XtalPi and JinkoSolar Establish Joint Venture to Advance Photovoltaic Efficiency Limits with AI

Jan 08, 2026 08:57 CST Updated 08:57

XtalPi

Computation-Driven Innovative Drug R&D Provider

XtalPi Holdings ("XtalPi", "XtalPi")2228.HK) Recently announced the signing of a strategic cooperation agreement for the research and development of artificial intelligence (AI) + automation high-throughput tandem solar cells with a subsidiary of JinkoSolar Holding Co., Ltd. (688223.SH). Both parties will jointly establish a joint venture to build the world’s first fully closed-loop tandem cell intelligent manufacturing line featuring “AI decision-making - robotic execution - data feedback,” developing highly efficient and stable solar cell products for various application scenarios. This move marks the collaboration of two leaders in different technological fields, officially initiating deep synergy in next-generation photovoltaic technologies such as perovskite tandem cells. The joint laboratory will achieve breakthroughs by encoding key parameters such as material structure, formulation, process, characterization results, and device performance (tokenization), realizing an iterative cycle based on large language models (LLM) and multi-modal AI reasoning and evolution, reshaping the photovoltaic R&D paradigm, and accelerating the industrialization of disruptive technologies.

XtalPi:

The Game-Changer Empowering Material Innovation with AI

Since its establishment in 2015, XtalPi has consistently focused on the cross-innovation of AI and materials science, creating an industry-leading AI for Science R&D platform. By integrating quantum physics algorithms, AI predictive models, and large-scale robotic automated experiments, the platform achieves cross-scale innovation from microscopic mechanisms to macroscopic experiments, breaking through the barriers between virtual algorithms and real-world experiments. Its intelligent autonomous experimentation platform operates over 200 automated robots, boasting a globally leading scale, and continuously collects high-precision, industry-scarce R&D data at 40 times the efficiency of traditional laboratories. XtalPi has also developed more than 200 AI algorithms and domain-specific models that enable high-precision rational design, development, and process optimization of molecules, crystals, and polymers based on the required physical and chemical properties for specific functions. The platform serves 17 of the top 20 global pharmaceutical companies and has successfully transferred its deep technological expertise from drug discovery to new materials R&D, earning recognition from top enterprises and research institutions worldwide. This establishes its cross-industry foundational technology leadership.

In terms of algorithms, XtalPi has developed a dual-driven algorithm engine powered by quantum physics and AI, which can optimize perovskite materials and interfaces at the molecular and atomic levels. It predicts with high precision the band structure, carrier mobility, and interface defect states of materials, enabling the design of more efficient and stable material systems. Combined with large models in vertical fields, it can lock the optimal solution from hundreds of complex experimental parameters related to materials, compositions, and processes without the need for physical experiments.

XtalPi has also independently developed an automated robotic data infrastructure designed specifically for AI algorithm engines. Its high-throughput, modular intelligent production line can cover the entire chain process loop in photovoltaic material R&D, from precursor synthesis, device preparation, in-situ characterization, laser processing, vacuum evaporation to aging tests, replacing traditional manual experiments. It collects high-precision R&D data quickly 24/7 and provides real-time feedback to AI algorithms, forming an iterative closed loop.

In this collaboration, XtalPi will leverage its PatSight literature data mining platform along with robotic automated experimentation and in-situ characterization capabilities to create a dedicated structured experimental database for perovskite-silicon tandem solar cells. By integrating an AI R&D engine that combines quantum physics models with large language models specialized in vertical domains, a closed-loop workflow encompassing database-algorithm prediction-automated experimentation will be established. This approach aims to efficiently address key challenges in the R&D design of perovskite solar cells, such as data scarcity and non-structurization, vast experimental parameter variable spaces, and predictions of interfacial stability and material lifespan across layers.

Jinko Energy: Photovoltaic Technology Innovation

Pioneers and Leaders in Implementation

JinkoSolar, as a technology leader in the photovoltaic industry, excels in research and development and technological innovation. In recent years, the company has invested over 20 billion yuan in cumulative R&D, with a research team of nearly 2,000 people. It owns more than 5,500 patents, of which over 3,100 have been authorized, making it one of the companies with the highest number of patent applications and authorizations in the industry. It is particularly prominent in the N-type TOPCon technology field, holding 731 related patents and driving this technology to become the industry mainstream.

JinkoSolar's R&D infrastructure is well-established, having received numerous national-level qualification certifications, including the National Industrial Design Center, National Enterprise Technology Center, and National Intellectual Property Demonstration Enterprise. It also hosts 20 provincial research platforms, such as the key laboratory for stacked-layer technology. The company has recently achieved significant milestones in technological breakthroughs: the conversion efficiency of high-efficiency cells based on the TOPCon technology platform has reached 27.79%, while the conversion efficiency of N-type perovskite tandem cells has reached 34.76%, both breaking world records. This demonstrates JinkoSolar's deep accumulation and forward-looking layout in technology R&D.

