Walk past any rooftop or field-scale solar project in America and you’re looking at essentially the same technology the industry has shipped for decades: a slab of silicon that turns sunlight into electricity. The panels got cheaper, the installations got bigger, but the fundamental architecture stayed the same.
That is now quietly changing inside a factory that opened this spring.
The shift involves a crystal compound most Americans have never heard of, bonded to the surface of silicon in a layer so thin it is almost invisible.
And the efficiency number it enables clears a limit that physicists set back in the 1960s.
Why every silicon solar panel eventually hits the same wall
Silicon solar cells are approaching their theoretical efficiency limit of around 29%. That ceiling is not a manufacturing problem. It is a physics problem.
A single-junction silicon cell can only convert photons whose energy falls within a specific band.
High-energy photons, the blue end of the spectrum, slam into the cell with more energy than silicon can use, and the surplus bleeds away instantly as heat.
That waste is baked into the physics of a single material trying to do the whole job alone.
This limitation can be exceeded with advanced device architectures, where two or more solar cells are stacked to improve the harvesting of solar energy.
What a 40-megawatt factory in Fremont just started building
Tandem PV announced the opening of its commercial demonstration factory in Fremont, California, a facility that marks a major step toward commercializing U.S.-manufactured next-generation solar panels for utility-scale projects.
The factory marks a transition for perovskite-silicon technology from laboratory development to repeatable manufacturing at scale.
The company has begun producing initial modules, with shipments planned to support customer validation trials later this year. It is targeting high-volume manufacturing in 2028.
It has received support from the U.S. Department of Energy and the California Energy Commission.
The 40-megawatt Fremont line is the first of its kind operating in the United States. But to understand why the industry is watching it closely, you need to look at what is actually inside each panel it builds.
The strange new supply chain behind the next panel boom
America’s solar module factories have expanded at a remarkable pace. The United States currently has more than 65 gigawatts of solar module manufacturing capacity, well above the country’s 2025 annual installation total.
Yet the industry faces a stubborn upstream gap, as domestic cell supply falls far short of what those assembly halls can consume.
Tandem PV’s perovskite layer is a thin photovoltaic coating applied to glass and layered atop a conventional silicon cell, requiring a fraction of the material of a standalone silicon wafer.
That manufacturing efficiency matters because raw material costs are already squeezing the broader panel boom.
Solar panel manufacturing is one of the fastest-growing consumers of silver globally, as each conventional cell relies on silver paste for its electrical contacts, a demand that grows with every new gigawatt of capacity. A panel that slots into existing silicon infrastructure while delivering higher output changes that equation.
The crystal layer that cleared a 1960s physics limit
Here is the mechanism the Fremont factory is betting on. Tandem PV’s proprietary technology combines a thin perovskite light-absorbing layer with a conventional silicon solar cell. By capturing more of the solar spectrum than silicon alone, tandem panels generate more electricity from the same footprint.
Each material handles the part of the spectrum it is actually suited for.
The tandem architecture captures a broader portion of the solar spectrum by stacking a wide-bandgap perovskite top cell over a silicon bottom cell, enabling energy extraction from photons that a single-junction cell would waste as heat.
The result in the lab is striking. The best-performing perovskite-silicon tandem cell has reached 35.5% efficiency, announced by LONGi in July 2026 and independently certified by the European Solar Test Installation (ESTI). That figure surpasses the Shockley-Queisser limit of 33.7% for any single-junction solar cell, the fundamental physics ceiling established in 1961. The previous benchmark, a 34.85% result also achieved by LONGi, was confirmed by the U.S. Department of Energy’s National Renewable Energy Laboratory in April 2025.
Tandem PV’s production panels have reached 29.7% efficiency based on internal testing, already approaching silicon’s single-junction theoretical ceiling and delivering meaningfully more power from the same area than today’s leading conventional modules, according to the company’s own data.
Increasing module efficiency and expanding manufacturing capacity play complementary roles in reducing costs of perovskite-silicon tandem solar modules, according to researchers at the National Renewable Energy Laboratory.
What has to happen before these panels reach your utility’s next project
The Fremont factory is a demonstration line, not a mass-production plant yet. The company plans to sell its first commercial panels in 2026 from this facility and is targeting high-volume manufacturing in 2028.
Durability is the remaining hurdle. Perovskites are crystal materials that can push solar efficiency beyond silicon’s single-junction limit and are cheaper to manufacture.
The problem has always been that they degrade when exposed to heat, moisture, and oxygen. In accelerated lifetime testing, the latest-generation panels show less than 1% average annual power loss, and Tandem PV is targeting 25+ year performance consistent with industry standards and warranty requirements for utility-scale solar projects. Solving that at commercial scale, across a 25-year panel lifetime, is the engineering task the Fremont factory is designed to test under real production conditions.
Tandem PV plans to have tandem solar panels ready for utility-scale customers in 2026, backed by ongoing R&D investment and government support.
As electricity demand from data centers and AI workloads continues to push utilities toward faster capacity additions, the pressure to move past silicon’s physics ceiling has never been more commercially urgent.
The crystal layer is thin enough to be almost invisible. The efficiency gain it unlocks is not.
