Solar’s origin story starts in 1954 at Bell Labs with a silicon cell running at roughly 5% efficiency. Its first major customer was a satellite. Sixty years later, the technology is the cheapest source of electricity in much of the world — and almost all of it is manufactured in China.
But the silicon cell that built this industry is running into a wall. Commercial cells today operate at 24–26% efficiency against a physics limit of about 29.5%. The world record sits near 27.8%. The trajectory that improved panels by roughly half a percentage point per year for decades is flattening.
According to Joel Jean, co-founder and CEO of Swift Solar, speaking on the Core Memory Podcast, the industry has exactly one credible answer for what comes next: stack a perovskite cell on top of a silicon cell to create a tandem that splits the solar spectrum into two pieces. The theoretical ceiling jumps from 30% to 45%.
“It’s really a unique time in the history of this company and also in the history of solar where we’re taking that first leap beyond the 30% potential of solar,” Jean said. “We better get building.”
The reason any single material stops at 30% comes down to a fundamental tradeoff between how much of the broadband solar spectrum you capture and how much energy you extract from each photon. Silicon does its job efficiently below about 1,100 nanometers. Longer wavelengths pass straight through and are lost. Pick a material with a lower band gap and you capture more of the spectrum — but you give up energy on every conversion.
A tandem stack solves this by assigning each cell a different slice of the spectrum. One absorbs visible light. The other handles infrared. The lab record for such a cell is 34.8%, against the 45% theoretical ceiling.
The efficiency numbers matter less than they seem, Jean argued. What actually determines whether solar gets built is levelized cost of electricity — dollars per kilowatt-hour, including labor, racking, wiring, transformers, and everything else spread over a project’s lifetime. Efficiency gains help. But the system costs are why panel prices in the US sit around 30 cents per watt even though Chinese domestic prices are far lower.

Perovskite — a crystal structure that can be synthesized from various elements — became solar’s most consequential word in 2012, when papers from Oxford and a team in Japan showed that the material could absorb light and transport charge at 10% efficiency on the first demonstration.
Jean puts that in personal perspective. He spent his MIT PhD making solar cells from quantum dots and organic materials. “I’ve made a lot of crappy solar cells in my lifetime and like none of them hit 10%,” he said. The 2012 perovskite results arrived “from day one” at a level his field couldn’t reach in five years.
What makes perovskite special is defect tolerance. Conventional semiconductors demand five-to-six-nines purity — every millionth atom in place — or energy escapes as heat. Perovskites tolerate missing atoms, moving ions, and general imperfection while still performing like a perfect crystal. They can be made at roughly 100°C instead of the 1,400°C silicon requires.
The tradeoff is vulnerability. Perovskites form at low temperatures and degrade at low temperatures. They are ionic semiconductors with moving parts. Unsurprisingly, the early hype — driven by university press offices amplifying lab demonstrations — collided with the harsh reality that panels are expected to survive for 20 to 30 years outdoors.
“Everyone and their mom can make a perovskite that’s like decently efficient and many people can make very efficient ones,” Jean said. “Almost no one in the world can make a stable perovskite.”
Then the caveat a CEO rarely volunteers: “I don’t think anyone in the world could say I can make a perovskite cell that can match silicon on durability right now.”
The stability problem is the entire ballgame. Every economic model for solar assumes decades of field life, and perovskite has never demonstrated that.
Jean’s counter is that every solar technology ever commercialized walked the same path. Cadmium telluride and CIGS thin films are 50-plus years old and showed identical early instabilities. Silicon itself needed a government-funded program run through JPL from 1975 to 1985 to push module lifetimes from roughly one year to more than ten.
“After this decade of development, we have shown a thousandx improvement in stability over the last few years,” Jean said. “We believe that it’s going to work and it’s a matter of scaling that up and proving it to the world.”
Swift’s strategic pivot, he said, was to stop chasing efficiency world records — after setting the first record for perovskite-on-silicon tandem cells — and pour everything into durability. He distinguishes two levels of “solved”: having enough data to have conviction the problem is solvable in reasonable time, versus having the data to back a 30-year warranty. Swift has the first, not the second.
“That’s true of every solar technology that’s ever been commercialized,” he said of the instability problem. “They’ve been around for 50 plus years and they took a lot of time. They showed these same kind of instabilities early on.”
The competitive picture is asymmetric. The US has on the order of five perovskite startups. Europe has a couple, with Oxford PV as the early leader. China has “a bunch” — and Chinese companies are already manufacturing perovskite cells at hundreds of megawatts to gigawatt scale, accepting degradation and iterating in the field. Jean believes continuous manufacturing improvement is not the same as the binary step changes sometimes needed to solve technical problems.
Swift’s biggest strategic move came with the acquisition of core assets from Meyer Burger, the leading European solar manufacturer that shut down after running into trouble on its Arizona factory and planned Colorado facility. Meyer Burger pioneered heterojunction technology — one of the two leading silicon approaches — and was the Western leader in it.
The logic is elegant. Heterojunction turns out to be the best bottom cell for a tandem architecture. By acquiring Meyer Burger’s assets and bringing on its former CEO and technical team, Swift gets a bankable, proven silicon platform it can commercialize at gigawatt scale in the US now — while maturing the perovskite layer that drops on top later.
The pricing picture stretches credulity: most of the world cannot compete with China on silicon manufacturing cost. But US panel prices, including Chinese and Southeast Asian imports, sit around 30 cents per watt, far above Chinese domestic levels. American buyers are already paying a premium for supply-chain security after seeing shipments blocked over forced-labor and tariff concerns.
Swift’s intellectual property position is unusual. Every company in the world that has set a perovskite world record has used Swift’s IP, developed in academic labs before the company was founded and licensed in. Swift patents selectively and keeps most process know-how as trade secrets, reasoning that the exact manufacturing method cannot be reverse-engineered from cross-sectional electron microscopy.
On the timeline, Jean is candid in a way that separates him from the standard founder narrative: “These kind of things always take more time and more money than anyone projects and that’s why all these startups fail. We have to be clear-eyed about that.” He puts gigawatt-scale competitiveness three to five years out.

