The Gap Between the Record and the Roll
This decade is supposed to witness a couple of technologies shaping the energy system. Both have already been winners on paper. Neither have thus far won on a production line and why is the real story
At a conference held at Shanghai Jiao Tong University on 14 July 2026, LONGi presented the result of a two-terminal crystalline silicon-perovskite tandem solar cell converting 35.5 per cent of sunlight falling on it into electricity.[1] The outcome was independently certified by the European Solar Test Installation of the European Commission’s Joint Research Centre, in Italy.[2]
It’s an incredible number and it’s worth grasping right. The Shockley-Queisser limit for a single-junction silicon cell, which was set in 1961 by Shockley and Queisser, is based on the thermodynamics of absorbing a wide solar spectrum with a single bandgap.[3] When the photons have an energy less than the bandgap, they pass through. Photons with higher energy are absorbed, with the extra energy dissipated as heat in these picoseconds. A combination of stacking a wide-bandgap perovskite on top of silicon allows the perovskite to absorb the blue and green photons’ energy almost in full, while the silicon absorbs the red and infrared. The theoretical limit of this architecture is approximately 43 per cent.[1]
Therefore, the record counts! It’s also for those who are trying to calculate the when of this technology to make electricity more affordable, virtually useless.
No details of the detailed electrical characteristics of the record device were provided, and LONGi didn’t reveal the number of active devices in the module.[2] The figure refers to a laboratory cell which is certified, not a module and not a product. But the gap between them is where the whole story resides.
Nine percentage points of engineering
Compare against what is available for purchase.
Fraunhofer CalLab has certified 26.9 per cent as the record in a residential format, held by Oxford PV.[4] In September 2024, the company shipped its first commercial perovskite tandem modules at 24.5 per cent efficiency to a solar energy project in the United States, built in its 100 megawatts plant in Brandenburg an der Havel, a former Bosch thin-film facility.[5][6][7] In 2026, commercially available tandem modules range from about 24 to 29 per cent, with 30 per cent expected in 2027 and/or 2028.[8]
Thirty-five and a half in a laboratory. Twenty-four to twenty-nine on a pallet. That interval is not hype and it is not fraud. It is a stack of four specific engineering problems, and each one is worth understanding on its own terms.
Thirty five and a half in a lab. 24-29 on a pallet. It’s not hype and it is not fraud. It is a series of four distinct engineering challenges, and each is worthy of its own study.
Area. A record cell is typically about a square centimetre. In the M10 format, a production wafer measures 330 square centimetres and a module is even bigger, an additional two orders of magnitude.[9] The simple reason why efficiency doesn’t scale is that defects, non-uniformities and resistive losses all increase with area faster. Additional square centimetres mean additional pinholes!
Deposition. Research perovskite films are made by spin-coating, in which a droplet of precursor solution is spread by spinning the substrate. It makes a gorgeous film and beyond a few centimetres it’s a hopeless mess. The industry has recently turned to slot-die coating, blade coating and vacuum thermal evaporation.[10][11] That gap is a manageable one, and much of the engineering effort of the industry today is invested in closing that gap.
Yield. This is the number that nobody publishes. No one has published yield rates for Oxford PV at Brandenburg. Yield is the gap between a technology that is available and a technology that costs money. One line runs 30 per cent cells at 60 per cent, that is more expensive per watt than a line that is running 24 per cent cells at 95 per cent and there is no press release available that will tell you that.
Degradation. The industry has 40 years of field data and a set of well-established silicon accelerated tests that correlate with field performance and warranties of 25 years. Perovskites fail differently. The ions move in the presence of an electrical gradient. Under illumination, mixed-halide compositions separate and consequently the bandgap of the material changes during operation.[12] Moisture is aggressive. All of these mechanisms are not what the IEC accelerated test sequences have been designed to find, as the sequences have been implemented for crystalline silicon and currently there is no harmonised accelerated test standard for perovskite-silicon tandems.[13]
The last one might be the most important, and it’s the least flashy on the list, so it’s important to emphasize this point. A module not tested to a recognised standard will not be underwritten. Without underwriting, there isn’t a utility-scale financing option. Document may become the binding restriction for a technology that’s 35.5 per cent.
In the meantime, the old truth is that no one knows at this time about lifelong. Hanwha Qcells has announced a 31.6 per cent, while certified 30.8 per cent on industrial Q.ANTUM silicon bottom cells, with 95 per cent efficiency after 1,000 hours of MPP tracking at 25 degrees, with ANTUM silicon bottom cells.[14] It’s a solid stability measurement and a 42 days. There will be a delay of 2027 or 2028 to obtain statistically meaningful degradation curves for the modules deployed in 2024.
Same trend in dollars
The same structure also applies to sodium-ion batteries, except that it is efficiency that is the limiting factor here.
