Good afternoon, everyone, and thanks again for joining us at day three of TD Cowen Sustainability Week. This is Gabe Daoud, TD Cowen Senior Energy, Battery, and Charging Analyst. Next up, we're delighted to host SES, and presenting from SES is Dr. Qichao Hu, Founder, Chairman, and CEO. As folks on the line are probably aware, SES is a next-generation lithium-ion battery company attempting to commercialize lithium metal batteries for applications both on the ground and in the air. So with that, Qichao, thank you for joining us.
Thanks, Gabe.
Good to see you again. Looking forward to a good chat. Qichao, why don't you maybe just give us a little bit of an overview on SES? Maybe summarize the most recent investor letter that was posted, you know, a couple weeks ago, and then we can get into some Q&A.
Sure. Yeah, so we develop lithium metal batteries for EV and urban air mobility applications. And then recently we announced we entered into two B-sample joint developments with auto OEMs. And then also we are converting the A-sample lines to produce cells for urban air mobility applications. And then we're headquartered in Boston. That's where we do all the materials R&D. We do a lot of the AI development, and then also we do the A-sample, B-sample cells for GM, Hyundai, Honda, and also UAMs in our Shanghai facility and Korea facility.
Great. Great. Well, maybe at a higher level, Qichao, why do you think lithium-metal batteries are necessary? Maybe just talk a little bit about the opportunity, the need for lithium metal, and, you know, some of the puts and takes between, you know, lithium metal versus, you know, traditional lithium-ion batteries.
Yeah. So fundamentally, it comes down to cost. And then specifically, if you look at the EV applications, in terms of range. So there are two kinds of markets in the EV applications. You have value cars as well as premium cars, and then for premium cars, you want to increase the range, and then for value cars, you want to lower the cost. And you can achieve both by using a higher energy density battery. And then if you look at the history of batteries, lithium-ion today, with graphite and silicon anode, is basically the state-of-the-art. Lithium metal is the most practical battery chemistry that can realistically and practically give you a significant boost in energy density.
So for OEMs, and, and this is why, GM, Hyundai, Honda have been working with us for A-sample and, B-sample development, because we can, we can lower the cost for the vehicles. And then also for UAM, urban air mobility, it also comes down to cost. It's about, $1 per passenger per mile. So if you can have a, a lighter battery, then you can add more passengers, more suitcase, more payload, and also can fly farther. So that, that lowers the cost, for the UAM operation.
Got it, got it. Okay. Maybe just sticking to the tech and architecture and the IP of the battery for a minute, maybe just talk about, you know, we've hosted some solid-state battery companies at the conference. Solid-state continues to make a lot of noise with the promise, but it's really about advancing the anode, and obviously you guys are attempting to commercialize lithium metal. So maybe just talk about the IP. What's different from your design versus, like, a solid-state design? Maybe get into some of that, if you could.
Yeah. So really, it's lithium metal that gives you the boost in energy density, not solid or liquid. So you can have a lithium metal, solid or liquid, it will have the same energy density. And so we focus on the anode, and then we want to use a lithium metal as the anode, not graphite, not silicon, because of the energy density improvement. And then to enable safe and long operation and performance of lithium metal, then you really have to improve what's called the Coulombic efficiency of the electrolyte. And so, our core IP are really in how to make this anode, lithium metal anode, and especially when you have to make these large cells, A-sample, B-sample cells.
Some of these can be 500-600 mm wide, and no one in the world make these wide width, high purity, high uniformity, high quality anode. So we actually make these things in-house. So the manufacturing process for this anode is really key for us. And then also in terms of the electrolyte, we actually started out working on solid-state electrolyte, but then we pivoted to liquid around 2015 because we found the biggest benefit of liquid is you can use lithium-ion manufacturing process to manufacture lithium metal cells. And this was a big deal for us, because if we use solid-state, then you do have to innovate a lot at the manufacturing process, and we did not want to spend too much effort in the manufacturing process.
We want to use whatever is mature and scalable out there. So this liquid electrolyte allows you to do that because you can have an A-sample line, B-sample line, and then if you, if you change the cell dimension, you just change the dies. And this is why we converted the A-sample line to make UAM cells, and the B-sample line, now we're going to make EV B-sample cells. And we can come up with new electrolyte formulation, new electrolyte molecules, and then just drop in to that existing line. So think of you, you can test lots of different molecules, lots of different electrolyte formulation, and then at the end, you can have different cell designs for EV UAM applications. But in the middle, this pipe, this A-sample line, B-sample line, stays the same.
