SES AI Corporation (SES) Earnings Call Transcript & Summary

My name is Mark Newman, Founder of Electric Revolution Ventures and Electric Revolution Insights. This is a third edition of our battery world event and one that I'm particularly excited for. Thanks for joining us today. If you remember, 2 years ago, SES released some pretty amazing data on the 4 mPOWER cell and a first look at a huge 100 mPOWER sale. And last year, that was followed up with data on their 50 mPOWER cell. Unlike many of the other battery companies out there, SES has consistently been very transparent with their data, providing not just cherry picks data points in selected conditions, but a huge amount of performance data in various conditions, various temperatures and power requirements. I challenge other companies to do this for their next-generation batteries, lithium metal, silicon or solid-state batteries. But sadly, I do not see much of this coming from others. Today, SES will finally reveal the highly anticipated data on their 100 mPOWER sale. Even more anticipated, though, is details on the world's first automotive B-Sample JDA for lithium metal batteries. It is becoming quite clear that SES is way ahead of competitors in this space. And SES will be providing further updates on their road map to commercialization, including not just electric cars, but also for urban air mobility. As I mentioned last year, the world is yearning for better batteries, with high energy density that in turn allows longer range and less use for critical materials. SES is extraordinarily well positioned in this space with their leading lithium metal batteries. To explain more, I'd like to now hand over to Qichao Hu, Founder and CEO of SES. Mr. Qichao.

Welcome to Battery World 2023. This past year, we made several really exciting breakthroughs. The kind of breakthroughs that don't happen every day. They happen once every 2 to 3 years. And in the past decade, SES introduced several of these breakthroughs. In 2012, we introduced solid polymer ionic liquid lithium metal. It was a huge breakthrough compared to earlier versions of lithium metal. And back then, very few people would have been thinking about lithium metal. In 2015, we introduced high concentration solvent-in-salt electrolyte for lithium metal. And today, this technology is used in almost every lithium metal battery out there. In 2017, we introduced HERMES, high-energy rechargeable metal cells, for space. At the first ever Uber Elevate Conference, this was our initial entry to drones and space application. And we liked it so much that we gave it wings. In 2019, we introduced Avatar. It's a digital twin of lithium metal battery that tracks all 3 types of data, the DNA, which is material and cell design data. Pregnancy, which is manufacturing and quality data. And Lifestyle, which is cell testing data. And we use the Avatar to monitor battery health and predict incidents. In 2021, we announced the world's first automotive A-Sample for lithium metal, and we signed 3 A-Sample JDAs with Honda, Hyundai and General Motors. And we also introduced Apollo, the world's largest lithium-metal cell. And today, we're going to announce something of equal magnitude. And this is the culmination of years of research and development in materials chemistry, cell engineering and battery health monitoring algorithm. And I'm very excited to announce that we just signed the world's first automotive B-Sample Joint Development for lithium metal. This is a major milestone in the commercialization of lithium metal batteries. And currently, we're not disclosing the identity of the OEMs, but it's one of the 3 and you can guess. And we expect to sign a second one very soon. So how do we do it? Well, our core is material development. And we didn't just develop materials. We've developed new platforms for materials. One is human-based that tries to think like a human and speed up the idea discovery process. Another is machine-based deep learning that tries to not think like a human, but still solve the problem. Let's start with the human-based deep learning. So one thing that our human scientists do a lot is reading publications, and that's where they get their inspiration from. But human scientists read 3 to 5 papers a day, and they get tired, they forget things. They miss details, especially when the coffee runs out. What if we can build a machine, that can read tens of thousands of papers a day every day, meticulously analyzes all details, never forgets anything and learns from every paper in every publisher in every language. This machine will be a super intelligent scientist, smarter than any human scientist. And that's what we're doing. Now if you look at this example of a paper, and it has roughly 3 types of information, text, data graphs and illustration. For text, we already have very mature tools for comprehensive text such as GPT-4 and the machine can comprehend text information. But for data, these graphs, machines currently cannot comprehend this data, cannot comprehend the graphs, the physics, the chemistry, the material science that are represented by the graphs. So we developed this internal tool that can reconstruct data, information from these images, but without having access to the raw data. So now our machine can actually comprehend the data, the graphs and the physics and chemistry that's happening behind the graphs. And then for these illustrations, so these are typically cartoons that tend to be wrong and manmade. So we throw this into the trash can. And we try to train the model with only facts, not opinion. And this tool can really help comprehend information from a large volume of human authored papers, and this really helps speed up the process of idea discovery. But this is still trying to think like a human. One of the major flaws of human like deep learning is that it's human-like. It still thinks like a human. Now let's try not to think like a human. And a battery is a perfect problem for machine-based deep learning to solve because a battery is a black magic. We don't really know how to do anything until we actually figure out how to do anything. And we don't really know how to improve performance or safety. We have some knowledge of the bulk properties of materials, but we really have very limited understanding of the interfacial properties. And it's sort of like the blind men and the elephant. Some think we need to increase connectivity, some think we need to control mossy lithium and some think it's about coulombic efficiency. And -- so even if we develop a tool, they can read millions of human authored papers, we're still trapped in this human-based subjective science, discovery and reasoning. So we want a more pure play, and that's where machine-based deep learning comes in. For example, here are the chemical structures of 10 electrolyte solvent molecules. And some are blurred for confidentiality reasons. Now instead of training the model with human input like viscosity, connectivity, electrochemical window, solubility, we use a graphical neural network model to process these chemical structures, and then we let the model find patterns itself. And then we screen a larger database using these patterns so we can find promising new candidates. What we're doing has never been done before in the battery industry and we are collaborating with several preeminent research labs. We've already seen some very interesting results, and we're really excited about the potential of this approach. And this is why we're building a new facility right next to our Boston headquarter. This facility will be dedicated to synthesizing and testing new electrolyte solvent molecules recommended by both human-based and machine-based deep learning. It's called Electrolyte Foundry. Electrolyte is arguably the most important and complex material in the battery. It's also our core. And this platform allows us to strengthen our core competence in a very fundamental way. At the cell level, we have 5 lines. We have 3 A-Sample lines, and we are preparing 2 B-Sample lines. In Shanghai, we have Line 1 and that's mostly for internal R&D. And Line 4, that's for a B-Sample JDA. In Chungju, South Korea, Line 2 and 3 are 2 A-Sample JDAs, and we're preparing Line 5 for a B-Sample JDA. And this last month, we broke the record and we built 1,100 mPOWER lithium-metal cells per month per line on our Line 2 in Chungju, South Korea. We also built these advanced testing bunkers with automated cell loading so that we can test different materials, different cell designs, different manufacturing process quickly. [Presentation]