The physical case is simple. Unlike lithium, which is found on only a handful of continents, sodium is everywhere, and can be extracted from seawater and salt deposits on every continent, and is not held up by geography in shaping its politics. The chemistry, on the other hand, is said to have much greater cold tolerance: cells made in large numbers can discharge more than 90 per cent of their capacity at minus 40 degrees, and have a discharge power more than triple that of an equivalent lithium iron phosphate cell at minus 30 degrees.[15] That isn’t a small plus for continental winter storage.
This achievement gap has decreased significantly. CATL quotes Naxtra cells at about 175 Wh/kg in mass production, while lithium iron phosphate is in the 200-205 Wh/kg range.[16] The volumetric energy density is still bad, and that’s important for vehicles, but not for a substation pad container. CATL and Changan announced its first mass-production sodium-ion passenger vehicle in February 2026.[17] BYD has been working towards mass production of tens of gigawatt-hours of sodium.[18] The global sodium-ion capacity is over 95 per cent Chinese.[19]
There’s the price number, too, and with that one the discipline of a reader is significant.
Its manufacturer quoted about 19 dollars per kWh at the cell level for 2026, and independent analysis quoted it at 70-120 dollars per kWh, with second-generation Naxtra cells quoted at around 77 dollars.[20] The Chinese lithium iron phosphate cells are in the area of about 70 to 80 dollars.[18]
There are arguments for and against both numbers. The low number refers to a Bill of Material based on assumed volume, in which cheap sodium precursors have been used to replace lithium carbonate. The high figure is what a cell really costs to produce and sell today, and hard carbon anode material is still more costly than lithium production, while production volumes are still less than lithium.[19] This is the distance that lies between a chemistry and an industry.
It’s not even sodium, it’s its progress. But when lithium carbonate prices plummeted following their peak in 2022, so too did the economics behind sodium, and manufacturers quietly dropped their programmes even though the technology was working.[21] The sodium programmes returned when Chinese authorities halted some of the mining licences and lithium prices rose.[22][16] The commercial fate of a technology that is based on independence from a volatile commodity is determined by the price of that commodity.
The losses are illustrative. Natron Energy built sodium-ion cells in Michigan (with a claimed cycle life over 50,000), but then the factory was shut down and plans were abandoned on the factory were worth 1.4 billion dollars.[23][24] The product worked. The business failed to.
For those reading a press release, what this means
So, there are three practical lessons to be drawn and they are applicable to more than these two technologies.
First, certified records are no longer a good predictor of deployment. They are true, incontrovertible, scientifically significant and address the question of physical possibility, not industrial reality. Any record is answerable to a few questions that are often useful: What area was it measured over, and what is the best full size module that the same organisation does?
Second, no standards are now a greater obstacle than the science that still remains. This is a problem that can be solved, but not by the brilliance of the laboratory, it requires the work of the committee, a committee that is not assembled by the laboratory but by the IEC and the certification bodies, and that has its own time.
Third, the benefit is starting to accrue to those who can maintain yield on a large scale, and that’s a manufacturing skill not a research skill. Germany has the pilot infrastructure. South Korea and the United States are developing the next generation. China has the scale to be the number one producer in the future, after the pilot phase of single-junction silicon, and when they become the number one producer, capacity will be overwhelming, as it was in silicon. But the perovskite race could be won the same way as the previous one was won, and if so, the laboratory records would have been a good precursor to nothing.
The honest summary
So in my opinion, all this does not negate either technology. Perovskite-silicon tandems are the most obvious way to break a physical wall which has limited solar since its inception and any additional kilowatt from the same square metre of land is invaluable in such markets. Sodium-ion is already a proven, safe, cold-usable storage option for stationary applications, where the grid really needs cheap storage solutions.
What it is against is a mindset. The energy transition is often told in terms of breakthroughs, as breakthroughs are easier to read, and are dateable. What really dictates the timing of a technology are yields that companies are not willing to release, four-year degradation effects to be seen, standards that aren’t available yet, and the price of another commodity.
Thirty-five and a half per cent is an excellent bit of physics. There’s one number that hasn’t been disclosed that will mean the difference between it being significant and not.
