Perovskite Solar Research: The Newest Results and What Is Still Unproven

Perovskite solar cells are having a loud few weeks. In the last ten days, research groups and companies have reported a string of efficiency records, new tandem designs and a satellite test in space. The tone of the coverage is celebratory. The science is more mixed.

What is new

The freshest dossier item is not a journal paper but a test mission. NTU Singapore launched the CRIMSON-1 satellite on 2 October to test next-generation perovskite solar cells and AI computing in orbit, according to EurekAlert!. The same week, pv magazine Global reported that an Italian startup had unveiled a double-layer PV system prototype for space-constrained applications, and that Trina Solar and Microquanta are testing perovskite and tandem modules at a UNSW site.

On the efficiency side, the numbers are real but need reading carefully. A perovskite-silicon tandem cell built on textured silicon reached 30.77%, pv magazine Global reported on 28 September. A three-layer cell reached 30.1% after smoother perovskite films were used, according to Tech Xplore, while another group reported 30% efficiency for the first time in late September. A Chinese solar cell hit 27.39% with a new molecular design, The Cool Down reported on 1 October.

Ember's 2025 solar growth estimate is twice that of international statistics: 12.0 GW in 2025, compared with 6.2 GW from the International Energy Agency (IEA) and 4.6 GW from the International Renewable Energy Agency (IRENA).

These are single-cell or small-module results, not commercial modules. The gap between a record efficiency in a lab and a product on a roof is where most perovskite coverage goes quiet.

The warning from the review

The most useful recent document is a landmark review warning that AI alone will not fix perovskite solar cells. It was covered by bioengineer.org on 1 October and summarised by several outlets. The argument is straightforward: machine learning can guide materials discovery and process optimisation, but it cannot substitute for solving the physical problems that limit perovskite stability and lifetime. That warning matters because much of the recent excitement has been driven by data-heavy methods. A separate result, reported by bioengineer.org on 30 September, described an AI-guided recipe for high-entropy perovskite materials. The same week, pv magazine Global reported a Prussian blue treatment that improved perovskite solar performance from small cells to larger modules, and a US team claimed a stability record for a 16.19%-efficient 2D/3D tin perovskite solar cell. Those are process and materials results, not AI results.

The stability question is not academic. Perovskite cells degrade under heat, moisture and light. A 30.77% tandem result is impressive, but if the module loses a meaningful share of its output in the first two years, the economics change. The review's point is that AI can accelerate the search for better materials, but the search still has to end in a cell that survives a rooftop summer.

Space and water, the new test beds

Two of the newest experiments push perovskite cells into environments where conventional silicon is awkward. CRIMSON-1 is testing perovskite cells in orbit, where the solar spectrum is different and thermal cycling is severe. In the South China Sea, scientists demonstrated solar power generation 10 metres underwater, euronews and Gadget Review reported in late September, with perovskite cells operating at that depth. Physics World covered polymer-tuned perovskite for underwater solar cells on 29 September.

These are early demonstrations, and none of the coverage in the dossier claims commercial readiness. They are useful because they stress the material in ways a lab bench does not. If a perovskite cell can survive a satellite launch or a month under water, that tells researchers something about encapsulation and mechanical robustness that a standard damp-heat test may miss.

The space angle also connects to a broader trend. Google launched its first datacenter satellite on 2 October, The Register reported, and is researching whether orbital compute can scale. NTU's CRIMSON-1 is a much smaller experiment, but it sits in the same conversation: moving power generation and computing off the ground requires solar hardware that works in space, not just in a field in Spain.

What the field data says

On the ground, the news is about field testing rather than records. UNSW received AUD 18M for a perovskite field test involving commercial modules, with a 2031 goal, according to IndexBox on 1 October. Trina Solar and Microquanta are testing perovskite and tandem modules at that site, pv magazine Global reported. The point of these tests is not to set a record. It is to find out how modules behave over years, under real weather, with real maintenance.

That is where the perovskite story will be decided. A 30.77% tandem cell is a scientific result. A module that holds 90% of its rated output after five years in the field is a product. The dossier does not yet contain that second result for perovskite, from any group or company.

The money and the market

Policy is moving in parallel. The US Department of Energy announced up to $400 million for basic research on 2 October, HPCwire reported, and the American-Made Challenges Perovskite Startup Prize remains open, according to the Department of Energy. In Europe, T&E analysis published on 30 September argued that the EU will have enough battery cells to meet demand by 2030, a reminder that clean-tech scale-up is a supply-chain problem as much as a materials problem.

Solar deployment is already large. Ember estimates that Chinese exports of solar panels to Africa surged to 23 GW in the 12 months to June 2026, a 53% rise on the previous 12 months, CleanTechnica reported on 2 October. Ember's 2025 solar growth estimate for Africa is 12.0 GW, twice the IEA's 6.2 GW and well above IRENA's 4.6 GW. That is a market signal: the world is installing a lot of silicon. Perovskite will have to compete on cost and lifetime, not just on a headline efficiency number.

For now, the honest summary is this. Perovskite research is advancing quickly, with new records, new test environments and new funding. The unresolved questions are stability, scale and field performance. The newest results are worth reporting. They are not yet a reason to expect a perovskite roof next year.