Microsoft and PNNL collaboration. AI and supercomputers to revolutionize lithium batteries

PNNL is part of the US Department of Energy, which collaborated with Microsoft for this endeavor. Utilizing the resources of the company founded by Bill Gates, they narrowed a list of 32 million potential inorganic materials down to 18 promising candidates. Microsoft scientists used artificial intelligence and supercomputers to reduce the trial and testing cycle from a twenty-year process to less than a week.

Often referred to as "white gold" due to its color and substantial market value, Lithium is a critical component in the manufacture of lithium-ion batteries. With the ongoing electrification trend and the increase in the use of this power supply by cars, houses, and phones, there is a fear of potential difficulties with lithium supplies in 2025. Experts in the US believe that by 2030, the demand for lithium will continue to grow as manufacturers erect new factories to produce batteries.

Lithium extraction is fraught with challenges and is harmful to the environment. Large quantities of energy and water are used during this process, often generating toxic waste and causing significant damage to the natural landscape. If a mixture containing less lithium but maintaining its energy storage properties is widely adopted, it will result in substantial financial and environmental savings. Scientists note that the new AI-discovered substance allows a single battery to use up to 70% less lithium.

The AI-selected mix prototype battery was developed over nine months and has continuously powered a test light bulb since its creation. The high computing power of the supercomputer checked the chemical properties of millions of tested substances, allowing AI to select promising candidates.

Despite its advantages, this method can sometimes be misleading and generate false results. Researchers warn that AI might mistakenly identify existing compounds or ones that are impractical to create in a lab as appropriate candidates. Potential errors were avoided in this case, and the scientists could synthesize the proposed compound, tentatively named N2116. A Microsoft representative mentioned that such scientific approaches will aid the company in achieving its mission to compress the scientific discoveries of 250 years into the next 25 years.