And in keeping our tradition of data transparency, today, we are releasing data on our latest 100 mPOWER lithium-metal cells. Back in 2021, we released data on 4 mPOWER lithium-metal cells. And in 2022, we released data on 50 mPOWER lithium-metal cells. We're very excited to see that the data are very consistent as we scale from 4 mPOWER to 50 mPOWER and now to 100 mPOWER lithium-metal cells. So let me walk you through the data in details. First, these 100 mPOWER lithium-metal cells have excellent performance in cold weather, low temperatures. And this shows that even down to negative 30 degree Celsius, we can still achieve 80% capacity. And this shows that our lithium-metal cells have excellent performance in high power. And this is at high C-rate, 3/C, that's 20 minutes full discharge, we can achieve 90% capacity. We can pass overcharge, safety test, and this is done at a third party. We can pass no penetration also done at third party. We can pass external short circuit, again, third-party test. And we can pass thermal stability and all these safety tests were done at third parties. And now not 1 but 2 of our lithium-metal cells, both the 50 and the 100 mPOWER lithium-metal cells have received UN38.3 certification. So far, our focus has been on pouch cell formats. And these pouch cell formats are still the preferred form factor for our current automotive customers. And we're doing A-Sample and B-Sample joint developments with them. And we can build these modules with these A-Sample pouch cells. But sometimes, our customers want a choice. And today, we are introducing an entirely new form factor for lithium-metal cells. [Presentation]

It's a prismatic lithium-metal cell. This prismatic lithium-metal cell has the same energy density benefit of the pouch lithium-metal cell, but safer. And it doesn't replace the pouch lithium-metal cell. It offers a choice to automotive customers that may want to build a prismatic lithium-metal cell. And both this pouch and prismatic lithium-metal cell have about 100 mPOWER capacity. So we're really excited about this prismatic form factor. Going forward, we are going to convert our lines so that they have dual capabilities to produce both the pouch and prismatic form factor for lithium metal, so that we can serve more automotive customers better. And to ensure manufacturing quality of these cells, it's really important that we systematically enhance the rigor of our specs and improve our quality manufacturing control plan. So we are increasing the number of control points from less than 400 in A-Sample to over 1,400 in B-Sample. And we're incorporating imaging tools with AI processing throughout our assembly process, such as vision, X-ray, CT and ultrasound. And all these data are used to train our Avatar and safety is quality. To ensure practical safety we need to build a rigorous manufacturing quality system with full traceability. For example, here are 25 large 100 mPOWER lithium-metal cells that had incidents in the past. And we collect data, DNA data, that's the first column, cell design and material information. And we collect life style data, that's these 2 columns, that's the cell test data. And we collect pregnancy data that's the manufacturing quality data. And by combining all the data, we are able to predict 23 out of the 25 cells that had incidents. That's a 92% accuracy. And last year, our accuracy was only 60%. So this is a really exciting improvement. And we will continue to meticulously collect all data of all steps, all details of all process. And it's a lot of work, but it's very important for the safe commercial use of lithium-metal cells. And the more cells we manufacture, the more cells we use in the field, the more data we collect, the more accurate our Avatar will become. [Presentation]