The reference around solar’s ability to power AI data centers has changed dramatically. Jean is more bullish on the prospect than host Ashlee Vance expected. Hyperscalers — Google, AWS, Microsoft — are deploying large quantities of solar through power purchase agreements, often off-site rather than at the data center itself. Google acquired renewable developer Intersect. Jeff Dean, Google’s longtime AI leader, is a personal investor in Swift and shares the view that AI’s energy future is heavily solar.
Solar’s modularity helps. A facility doesn’t need to be a 10-by-10-mile array to be efficient; it can scale from a single panel to a utility farm without meaningfully losing performance. Solar costs have declined roughly 20% with every doubling of deployment, tracking the growth curve AI infrastructure is now riding.
The darker subplot is equipment access. In the week before the conversation, news emerged that the Chinese government was pressuring Chinese manufacturers not to sell leading-edge solar cell manufacturing equipment to the United States — reportedly triggered in part by Tesla’s announced $2.9 billion purchase of solar equipment for domestic manufacturing. Jean couldn’t confirm the details but treats the threat as real. Vance drew the parallel to wind: a Massachusetts company went from world leader to bankrupt in about 18 months after China obtained the source code to a windmill controller.
The weakness is systemic. Almost all generic solar factories in the US and elsewhere depend on Chinese manufacturing equipment. The same story applies to batteries, where China is completely dominant.
The episode’s most important calculation is a simple one, and Jean delivers it without drama: the cost of rebuilding a domestic solar industry, he argues, is a rounding error compared to what any single hyperscaler will spend on data centers next year. Whether anyone is actually prepared to make that bet at scale — rather than simply buying the cheapest panel available and signing a PPA — is the unresolved question.
References:
Once added, BigGo Finance appears first in Google Search Top Stories, so you get the broadest, most up-to-the-minute, and most comprehensive global financial news first.
The news and data on this website are for reference only and do not constitute investment advice or an offer to buy or sell. Information is sourced from exchanges and public sources, and may be delayed, interrupted, or updated. While we strive for accuracy, we do not guarantee timeliness, correctness, or completeness. Content may include external links for which we are not responsible. By using this site, you agree that we and our partners are not liable for any losses. Investment carries full responsibility; please carefully assess risks and consult professionals. If there are errors in the content, please contact us for correction.

source