Following our 2024 study conducted by Forrester, we had the pleasure of hosting a webinar featuring an esteemed panel of experts from the industry.
The discussion centred on the current trends and future directions in AI and battery validation, providing valuable insights into the industry's advancements and challenges.
The expert panel included (bios included below the blog):
Summary of key points:
Starting with current trends in AI and battery validation, our experts highlighted the transformative potential of AI-driven techniques. Dr. Richard Ahlfeld, who has specialised and explored artificial intelligence and machine learning for years already, shared his perspective on the growing role of AI in battery development:
"From my perspective, the fact that there's more and more data in testing and in product development is exciting. The ability to actually apply AI to testing because of the availability of data in the R&D process has only really blossomed and come to fruition in the last couple of years. We now have a huge amount of data, which means we can actually leverage machine learning within the battery development cycle."
Richard also highlighted a common misconception:
"What we call AI today is essentially an algorithm that is learning specific patterns and relationships from data in a similar way that a human would. If the answer is not in the data and we don't have any data on what would happen, for example after thousands of people use solid-state batteries for three years, then the answer that I'm going to get from the AI is obviously useless.
What's sad is teams then decide AI doesn't work but there are a lot of applications today in machine learning that are actually very powerful and very useful. One of them, for example, is anomaly detection. AI is very good at looking at specific test data, finding an error, saying ‘hey, this is wrong, you need to have a look at this.’ It can save you three to four months of work every year and that makes you faster and more effective and provides more insight."
Having seen large scale applications of AI up close, particularly in China where operations are “remarkable”, Bob Galyen believes AI is already a key tool at multiple points along the supply chain of many industries including battery technology, which is driving innovation for EVs:
“I saw some of the most remarkable uses of artificial intelligence over the last two weeks while I visited my past colleagues [at CATL] in China. The machines are collecting the information, they're posting it online, so to speak, it talks to the machine upstream and the machine downstream, and you look at the overall picture of quality control in the manufacturing area. It's remarkable what AI can do if it's exercised correctly.
Electric vehicles are remarkably difficult machines, particularly the battery pack system. You have almost every discipline of engineering that goes into making batteries. It's chemistry, it's electrochemistry, it's chemical engineering, it's packaging engineering, it's structural engineering, it's the use of software, hardware, making circuit boards, making electrical circuits within the battery pack system, proper ground protections... There are thousands of applications that can be used, but I believe that these are the trends that I'm seeing in the EV world, that everybody is using AI and it doesn't matter which discipline you're in. It's just a matter of maturity level.”
Bob Zollo describes EV and battery as transformative to society but highlights that the industry still has a way to go:
“If we think about where EVs are today, they're a small percent of the market compared to the internal combustion engines. So as EVs move into the mainstream, we're talking about transformative changes that are going to be an order of magnitude bigger than they are today. We're going to need more batteries. We're going to need better batteries. Batteries that achieve things like bringing the cost of the vehicle down to where everyone can afford them. We're going to need increased safety and better range which means higher capacity so that vehicles can go longer distances between charges.
In order to carry out that battery innovation, you're going to need to run R&D with more engineers testing. So, everything is more, more, more – and we're going to need to use tools to make us more productive. AI is one of those key tools that will deliver faster insights, make better use of the data, make people more productive, and figure out how to make the machinery and systems more productive through optimisation.”
Turning our attention to the study AI in Battery Validation, the panellists discussed the findings that most stood out to them. A primary theme that emerged was the need to get to market faster without compromising quality. In considering the use of AI specifically to address this challenge, the study delved into how AI will be adopted?
In this light, a few relevant findings emerged that the panellists discussed, including that:
Yen T. Yeh: "The report underscores a pivotal insight: AI is viewed as essential by engineering leaders for staying competitive in EV battery development. Leaders are recognising the potential to optimise resource-intensive battery testing and streamline battery validation processes through AI, maintaining high quality while reducing time and effort. This acknowledgment of AI's role marks a trend towards more innovative development practices."
Bob Galyen: “If you look back 12 to 15 years ago, the cost of a new battery was over $1,200 per kilowatt-hour. Today, you can buy LFP and NMC under $100 per kilowatt-hour, and this is because of the amazing research that's been done. Artificial intelligence has played a big role in developing many of the new active materials.
This same storyline applies throughout the entire value chain, from the raw materials, through to processing, to the building of the actual cells. AI is intertwined in every segment of this process when applied the right way. AI is really integrated across the entire battery development and production process.”
The future of AI in battery validation appears promising, with common themes pointing towards more data-driven innovation.
Yen Yeh emphasises the importance of a dedicated infrastructure going forward:
“By analysing datasets from battery testing, AI-enabled methods have the potential to forecast cycle life and identify the most promising cells much faster than traditional approaches. This reduces the time and cost associated with product development, allowing companies to bring more efficient and reliable battery-powered products to market more quickly.
More batteries mean more data, and we’ll need a dedicated infrastructure to track, manage, and optimise for the safety, performance, and even the financeability of batteries. This intersection of batteries, data, and AI will be fertile grounds for innovation in the foreseeable future."
Richard: “When I fast forward in my mind three years, five years, 10 years in the future, I try to figure out where language models and AI models are going to be in 2030, which is a crazy thought experiment. It could go very far. If we assume that AI can figure out what test you should do next and what the optimal tests are in your lab, and AI can tell you based on that, whether this is the right battery or not, whether it’s going to fail or have a thermal incident... you can figure out what you should do and it can conclude from what you've done whether it was the right thing or not. The next logical thing for me is then why do I even need to operate labs in the normal way? Couldn't I just do what we do in manufacturing, which is put a lot of robots into my battery test lab and let the robots do all of the testing?
I've seen some prototypes of high throughput labs, some of them are only in material screening, but I think they could work anywhere where you have one AI brain literally coordinating all of the tests that you want to do on all of your batteries, 24/7, forming various experiments, screening the materials, making recommendations of what works, and slowly but surely iterating those batteries faster and more efficiently to high performance. I think we're not that far away from that, even in the next five years.”
Bob Zollo rounded off with a prediction based on the practical application of AI for companies, speaking on a common challenge he thinks will become much easier to navigate once the benefits of AI are more measurable:
“Whenever you bring a new technology, there's an aversion to risk that says, I need to make sure this is really going to do the job for me before I spend money. If I'm not saving anything, I'm not going to take the risk of using new technology [...] If it's simply measurable, the decision is going to get made and AI is going to get applied there. My prediction is that’s the entry point: where it's super easy to measure the positive benefits, financially, whether because of a reduction in investment, in equipment or reduction in test time or similar.”
The experts on this panel share the belief that the integration of AI in battery validation represents a transformative leap forward, particularly for the EV industry. Our expert panel highlighted how AI can drive innovation, enhance productivity, and reduce costs, from R&D to large-scale manufacturing. As the demand for safer, more efficient batteries grows, AI's role becomes increasingly critical, promising faster, data-driven advancements.
By leveraging AI's capabilities, we can expect a future where battery technology not only keeps pace with the rapid adoption of electric vehicles but also sets new standards in performance and reliability. Embracing this technology will be essential for staying competitive and leading the charge in the next generation of energy solutions.
Watch the full webinar and download the full study to access the findings and statistics shared in this post:
Watch the webinar