The above technological breakthrough validates the compatibility and potential of TOPCon technology evolving into perovskite tandem stacking, drawing global tech attention, including from Elon Musk. JinkoEnergy, with its leading efficiency metrics and forward-looking technological layout, has solidified its leading position in photovoltaic technology innovation.

Powerful Combination:

AI Reconstructs the Next-Generation Stacked Battery R&D Paradigm

The R&D model that integrates AI and robotics will directly drive the maturation of perovskite tandem solar cell technology, thereby reshaping the imaginative boundaries of photovoltaic applications. In terrestrial fields, perovskite-silicon tandem modules are expected to significantly reduce the levelized cost of electricity (LCOE), improving system economics by approximately 15% compared to traditional technologies. More crucially, its potential for space applications: perovskite tandem solar cells, with advantages such as "high efficiency, low cost, lightweight, and flexibility," are more suited to the demands of space photovoltaics compared to other technological approaches, making them the optimal solution for space photovoltaics in the medium to long term. This lightweight, highly radiation-resistant energy system solution can significantly reduce the deployed area of satellite solar panels, freeing up more payload space for other critical components.

Despite the enormous technical potential, the industrialization of stacked-layer batteries still faces multiple challenges, making it difficult to simultaneously optimize photoelectric conversion efficiency, long-term stability, process consistency, and cost. Traditional research and development methods are time-consuming and highly discrete, while the implementation of AI technology is constrained by data scarcity and insufficient adaptability of general large models, hindering the efficient exploration and identification of optimal material formulations and processes.Art Combination.

XtalPi and JinkoSolar Collaborate to Build Next-Generation Photovoltaic LabFor the first time, encoding hundreds of perovskite material formulations with complex thin-film fabrication process parameters., to establish an exclusive perovskite-silicon tandem experimental database for the project. Based on key parameters such as material structure, formulation, process, characterization, and device performance in the database, XtalPi will develop a dedicated small model and a large language model-driven AI R&D engine. This engine will integrate multitask deep learning models with physics-informed machine learning models to achieve multimodal AI reasoning and evolutionary iteration loops. This AI engine will focus on critical tasks such as tandem structure design, material screening, electrical and optical performance prediction, and process parameter optimization.Continuously iterate to design superior perovskite tandem solar cells.

This collaboration will also buildThe First Thousand-Square-Meter-Level AI High-Throughput Stacked Solar Cell Experimental Line in the Industry，The experimental throughput is expected to reach 1,000 pieces/day, bringing a hundredfold increase in experimental throughput., driving the rapid iteration from algorithm design to experimental validation. The robotic workstation can automatically execute key steps such as precursor synthesis, thin-film preparation, and in-situ characterization. The generated experimental data is encoded and then fed back in real time to a dedicated database and AI engine, forming a closed-loop mechanism of "design-experiment-feedback-optimization" to complete material formulation and process design in a one-stop manner, significantly improving R&D efficiency.The high-performance perovskite-silicon tandem solar cells developed through this automated platform will have the potential to expand into various application scenarios, providing better continuous power supply support for high-density computing infrastructure, and accelerating the technological integration and industrial realization of next-generation photovoltaic technologies in extreme environments and high-reliability applications.

Industry Outlook:

The Innovative Revolution in the Era of Grid Parity for Photovoltaics

In this collaboration, XtalPi and JinkoSolar will adopt"Technology Introduction + Joint Development" Model, and Jointly Establish a Joint Venture, the intellectual property generated in the perovskite field will be commercialized by Jinko-led and XtalPi jointly. Based on the solid technical foundation and cooperation planning of both parties,JinkoEnergy Expects Perovskite Tandem Solar Cells to Move Towards Large-Scale Mass Production in About Three Years。

XtalPi Chairman Shuhaotem Wen pointed out: This is a collaboration of strengths, mutual benefit, and win-win results. It is also a classic closed-loop cooperation example where AI-driven material development supports space energy and space computing power, which in turn nurtures the evolution of AI. The potential application value of AI in future materials is limitless.

In today's era of AI empowering various industries, the foundation of technological development lies in energy. As a new generation of energy material, perovskite has the potential to significantly expand the application scenarios of photovoltaic cells.This collaboration represents not only a breakthrough in a single technology but also heralds a new era where AI and automated robotics replace the traditional "trial-and-error" approach, redefining the pace and boundaries of innovation across the photovoltaic industry and initiating a paradigm shift in PV research and development.XtalPi andThe collaboration with Jinko Power is writing the latest chapter of this transformation, promoting the deep integration of AI computing power and energy technology.