So, your CapEx investment, your manufacturing hurdles and challenges are much less compared to solid-state. So the biggest benefit to a liquid electrolyte is just you can use current mature lithium-ion manufacturing process to make lithium metal cells and to achieve the same energy density as solid-state lithium metal.
Got it, got it. Okay. So maybe then talk about, though, the manufacturing process for the lithium metal anode, and maybe some of your supply chain partners there, if you can. I know, you know, Applied Materials is an investor. I imagine there's some work-
Yes
... being done with them. But yeah, maybe just talk about the lithium metal anode manufacturing process.
Sure. So, we've experimented with probably all of the different manufacturing approaches. You have extrusion, lamination, you have thermal evaporation, you have slurry coating, and as you mentioned, Applied Materials has been an investor and a partner since 2015, for almost 10 years. And then Applied Materials is a leader in thermal, in making thermal evaporation tools. So, for now, our workhorse approach is extrusion and lamination. And this approach historically had challenges in the sense that you could only extrude thin and uniform lithium, but then in a narrow width, typically about 100 mm or so.
But then when we had to make these 500-mm wide A-sample and now B-sample cells, we really had to optimize this process so that we can actually make this much wider, but then still achieve the same quality, same surface, same uniformity. So that was quite challenging. And then in the early days, we tried to rely on the vendors, but then we moved to basically in-house. So actually, now we buy the ingots. We buy high-purity, better grade ingots, and then we actually extrude it to the thickness and the width that we want, and then we laminate that onto copper.
And then we found by doing this in-house, actually, that improves quality a lot, because if you ask the vendors to do it, then they have to extrude it, and then they package it, ship it to the next location, then open it up, and then laminate it a few times to get to the thickness, and then package it, and then ship it to the next location to laminate it onto copper, then package it, then ship it to us. So you have a lot of these steps in the process that actually hurt the quality. So by doing this whole thing in-house, we can really ensure the quality. And then down the road, we are still trying thermal evaporation.
We think that could be an even better approach and potentially lower cost down the road, and more scalable than the current extrusion.
Got it. Got it. Okay. Okay, that's helpful. That's helpful. Okay, and then, you know, you've laid out some pretty impressive data on the large format cell, the 100 Ah cell, I think it's a 105 amp hour cell, so pretty attractive energy density. Can you, so maybe put into context, like the energy density targets, for the 100Ah cell versus, you know, what best-in-class NMC cathode cell is today? You know, maybe that's like 300 watts per liter, or per kilo, and you're maybe targeting 400. Just talk a little bit about the energy density improvements that you're targeting.
Sure. So for example, if you look at a high-nickel NCM cathode, if you use graphite, typically it's around 280 watts per kg, and then if you use a silicon anode, you are at about 300-320 watts per kg. So that same cathode, once we switch to a lithium metal anode, we start at 400 watts per kg at the cell level, not the stack level, not electrical level, but at the full cell level. And then we've already demonstrated that with the EV and UAM OEMs.
Okay. Okay, yeah, that's helpful. And, okay, and so maybe talk a little bit about your impressive JDA partners. Obviously progressing to B- sample with several of them. Maybe just talk about Hyundai, Honda, and GM, and where you are with each of those.
Yeah. Yeah, so we announced 2 B- samples so far, and then we're still working on third and potentially fourth and fifth. So some of the OEMs are very interested in LFP cathode, and LFP combining with lithium metal in a prismatic format for the value vehicles, because LFP lithium metal can give you the same energy density as a high-nickel lithium-ion-
Mm
... but it's actually cheaper because LFP is cheaper than the high nickel. So LFP lithium metal is very interesting, very appealing to the value vehicles. And then one other OEM is very interested in the Avatar aspect, especially from A- sample to B- sample. We build more cells. We go from less than 1,000 a year to now more than 1,000 a month. And then we go from just 200 quality checkpoints per cell to now 1,500 quality checkpoints, and the more image data. So one OEM is also very interested in the improvement in the collection of data and how that helps the accuracy of that Avatar model. And then one other OEM is interested in EV, but at the same time, also basically applying that similar cell design for UAM applications.
