As interest in AI surges, so too has scrutiny of the technology's energy intensity and its corresponding burden on data centres and power grids. But alongside those challenges are also opportunities for AI to help speed decarbonisation by accelerating the development of sustainable materials.
Battery development is typically a slow and laborious process, with extensive testing and retesting. Measuring how long a battery will last, for example, can involve charging and discharging thousands of times. Using AI, developers can cut down the time and expense of battery design by iterating faster and more efficiently. One early-prediction model helped reduce the time it took to identify fast-charging techniques from over 500 days to 16 days, in part by predicting the final cycle life using data from the first few cycles, according to a 2020 Nature study.
"The pharma industry, which is where you have very expensive R&D and testing costs, is a sector that has embraced this type of approach for years," said Alp Kucukelbir, co-founder and chief scientist at AI company Fero Labs, who authored a report on AI's potential to spur climate solutions. Now, there are financial and societal incentives to apply the same techniques to the battery and EV industry, he said.
Sunnyvale, California-based Chemix says its algorithms can help customers create custom batteries with specific properties, such as an ability to withstand high temperatures or exclude certain materials.
In one case, Chemix helped a client develop a novel electrolyte - the substance that transfers ions between the battery's cathode and anode - that could extend the battery's cycle life by 400% within the span of two months, according to co-founder and Chief Executive Officer Kaixiang Lin. The process would ordinarily take twice that amount of time without the use of AI.
Lin wouldn't disclose Chemix's current customer base but said that it includes premium EV makers who want to move beyond off-the-shelf batteries. Currently, the startup's focus is on electrolyte design for EVs but it plans to turn to cathodes, anodes and slurry - all critical battery components - as well as home batteries.
The EV market is evolving quickly, presenting a challenge for battery manufacturers aiming to keep up with the pace of development.
"Chemix isn't inventing new chemistry, it's working to make what's already out there better," said Yann Lagalaye, managing partner of BNP Paribas Solar Impulse Venture Fund, which invested in the round.
"And just improving the existing [chemistry] is huge, because by doing that they can tackle underserved markets" that currently rely on mass-market batteries that aren't custom built to their needs, such as electric sports cars and motorbikes. Chemix is growing a proprietary dataset gathered in the company's physical lab, which it uses to help train the AI model, Lin said.
"To really practically apply AI and also unleash the full potential of AI, what the industry is missing is availability of data," he said. "If you look at genAI, ChatGPT-type of application, they can screen through massive data from the internet, but when it comes to battery or hardware development, there is not enough data."
Disclaimer: The copyright of this article belongs to the original author. Reposting this article is solely for the purpose of information dissemination and does not constitute any investment advice. If there is any infringement, please contact us immediately. We will make corrections or deletions as necessary. Thank you.
