|---|---|
| Perovskite PV | Visible light | 15–20% | 8–12% (transparent, 2025) |
| Bismuth telluride TEG | Road surface heat | 5–8% | 1–2% |
| Combined system | Solar + thermal | 20–28% (ceiling) | ~10–14% (today) |
Conventional tarmac: 0%
A note on the TEG layer. "Harvest the same sunlight twice" is a real and elegant principle, but worth sizing honestly. Lab TEGs reach about 11% conversion; on a real road they deliver a few percent, because the temperature gap between surface and substrate is modest (often 20–40°C). So in the near term the TEG layer is a bonus rather than a main supply — it pulls some electricity from heat that's otherwise wasted, and the same hardware doubles as thermal management and self-de-icing. The bigger conversion gains come as thermoelectric materials improve. The combined figure to quote today is around 10–14%, with 20–28% the long-horizon ceiling.
This is a generational infrastructure upgrade — the same category of change as electrifying railways, rolling out the internet, or the transition from analogue to digital. Each of those transitions seemed impossibly large before they happened and inevitable in retrospect.
Phase 1 — 100m test road segment. Solar composite surface, inductive charging, embedded sensors.
Phase 2 — 1km urban corridor. Full gutter conduit and underground superconducting backbone.
Phase 3 — Relay station network at 500m intervals. Day/night storage cycle validated.
Phase 4 — Autonomous vehicle integration. AI fleet management, maglev guidance, emergency priority routing.
Phase 5 — Metropolitan pilot. 500m city road plus one high-rise with solar PV glass.
Phase 6 onwards — City-by-city, road-by-road expansion. Existing EVs retrofitted. No forced obsolescence.
A proposal that names a benefit should name a price. These are order-of-magnitude figures from real projects, so the costs can be argued with rather than assumed.
| Item | Evidence-based cost | Source |
|---|---|---|
| Dynamic inductive charging lane | ~US$1.25M / km (single lane) | Electreon Detroit (~$2M/mile) |
| Solar PV pavement | ~$39,000 / kW installed | Solar Roadways Idaho pilot |
| Conventional ground-mount solar | ~$800–1,100 / kW | 2025 utility-scale benchmark |
| Wattway-class PV road | €5M / km | Colas, France |
Two things follow. The inductive layer is roughly cost-competitive with other electrification infrastructure — cheaper per km than the overhead catenary used for trolleybuses, per Volvo's eRoad Arlanda findings. The solar driving surface is not yet economic; at today's prices it costs 10–40 times more per watt than a field of panels. That gap is the financial case for the design above: spend on the inductive road and on cheap roadside and facade solar now, and add the transparent PV surface where it's structurally cheap (new builds and premium corridors) while its cost comes down.
The phased rollout follows the same logic. Phases 1–4 prove the inductive, AI, and storage backbone, the parts that already work and pay back through fuel and battery savings. The expensive solar-surface capital comes last, once transparent-PV prices have fallen — the opposite of Wattway, which spent it first.
Big infrastructure projects work best when their hardest problems are named in advance. These are SolarRoad's, set out plainly so the engineering can aim straight at them. None is a dead end.
- Transparent solar pavement (TRL 3–4) is the boldest piece, and the design is built so it doesn't depend on it. The system delivers while PV prices fall, and the road's solar contribution grows as they do.
- Thermoelectric harvesting is a bonus today and a larger one as thermoelectric materials improve. In the meantime it pays its way as thermal management and self-de-icing.
- Road-scale maglev is the longest-horizon piece and is optional. Inductive charging and AI control already provide most of its safety and efficiency benefits, so maglev can come later as a high-speed-corridor upgrade.
- The Dynamic Propulsion Matrix is the most ambitious idea here, and the sensible path is a closed test loop first — the way maglev trains and linear motors were proven before they carried anyone.
- Superconducting backbone needs cryogenic cooling, though high-temperature superconductors keep shrinking that cost. Early phases can use conventional HVDC or MVAC and save superconductors for high-density trunk routes.
- Grid intermittency is the challenge shared by all solar, handled here by the hybrid storage stack and a fleet-wide V2G battery.
The strategy is straightforward: lead with the proven pillars (inductive charging, AI traffic management, roadside and facade solar, modern storage), which are worth building today, and let solar pavement, maglev, and the propulsion matrix mature on top of them over time.
| Factor | Current system | SolarRoad |
|---|---|---|
| Road transport CO₂ (global) | ~8 billion tonnes/year | Zero |
| Road deaths (global) | ~1,350,000/year | Approaching zero |
| Road injuries (global) | ~50,000,000/year | Approaching zero |
| Grid transmission loss | 8–15% average | <1% (superconducting) |
| Road surface heat waste | 100% wasted | Harvested via TEG |
| Urban noise pollution | Significant | Dramatically reduced |
| City electricity (Sydney model) | Grid-dependent | 84–137% self-generated |
| Road maintenance | Reactive, frequent | Predictive, extended lifecycle |
All component technologies exist in some form, and none requires new physics. But they sit at very different maturity levels, and integration is its own hard problem (see The Frontier Challenges). "No new physics" is not the same as "ready tomorrow" — it means the remaining barriers are engineering, not discovery.
| Component | TRL (1–9) | Primary challenge |
|---|---|---|
| Transparent PV road surface | 3–4 | Efficiency + durability under load |
| Inductive road charging (moving vehicles) | 5–6 | Standardisation + cost |
| Embedded maglev (road-scale) | 4–5 | Integration with road formats |
| Superconducting underground grid | 6–7 | Cooling infrastructure cost |
| Solid-state batteries | 6–7 | Manufacturing at scale |
| Graphene supercapacitors | 5–6 | Energy density improvement |
| Autonomous vehicle AI | 7–8 | Regulatory frameworks |
| Transparent PV building glass | 5–6 | Efficiency at high transparency |
TRL = Technology Readiness Level. 9 = fully deployed commercial technology.
Sources for the field data, costs, and yield figures used above:
Prior solar-road projects
- France's failed solar roadway (Wattway) — Big Think · ScienceAlert
- Solar Roadways Inc — Wikipedia · Interesting Engineering
- China / Jinan solar expressway — NS Energy · pv magazine, China highway solar 2025
Dynamic wireless (inductive) charging
- Electreon France A10, 200–300 kW in motion — Electrek · New Atlas
- Detroit pilot, cost & efficiency — Interesting Engineering
Transparent photovoltaics (2025)
- Transparent tandem cell, 12.3% at 30% transparency — pv magazine
- Hybrid solar window, 8.3% at 75.6% AVT — ScienceDirect
Road thermoelectric (TEG) harvesting
- Asphalt-pavement TEG field studies & real-world efficiency limits — ScienceDirect review · J. Electronic Materials field study
The figures in this document are order-of-magnitude estimates meant for discussion, not engineering guarantees. The sources are here so any claim can be checked.
Released under Creative Commons Attribution 4.0 International (CC BY 4.0).