Sign up to our newsletter

Get the latest articles on all things data delivered straight to your inbox.

Share on

LinkedIn
Twitter
Facebook

Aionics Battery AI Partnerships Featured in Steve LeVine’s The Electric

Aionics - Battery AI Partnerships Featured in Steve LeVine’s The Electric

“If you think you have to sample 100 electrolytes until you find one that works…we can come along and say, ‘Well, now it’s 20 you have to try instead of 100,’” Sendek said. “That’s a total win.” – Why EV Battery Makers Need to Start Using Machine Learning, The Electric

Should battery makers start using machine learning? According to a new piece by Steve LeVine in The Electric, harnessing computational power is necessary to stay alive as the increasingly competitive battery industry enters the Data Age — and Aionics is in the center of the action.

“For instance, no current algorithm can give instructions for how to create an affordable lithium-metal battery that reliably powers an EV for a decade and longer,” LeVine writes in Why EV Battery Makers Need to Start Using Machine Learning. “But the smaller things that AI, used as a tool with the active assistance of human researchers, can already do show that computers are likely to make the difference between battery and EV companies that win or lose.”

The article highlights Aionics’ successful development of an AI-driven cycle life prediction model in partnership with an undisclosed battery manufacturer, and also discusses a 2019 electrolyte development parternship with California-based Cuberg, previously featured in AI Trends.

The article underscores the importance of having robust datasets and well-designed objectives in order to fully harness the power of machine learning modeling. “We can sometimes see weird predictions pop up when screening thousands of candidates with data-limited models,” Aionics CEO Austin Sendek said yesterday in response to the article. “The field is really just now coming out of its infancy, and data is key: This is why Aionics now runs quantum mechanical simulations in the cloud alongside real-world experiments to construct information-rich electronic structure features; more information means better machine learning models.”

The full article is available to subscribers of The Electric here.