NVIDIA is driving scientific breakthroughs in materials discovery through accelerated computing, with new data processing pipelines and AI microservices unveiled at the SC25 conference in St. Louis. These innovations are helping researchers develop advanced materials for future technologies such as liquid-cooled data centers, high-resolution digital displays, and long-lasting batteries. A key highlight at the event is a demonstration by the U.S. Department of Energy’s Brookhaven National Laboratory, which uses the NVIDIA Holoscan AI sensor processing platform to achieve nanoscale imaging at resolutions under 10 nanometers. This capability is enabled by the platform’s ability to process streaming data in real time, allowing scientists to receive immediate feedback during experiments. Hanfei Yan, lead beamline scientist at Brookhaven’s Hard X-ray Nanoprobe, noted that this real-time insight allows researchers to identify regions of interest on the fly and observe material property changes during measurements—critical for making rapid, informed decisions. Daniel Allan, group leader of data engineering at NSLS-II, emphasized that faster processing also improves instrument efficiency, enabling more users and more scientific output. NVIDIA’s ALCHEMI suite, a collection of AI microservices and toolkits for chemistry and materials science, is also making a major impact. Two new microservices now available in NVIDIA NIM are designed for high-throughput batched conformer search and batched molecular dynamics simulations—key processes for predicting atomic-level material properties. These tools are being used by early adopters including ENEOS, a Japanese energy company, and Universal Display Corporation (UDC), a New Jersey-based OLED technology leader. ENEOS is leveraging ALCHEMI to discover new immersion cooling fluids for data centers and efficient catalysts for hydrogen production. Using the microservices, the team evaluated up to 10 million liquid candidates and 100 million oxygen evolution reaction candidates in just weeks—more than 10 times the scale of previous methods. Takeshi Ibuka, general manager of AI innovation at ENEOS Holdings, said the speed and scale of these calculations have transformed their research, allowing them to focus more on analysis and less on computation. At UDC, scientists are using ALCHEMI to accelerate the search for next-generation OLED materials. With over 10^100 possible molecules to consider, traditional methods were limited by computational capacity. Now, with GPU-accelerated conformer search, UDC can assess billions of candidates up to 10,000 times faster than before. Julie Brown, UDC’s chief technical officer, said the technology has removed previous bottlenecks, enabling broader exploration and faster discovery. Molecular dynamics simulations, accelerated up to 10x per run and further sped up through parallel GPU processing, now take seconds instead of days. The team is applying this to develop blue phosphorescent OLEDs that promise higher efficiency and performance. These advancements are part of NVIDIA’s broader effort to accelerate science through its CUDA-X ecosystem, which includes over 150 libraries and frameworks designed to solve complex real-world problems. The collaboration between NVIDIA and leading research institutions and companies is enabling faster, more efficient materials discovery—paving the way for more sustainable, high-performance technologies across industries.