Is next-generation AI creeping into EV battery development | TechCrunch

Is next-generation AI creeping into EV battery development |  TechCrunch

Ten billion. That is Are there any commercially available molecules these days? Start looking at them as a group of 5 common combinations used to make the electrolyte in the battery, and it increases from 10 to 47th energy.

For those who count, that is a lot.

All those combinations are important in the world of rock. Find the right mix of electrolyte and you can end up with faster charging, thicker battery power for EVs, power grids or even electric jets. Disadvantages? Similar to the drug discovery process, it can take more than a decade and thousands of failures to find the right fit.

That’s where Aionics developers say their AI tools can speed things up.

“The problem is that there are too many candidates and not enough time,” Austin Sendek, co-founder and CEO of Aionics, told TechCrunch during the recent Up Summit in Dallas.

Dr. Lenson Pellouchoud, co-founder and CTO; Dr. Austin Sendek, Co-Founder and CEO, and Dr. Venkat Viswanathan, Co-Founder and Chief Scientist Photo Credit: Aionics

Electrolyte, Meet AI

Lithium-ion batteries have three main building blocks. There are two electrodes, the anode (negative) on one side and the cathode (positive) on the other. The electrolyte usually sits in the center and acts as a carrier to move ions between the electrodes during battery charging and discharging.

Aionics focuses on electrolytes, and it is using AI kits to speed up detection and ultimately provide better battery life. Aionics’ approach to catalytic discovery has also attracted investors. The Palo Alto-based startup, established in 2020, has raised $ 3.5 million to date, including $ 3.2 million in seed collection from investors including UP.Partners.

The startup is already working with several companies, including Porsche Cellforce battery subsidiary. The company also worked with energy storage company Form Energy, a Japanese materials and chemical company Showa Denko (now Resonac), and Cuberg Battery Technology.

The whole process starts with the company wish list or the operating profile for the battery. Aionics scientists using AI-accelerated quantum mechanics can perform experiments on an existing database of billions of known molecules. This allows them to consider 10,000 candidates every second, Sendek said. The AI ​​model learns how to predict the outcome of the next simulation and helps select the next molecular candidate. Each time it runs, more data is generated and it gets better at troubleshooting.

Integrate AI generation

Aionics has taken it one step further in some cases, bringing a new generation of AI into the mix. Instead of relying on billions of known molecules, Aionics has begun using this year’s next-generation AI prototype, which is trained on existing battery material data to create or design new molecules focused on a specific application.

The company is charging for its efforts using software developed in the Accelerated Computational Electrochemical System Discovery at Carnegie Mellon University. Venkat Viswanathan, an associate professor at CMU and co-chair of the program, is the co-founder and chief scientist at Aionics.

Aionics has also started using large language models built on GPT 4 from OpenAI to help its scientists figure out millions of possible formulas before they start processing them through databases. This chatbot, trained in chemistry textbooks and science papers selected by Aionics, is not used for practical discovery, but it can be used by scientists to eliminate some of the useless molecules in A specific program Sendek explained.

When trained with those textbooks, LLMs allow scientists to question models. “If you could talk to your textbook, what would you ask it for?” Sendek said. But he immediately noticed that this did not do anything wrong with the people who prepared the scientific papers. “This is just the next level of interaction,” he said, adding that everything is verifiable by pointing back to the source used to train the chatbot.

“I think the good thing about our sector is that we are not looking for specific facts, we are looking for design principles,” he said as he explained the chatbot function.

Winner Selection

When billions of candidates are screened and narrowed down to a few or designed using Aionics AI models, send their client models for validation.

“If we do not get to the first round, we repeat and we can run some clinical trials to confirm it until we reach the winner,” Sendek said. “And when we find a winner, we work with our production partners to scale that production and bring it to market.”

Interestingly, this process is even used in some novel areas, such as cement. Chement, a startup co-founded by Viswanathan and partnered with Aionics, is working on ways to use renewable energy and raw materials to stimulate chemical reactions to create non-emission products such as cement.

#nextgeneration #creeping #battery #development #TechCrunch
Image Source : techcrunch.com