So the platform technology that we're using for all three is quite similar, but then different OEMs will have different flavor to it.
Got it. Got it. Okay. You mentioned Avatar, the AI tool. Maybe talk a little bit about that and how you're incorporating the AI into the business. Obviously, you mentioned safety checkpoints and improving the accuracy of Avatar, but what makes Avatar AI? Could you maybe just talk a little bit about that?
Sure. So there are two parts. One is the data collection, which is not AI, and then another is the learning aspect. So the data collection, and then we started doing this in the A-sample line, and then we are doing this basically at full scale in the B-sample lines. We actually collect all the manufacturing processes in the line. So in the A-sample line, you had what's called islands of automation. So you had small equipment, and then each equipment would run the process, but then you have a human that would basically hand-carry the cells to the next machine. So you could collect data from the machines, but not from the human step in the middle.
Now, B-sample, we have these islands of automation, so several of those machines are integrated, so we can collect a lot more data. And then we have far fewer steps that we miss in terms of not able to collect data. So this Avatar can collect all the manufacturing data, and then we pre-train a machine learning model. That's where the AI comes in, because this machine learning model can actually learn from this data and then come up with new features. For example, say, if you feed a traditional quality team... Actually, in our company, the quality team is actually under Avatar.
It is actually under the Avatar AI team, because we need the traditional quality team to collect the data, but then if you just show that data to a quality person, he or she will only see, say, several hundred features. But if you show that same set of data to this machine learning model, that model can come up with several tens of thousands of features. Some cannot be explained by human, but then they are very accurate in terms of telling you. So before, it was like binary, "Okay, this cell is good or reject." Now, you have a spectrum, and then even the cells that you accept, you have cells that are better than others. So we want to identify these false positives.
In the past, lots of cells that get accepted still have defects, and then if you put those cells, those false positive cells, in a car, and then you combine that with an abusive driver or abusive pilot, you will end up with an incident. But now we can actually model that. We can have these cells that are accepted but are close to the threshold, and then also model that with the actual driver or pilot behavior, mission profile. And then once you combine those data to a model, that model will learn, and then that model will figure out, okay, once you have a certain combination of these different features, then you will have an incident. So this model allows you to predict these incidents before they happen. So this is really powerful, and this-
Mm-hmm
... is very, very valuable to the OEMs for lithium metal. And actually, even beyond lithium metal, some of the UAM OEMs actually have asked us to apply this model to their lithium-ion batteries, their current lithium-ion batteries. Because why? Because that same aircraft flies the same mission profile with lithium-ion or with our lithium metal. So by training this model with lithium-ion, lithium metal, different chemistry data, but similar mission profile, actually, this model can actually learn a lot. So, going forward, this actually opens up some interesting business models. So in addition-
Mm-hmm
... to us making the batteries, the B-sample for UAM EV, we could also license this Avatar AI to beyond lithium metal batteries.
Mm-hmm. Yeah, that's actually that was going to be my follow-up. You mentioned one of the OEMs or maybe even multiple OEMs are really interested in Avatar. So how does that... Yeah, what are some of the monetization opportunities of Avatar? And I guess you said a licensing type agreement?
Yeah, I think so. Yes, those are still being discussed, but you can imagine, say, this OEM has 1 million vehicles and maybe 10% with lithium metal cells, 90% with lithium-ion cells, but we can apply Avatar to all cells.
Mm-hmm. Mm-hmm. Okay, yeah, that'd be very effective.
Yes.
Okay. Let's maybe talk a little bit about... We'll go back to EV, but maybe just talk about the UAM opportunity.
Okay.
Maybe just, like, size of the market, so the differences in the requirements between an EV cell and a UAM cell. Yeah, we'd love to just get some updated thoughts there.