There used to be a stigma that lithium-metal cell is not safe and lithium-metal cannot be safe. Will we change that? We have significantly improved the practical safety of lithium metal. And one thing that's really important is that we don't shy from telling people that ourselves sometimes blow up. And if we don't fail quick enough, often enough, we don't innovate fast enough. So now that we built this strong solid foundation with A-Sample and B-Sample automotive JDAs, and we've significantly improved the cell safety and incident prediction ability. Where do we go from here? Well, the answer is actually up there in the air, urban air mobility, UAM. UAM is a perfect fit for lithium metal. For anything that flies, weight is really important, and lithium metal has one of the highest energy densities in [indiscernible] per kg of [ any ] battery chemistry. And here's a cycling test that we did, basically, we charge the cell using Silver 3, 3-hour charge and we discharge the cell using a typical UAM mission profile, and we achieved more than 1,500 cycles. This is all with lithium metal. This is a game changer for UAM. Another important factor is the lack of serious competition. Most large battery companies are so focused on EV, land EV. And I think UAM is too small and unimportant to their business. Well, that's where we come in. It's our chance to set industry standards for lithium metal. 30 years ago in 1990s, Sony commercialized the first lithium-ion battery and then changed consumer electronics forever. And now it's the 2020s, SES is going to introduce the first commercial lithium-metal battery and change UAM forever. UAM for us is not just another market, UAM is our destiny. And we plan to approach UAM in 3 phases: First, sub-size aircraft; second, full-size aircraft, but unmanned; and third, full-size aircraft, manned. And our goal is to use these aircraft to field test our A-Samples and B-Sample lithium-metal cells for UAM applications. And many people are so impressed by the capabilities of these amazing aircrafts with our lithium-metal cells, the low-temperature performance, the high power performance. So we thought why not use this capability to do good in the world. So we launched this new initiative called SES Cares. Our mission is to apply advanced drones powered by our lithium-metal cells, the A-Samples, the B-Samples to protect our humanity and environment. This year, we took a large drone, and we built a lithium-metal battery pack using these lithium metal A-Sample cells. And we took to the beautiful island of Jeju to study endangered offense. [Presentation]

It was a wonderful experience. It was really good to see our A-Sample lithium-metal cells coming off the pilot line and actually doing good in the world. And we want to provide SES Cares, this service to more people and more places around the world. So if you have interesting ideas, we would love to hear from you. So let me summarize. We covered a lot of topics today. First, we announced the world's first automotive B-Sample JDA for lithium metal. This is a huge milestone in the commercialization of lithium-metal batteries. Second, we released data on our latest 100 mPOWER lithium-metal cells. They are consistent with our previous 50 mPOWER and 4 mPOWER lithium-metal cells. Third, we discussed our new platforms for material discovery using human-based and machine-based deep learning. And we demonstrated that our Avatar prediction accuracy for large cells is now more than 90%. We also unveiled an entirely new form factor for lithium metal, a prismatic lithium-metal cell. And we introduced this new initiative called SES Cares, where we apply drones powered by our A-Sample and B-Sample lithium-metal cells to protect our humanity and environment. Lastly, there's a quote by the French poet Fontaine. He says, a person often meets his destiny on the road he took to avoid it. I like this a lot because 6 years ago, in 2017, when we first unveiled HERMES for drones and space applications, our destiny first called us. But then we tried to avoid it. we decided to focus on EV development with A-Samples and B-Sample JDAs. Now these JDAs are fantastic. They really helped us build a solid foundation and all the progress that we made, the performance improvement, safety, manufacturing quality, all these improvements that we made can all be applied to UAM. And destiny has a way of keep calling you. And today, we are finally answering destiny's call, and we are officially reentering UAM. Lithium metal was born to fly. Thank you all very much.