Nvidia unveiled a new suite of AI microservices at the SC25 supercomputing conference in St. Louis this week, claiming the technology can accelerate the discovery of new materials by up to 10,000 times. The platform, dubbed ALCHEMI (AI Lab for Chemistry and Materials Innovation), utilizes GPU-accelerated computing to virtually screen billions of molecular candidates in seconds, a process that would traditionally take weeks or months.
Universal Display Corporation, a developer of OLED technologies for consumer electronics, is among the first to adopt the platform. According to Nvidia, the company is using ALCHEMI to evaluate potential molecular candidates for displays up to 10,000 times faster than previous CPU-based methods. This acceleration allows researchers to navigate a search space estimated at 10 to the 100th power of possible molecules.
Japanese energy company ENEOS is also deploying the technology to develop next-generation cooling fluids for data centers and catalysts for hydrogen production. The company reported evaluating approximately 10 million liquid-immersion cooling candidates and 100 million oxygen evolution reaction candidates in a matter of weeks—a scale of research it had not previously considered feasible.
The ALCHEMI microservices are part of Nvidia’s broader NIM (NVIDIA Inference Microservice) platform and include specific tools for conformer search and molecular dynamics simulations. The announcement comes as Nvidia technology continues to solidify its presence in high-performance computing, with the company revealing that its systems now power 78% of the supercomputers on the TOP500 list.
In a related development, Brookhaven National Laboratory is utilizing Nvidia’s Holoscan platform to process streaming data from X-ray imaging experiments in real-time, allowing researchers to receive immediate feedback during experiments.