Yeah. Yeah, so the UAM market is actually growing much faster than we expect. For example, if you look at the FAA for North America, EASA for Europe, and the CAAC for China, different countries have already started providing certifications to cargo as well as manned UAM. And then, for example, this year at the Paris Summer Olympics, we expect Volocopter to potentially fly with athletes on those aircrafts. And then, next year, also, several California-based UAM companies are also expected to receive their type certification. In the market, I mean, the TAM, depending on the source, is expected to actually grow to be pretty significant, around $40 billion by in the 2030s. So-...
and then, so for lithium metal, it's actually really interesting because a B-sample for EV is quite, it's basically equivalent to commercial for UAM. Why is that? Because the scale is much less. For B-sample EV, when we make more than 1,000 cells per month, that translates to about two aircrafts' worth of batteries per month. And then, for now, most of these OEMs actually basically, that's about the demand from these OEMs in the next 1 to 2 years. So we can actually grow with the OEMs. And then in terms of demand, it's quite challenging, but then it's actually a very good fit for lithium metal. So UAM requires at least the same energy density as lithium-ion as EV, but then also on top of that, requires higher power density.
Because when you take off and then land, that power density can get as high as 5C, 7C, so that's even higher than a race EV. And then for lithium metal, high power density on discharge is actually very, very good for lithium metal. So... And then also the weight requirement, weight-to-power ratio for lithium metal is much higher, so at least 400 watts per kg for lithium metal, compared to about 300 for lithium-ion. So for UAM, commercially and technically, it's a very good fit for lithium metal. And this is why we are not building new lines for UAM, we're just converting the old A-sample lines for UAM.
Right. Right, okay. Yeah, it sounds like a great natural market to target, given the technology. And, you know, as you mentioned, B-sample in EV is effectively commercial product for UAM. But okay, so four lines then currently, right? Two in Seoul, two in Shanghai. Can you just give us an update on, I guess, building the B-sample lines, what—how's the progress, and just how's the manufacturability going overall this year?
Yeah.
Still at 1,000, 1,000 cells per month per line, right? Is that the, the number?
Yeah. Yes. So, with Hyundai, we already issued PO for that B-sample line, and then that B-sample line will actually be in their facility in Uiwang, South Korea. And then this is actually the first time that we build and operate a line inside an OEM's facility. The other B-sample is gonna be in our own facility. So for the Hyundai one, we issue PO, and then we expect the line to be up and running by Q4 this year. And then, in parallel, we're not waiting for the B-sample line to be ready to build cells to do B-sample development. In parallel, we are using the A-sample line, the old line, the one that we have not converted to UAM, to do all the B-sample development.
It's just that once the B-sample line becomes ready, then you can validate all the development on the large B-sample line.
Got it. Yeah, okay. And the 1,000 cells a month per line, like, is that the yield, or is that... What, what's the yield off of that?
Well, so we haven't built that many cells yet. But then in the end of last year, when we tried to build 1,000 cells per month per line on the A-sample line, on the improved A-sample line, we could get about 80%. And then, so we expect B-sample line, the yield... Now, Avatar is going to get more rigorous for the B-sample line. So even if we do everything the same, that yield will come down, and then that's intentional, because we are adding more quality checkpoints. And then, Avatar actually intentionally wants to lower the yield, because we want to filter out some false positive. So, it's a push and pull.
The improvement in the B-sample line will improve the yield, but the improvement in the Avatar quality control will also lower the yield. So I think by the time this B-sample line becomes ready, the yield will still be around 80-ish%. And we don't want to have a very high yield. We actually want to keep the yield low so that the Avatar is doing its job.
Right. Okay, okay. Got it, got it. Okay. Okay, so obviously pretty deep relationship with Hyundai. How much customization or, or, you know, back and forth is there with Hyundai on just, like, design parameters and any specific specs that they maybe want to see from you? Like, how much back and forth is there?
Yeah, so in the B-sample, because now, also as part of the B-sample, we also are going to build cells and put the cells into modules for several dozen demo cars. So there is a lot more customized specs in terms of they want the cells to be this particular dimension, particular cathode loading, particular power density, particular mission profiles, much more so than in the A-sample. Because A-sample, it was more just to show the performance at the cell level. But now, because the target is to put these inside the demo cars, it's much higher.
Okay, okay. What is the cathode loading that they'd like to see from-
... Yeah. So, so it depends on the car. I mean, we have cathode loading going from 20- 40, and then some even more than 40. So the higher cathode loading are typically for value cars. Lower, so, so the higher the cathode loading, the lower the cost, and then some of the lower loadings are more for the sports cars and, and, UAMs, because of the higher power density.