References
[1] LONGi, “35.5%! LONGi Once Again Breaks World Record for Crystalline Silicon-Perovskite Tandem Solar Cell Efficiency”, 14 July 2026. https://www.longi.com/en/news/crystalline-silicon-perovskite-tandem-solar-cell-new-world-efficiency-2026/
[2] pv magazine, “Longi sets new world record with 35.5%-efficient perovskite-silicon tandem cell”, 15 July 2026. https://www.pv-magazine.com/2026/07/15/longi-sets-new-world-record-with-35-5-efficient-perovskite-silicon-tandem-cell/
[3] W. Shockley and H. J. Queisser, “Detailed Balance Limit of Efficiency of p-n Junction Solar Cells”, Journal of Applied Physics, vol. 32, no. 3, pp. 510-519, 1961. https://doi.org/10.1063/1.1736034
[4] PV Tech, “Oxford PV unveils record 26.9% perovskite tandem module”, 19 June 2024. https://www.pv-tech.org/oxford-pv-unveils-record-26-9-perovskite-tandem-module/
[5] pv magazine, “Oxford PV starts commercial distribution of perovskite solar modules”, 5 September 2024. https://www.pv-magazine.com/2024/09/05/oxford-pv-starts-commercial-distribution-of-perovskite-solar-modules/
[6] pv magazine, “Oxford PV completes 100 MW factory build out”, 23 July 2021. https://www.pv-magazine.com/2021/07/23/oxford-pv-completes-100-mw-factory-build-out/
[7] PV Tech, “Oxford PV to build perovskite pilot line at old Bosch site”, 11 November 2016. https://www.pv-tech.org/oxford-pv-to-build-perovskite-pilot-line-at-old-bosch-site/
[8] Green Fuel Journal, “Perovskite Solar Cell Revolution 2026: Efficiency Records, Stability Breakthroughs & Market Growth Explained – Part I”, 2026. https://www.greenfueljournal.com/post/perovskite-solar-cell-revolution-2026-part-i
[9] Hanwha Qcells, “Hanwha Qcells achieves world record efficiency for commercially scalable perovskite-silicon tandem solar cell”, 19 December 2024. https://www.hanwha.com/newsroom/news/press-releases/hanwha-qcells-achieves-world-record-efficiency-for-commercially-scalable-perovskite-silicon-tandem-solar-cell.do
[10] R. Swartwout, M. T. Hoerantner and V. Bulovic, “Scalable Deposition Methods for Large-area Production of Perovskite Thin Films”, Energy & Environmental Materials, vol. 2, no. 2, pp. 119-145, 2019. https://onlinelibrary.wiley.com/doi/full/10.1002/eem2.12043
[11] Energies, “Recent Advancements on Slot-Die Coating of Perovskite Solar Cells: The Lab-to-Fab Optimisation Process”, vol. 17, no. 16, 3896, 2024. https://doi.org/10.3390/en17163896
[12] National Renewable Energy Laboratory, “Stability and Reliability of Perovskite Containing Solar Cells and Modules”, NREL/JA-5900-91506, 2025. https://docs.nrel.gov/docs/fy25osti/91506.pdf
[13] PatSnap, “Perovskite-silicon tandem solar cells hit 34% in 2026”, 2026. https://www.patsnap.com/resources/blog/articles/perovskite-silicon-tandem-solar-cells-hit-34-in-2026/
[14] Nature Communications, “Interfacial design strategies for stable and high-performance perovskite/silicon tandem solar cells on industrial silicon cells”, 2025. https://www.nature.com/articles/s41467-025-64467-y
[15] R&D World, “175 Wh/kg: How CATL’s Naxtra architecture moves sodium-ion beyond the lab”, 6 March 2026. https://www.rdworldonline.com/175-wh-kg-how-catls-naxtra-architecture-moves-sodium-ion-beyond-the-lab/
[16] International Energy Agency, “Sodium-ion battery momentum grows, but challenges remain”, IEA Commentary. https://www.iea.org/commentaries/sodium-ion-battery-momentum-grows-but-challenges-remain
[17] InsideEVs, “The World’s First Sodium-Ion Battery EV Is A Winter Range Monster”, 5 February 2026. https://insideevs.com/news/786509/catl-changan-worlds-first-sodium-ion-battery-ev/
[18] Henan Zhuowei New Energy Technology, “Sodium Ion vs LFP Battery for Energy Storage: A 2026 Data-Driven Comparison”, 2026. https://www.zvepow.com/news/sodium-ion-vs-lfp-battery
[19] DataDeep, “Sodium-Ion Batteries in 2026: Cost Reality vs. LFP, CATL’s Naxtra, and the Hard Carbon Bottleneck”, 3 July 2026. https://datadeep.tech/sodium-ion-batteries-2026/
[20] Electrek, “CATL’s 30-year sodium-ion battery takes over grid storage”, 16 July 2026. https://electrek.co/2026/07/16/catl-sodium-ion-15000-cycle-grid-storage/
[21] Tianxia Gongchang Research, “Sodium-Ion Battery 2026: How China Built the Industry’s Second Card Alone After Lithium Crashed”, 2026. https://faxiangongchang.com/en/reports/sodium-ion-battery-2026-china-supply-chain
[22] Trading Economics, “Lithium Extends Plunge from 1-Year High” (CATL Jiangxi mining permit suspension), 10 September 2025. https://tradingeconomics.com/commodity/lithium/news/484309
[23] Natron Energy, “Sodium-Ion Batteries & Sustainable Energy” (over 50,000 cycles). https://natron.energy/
[24] Manufacturing Dive, “Sodium-ion battery maker Natron Energy shuts down, halts $1.4B factory plans”, 8 September 2025. https://www.manufacturingdive.com/news/sodium-ion-battery-natron-energy-shutters-halts-NC-factory-plans/759479/
