The material that keeps disappointing on rooftops may get its first real business indoors.
RadianX, a two-year-old startup in Redwood City, California, raised $9 million this month to build a factory for solar panels meant for a warehouse sensor or a shipping label, where a roof panel would be absurd. The pitch is narrow: perovskite cells that keep working under the dim, artificial light of an office or a container, a job ordinary silicon has largely abandoned.
Look at a solar calculator. It has run on ambient light for forty years. The modern version of that trick is harder, because connected devices draw far more than a calculator ever did. It is also the one place where a small panel can beat the thing it replaces: a battery somebody has to change.
RadianX is not first to notice this. It is first to certify a product for it, and that distinction is what the $9 million is buying.
The bull case: power density where silicon quits
The panel that matters is 1.2 meters by 0.6, and RadianX says it converts 22.33% of incoming light into power. That is the company's own certified result, and it says no module of that size has scored higher. A smaller 300 by 300 millimetre version reached 24.24%.
Efficiency on a roof is half the argument. RadianX's real claim is about low light: 38% conversion under 500 lux, measured by a third party, and 50 microwatts per square centimetre at the same brightness. Silicon's output collapses in those conditions, which is why the indoor sensor market has waited.
RadianX AlphaGlass module, certified
Company-reported certified efficiency for its full-size module. A 300 x 300 mm unit reached 24.24%. RadianX / pv magazine, 2026
The engineering is deliberately unglamorous. The perovskite absorber is less than a micrometre thick, laid onto glass or film at low temperature, and the finished module can be as thin as 0.7 millimetres. Sizes and output voltages are custom, so a device maker can fit power into a product that already exists.
Durability was the old objection. RadianX says its modules are among the first to pass TÜV testing in all three categories of IEC TS 63163, the standard for solar built into consumer products, including the category for heavy outdoor exposure. It cites a projected life beyond ten years from accelerated tests, plus EU lead-release and RoHS compliance.
The commercial signals are early and real. AlphaGlass, the rigid line, has sold on DigiKey since September. RadianX is pairing modules with NGK's EnerCera batteries in wireless asset trackers and is in the 2026 Amazon Devices Climate Tech Accelerator. Gary Chen, former chief executive of the 250 MW producer Sunspark, joined as vice president of manufacturing.
As we wrote in November, the perovskite crowd is already having a funding moment: Noak's €5 million seed went toward tandem modules for utilities. RadianX is chasing a smaller, dimmer surface, and that is the more original bet.
The bear case: a 6 MW line is a rounding error
The first automated line will make 6 megawatts of panels a year. A single utility-scale solar farm is measured in hundreds of megawatts. By the standards of the industry RadianX is entering, six megawatts is a pilot with a press release attached.
Perovskite has been the next big thing in solar for a decade. It is still more expensive than silicon and less durable, even after years of improvement. RadianX argues that indoor duty avoids the worst wear, and that may hold. It also concedes that the product cannot yet survive the conditions its cheaper rival was built for.
The market it wants is real but thin. A sensor that never needs a battery is valuable to the operator who owns ten thousand of them. The same sensor is a rounding error to a device maker whose margin lives in a coin cell that costs a few cents. Adoption will be decided by procurement departments, not by physicists.
Then there is the size of the bet. Nine million dollars, five of it equity, is seed money. The company lists two employees. The panel is certified, but certified is a long way from shipped at volume, and a handful of design partners is not a market.
The wider funding weather is unhelpful too. Solar venture funding fell 31% year over year to $2 billion across 60 deals in the first nine months of 2026, according to Mercom. Cheap thin-film and better small batteries are improving at the same time RadianX needs buyers to switch.
What would confirm the bull case
+ Published field data shows life beyond five years in real fleets
+ A second automated line, not just the first
Confirmation criterion: repeat orders from customers who are not also investors or accelerator partners.
What would break it
− The 6 MW line stays a demo because unit economics never clear
− Perovskite stability fails in a hot, humid deployment
Refutation criterion: a mass-market device maker choosing an incumbent power source for cost, not performance.
The supply-chain argument, and its limits
The most underrated part of the pitch is not the efficiency number. It is geography. Perovskite is a nanomaterial that does not depend on the silicon supply chain, needs no rare earths, and can be made start to finish in a single facility, says chief executive Vivian Qu. A country without polysilicon plants can, in theory, print its own solar.
That matters more than it sounds. Device makers have spent three years pricing the risk of a concentrated silicon supply chain, and buyers in the Gulf and India are explicitly shopping for a second source. One of RadianX's investors, Qatar's quantumEDGE, frames the bet in exactly those terms: a silicon-independent power layer for sensor networks and industrial robots.
The Physical AI framing is the tell. A warehouse robot that harvests a little light between shifts needs less battery mass, and a lighter robot is a cheaper robot. RadianX is not selling panels. It is selling a smaller battery, sold as a package with the electronics tuned to perovskite's odd light-soaking behaviour.
The limit is that "can be made anywhere" is a claim about the recipe, not the factory. Thin-film deposition at high yield is the hard part of every solar scale-up, and a 6 MW line is where that difficulty gets exposed. Lead content adds a regulatory surface, even with REACH and RoHS compliance in hand. And a company with two employees cannot run a global supply chain and a device-integration business at the same time.
So the supply-chain story strengthens the bull case for where the technology could go. It does nothing to answer the nearer question of whether this company can build it.
Where the evidence actually lands
Two facts do most of the work. The module is certified at 22.33% on a real 1.2 by 0.6 metre area, and it is certified durable across the consumer-product standard, including outdoor exposure. Those are not lab cells in a paper. They are products a device maker can order today.
The third fact is harder. Nobody outside RadianX has published field data at scale. The ten-year life is a projection from accelerated testing, not a decade of operating history. The 38% indoor figure is a third-party measurement of a module, not of a finished device inside a customer's fleet.
That gap is the whole debate. The technology has cleared the laboratory. It has not yet cleared the spreadsheet.
A 6 MW line proves production, not demand
The winning question is procurement economics, not physics
Our read is skeptical without being dismissive. RadianX picked the one niche where perovskite's weaknesses matter least and its strengths matter most. That is a smart place to start. It is not evidence that the niche is large.
The honest gap sits in the buyer's spreadsheet. A sensor that never needs a battery saves labor, not hardware, and labor savings are the first line a procurement team cuts. Until a device maker publishes fleet data, the certification stands and the business case waits.
So the question worth watching is not whether a solar panel can run a sensor. It can. The question is whether enough buyers will pay a premium to stop changing batteries before cheap batteries make the problem disappear. RadianX has bought itself time to answer. It has not bought the answer.