Yeah, yeah. Okay. Okay, got it. Got it. Okay, makes sense. Okay, and maybe, yeah, talk to us a little bit about just the balance sheet and liquidity and funding needs as you, you know, progress towards commercialization. And maybe on that note, too, just remind us when you expect to be commercialized with one of your JDA partners.
Yeah. So in terms of commercialization for UAM, because like I said, B- sample for EV is equivalent to UAM. We actually expect to get revenue from supplying the cells to UAMs beginning of next year. Maybe end of this year. So the commercialization and revenue from UAM will happen much sooner. And then in terms of for EV, now we're in B- sample. We expect to get a C- sample second half of next year, 2025, and then SOP, start of production, in 2026. And the balance sheet as of end of Q1, we had about $320 million, and then guidance for this year is about $100 million. So we have about a 3-year runway, and that's sufficient to get to C and SOP.
Yeah. Okay. And Qichao, like, have you noted a change in, like, OEM appetite for, like, different parameters in EV batteries? Like, meaning, like, is this focus maybe shifting away from the most energy dense cell to maybe a cell that could just fast charge? Like, what, what are some of the conversations like? And has anything changed over the last, you know, couple of years in that regard?
Yeah. Yeah, I think definitely now compared to 2021. Back in 2021, a lot of emphasis on range, increase of range, but now it's about lowering the cost. But actually, that's interesting because to lower the cost, a higher energy density battery can actually help you lower the cost. For example, you pair lithium metal with LFP, you can get the same energy density as high-nickel lithium ion, but lower cost. So, a lot more focus on lowering the cost, but lower the cost by going to a higher energy density battery.
Got it, got it. Okay. Okay, yeah. That's definitely interesting. Okay, and what are some other goals or guideposts we should be watching out from you this year?
Yeah, I think, definitely one is, commercial revenue from UAM. I think that we're quite close. And then, of course, B-sample, the completion of the B-sample lines, and then the training of the data from the B-sample lines to improve the accuracy of the Avatar model that we have.
Okay, great. Great, and just, maybe going back to the tech side, or you showed a high energy dense, cell, like we talked about. Maybe just talk a little bit about, like, cycle life and what cycle life targets maybe your JDA partners want to see out of you and your cells.
Yeah. Yeah, so, the first generation is going to be about 500 cycles, and then, we can, we can achieve 500 cycles. And then, per cycle, the car, with lithium metal, is about 300 miles. So that's 150,000 miles total range. That's, that's, sufficient for the minimal warranty. And then for subsequent generations, then we want to get to 800 cycles, 1,000 cycles, 1,500 cycles. So you have, 200,000 miles, 240,000 miles. You can, significantly increase the, the total, miles and then lower the, the cost of the ownership of the car. And then the way we do that is basically by coming out with new electrolytes.
And this is the nice thing about this liquid electrolyte is we can always come up with new molecules, and then currently we are at 500 cycles, but then we only tested several hundred molecules. We're going to test several million molecules, several billion molecules, several trillion molecules. And the more molecules we test, then we will definitely find electrolytes that can give us higher cycle life. And then we just basically change the electrolyte without changing the B-sample line or C-sample line. That line stays the same, we just swap out the electrolyte.
Got it, got it. Okay. So the business model is to manufacture cells yourself, familiar with the JDA partner. Do you think about licensing any of the IP, you know, whether it's like the electrolyte or the protected nano core? Like, are you open to other forms of business models, or is just making the cell entirely yourself the way to go?
Yeah, I think for now, the business model is twofold. In terms of cell manufacturing, we do want to participate in cell manufacturing, because that's important, because we, we do have a lot of trade secrets and know-how in terms of how to actually build these cells well. And also, by participating in the manufacturing, we, we get access to good quality data. That's really important to train the Avatar. But then, we don't plan to build the cells purely ourselves. We do plan to have joint ventures, likely 3-party joint ventures: us, a car company, and then a large battery company. So we build the cells together. And then the other aspect is we can license that Avatar to the OEMs for other lithium-ion batteries.
Well, okay. Okay. All right, great! Well, well, thank you, Qichao. It's a great, great conversation, as always. Thanks for taking the time and best of luck to you and SES the rest of the way.
Thanks, Gabe. Thank you, everyone.