Good morning, everybody. You know, I was a little worried coming into this event because Tim was telling me, like, hey, between Super Bowl weekend and the ski weekend in New York, which one should we pick, right? And I was doubtful, like, oh my God, ski weekend, these New Yorkers, nobody will be here. I'm glad the turnout seems to be really, really good here. And thank you all for coming today. Really appreciated that you're spending time to hear more about what Lam has to say. Also, thank you for those in the webcast. It's great to have the audience here and tell our story today. Before I give the stage on to the stars of the show, I want to go through some housekeeping items first, and obviously the most important being the disclosures and forward-looking statements that we are going to do.
The safe harbor is behind me. We will be making forward-looking statements that are subject to risks and uncertainties, as laid out in the risk disclosures that we have in the SEC public filings. I memorized that. I'm not reading it, and we will also make statements on a non-GAAP basis on some of the financial disclosures that Tim, Doug, and Seshasayee are going to make. And the reconciliation between GAAP and non-GAAP will be at the end of the presentation in the appendix section of the slides. It's pretty straightforward, the presentation today. It's going to last two hours, including 15 minutes of break when you can take a step out and put your buy order in for Lam. Just kidding, but there'll be interactive sessions where there'll be good content. Our comms team has done a phenomenal job of putting all these visual things in there.
And we have those. And you guys always ask about the complexity and what's happening in there. And hey, where is this going? So they did a great job of putting that out there. So please spend some time learning about that. It tells why Lam is doing so great in some of the areas that Tim, Doug, and Seshasayee are going to talk about. The Q&A, the most important part. We'll have about 30 minutes of Q&A at the end of the presentation. Hold your questions until then. So it gives us a bit more be very efficient about how we handle everything and get you guys out in time. And as you leave, there will be more time for you to interact with management as well as the content that we have for you. And finally, there'll be box lunches. Please grab one as you head out.
And that's the main part. So without taking too much time, let me hand the show over to Tim Archer, our President and CEO.
Dream of the future. We engineer it. At the atomic scale, where breakthrough technologies take shape, two fundamental forces drive every advance: the science of perfect placement, the art of precise removal. This is where Lam Research leads. In the space between atoms, where angstroms define the boundary of what's possible. While others question if it can be done, we prove it. From advanced memory to logic, from cloud computing to artificial intelligence, deposition and etch drive progress in semiconductor manufacturing. And our innovations keep pushing the industry forward. Lam Research. Two critical enablers. One proven leader.
Good morning. And thank you all for joining us today. I'm really happy to be back here in New York City for this Investor Day. Many of you might recall that we did hold one here about five years ago. And at that time, I said that there had never been a more exciting time to be in the semiconductor industry. You might remember that Investor Day was March 2020. Didn't quite know how exciting it was going to get. We had the global pandemic, global supply chain shortage. We had the rising geopolitical tensions. It got pretty exciting in the semiconductor industry. But in the past five years, I'm really happy that Lam remained focused. We've executed. We've made transformational investments in our product portfolio. We've diversified our device segment exposure. And we've made and scaled our global infrastructure to be closer to our customers for speed and efficiency.
We've done all of this while returning over $17 billion in free cash flow to our stockholders. I think it's a pretty nice five-year run, but it's time for us to hold another Investor Day. Over those five years, our execution has been strong, and you can see this in our most recent results. We just reported that in 2024, we delivered record high system revenues in both Foundry Logic and DRAM, and last year, we also achieved our highest annual gross margin since we brought Novellus and Lam together, so we're back here in New York for Investor Day. And I can confidently say that semiconductors are still a great place to be, but today, we're going to take this one step further.
And we're going to show you that as deposition and etch intensity grows through each technology inflection, and with NAND entering what looks to be a multi-year technology-driven upgrade cycle, Lam has never been better positioned to outperform. And there is no better place to be in semi-cap than Lam Research. Now, I know all of you are likely suffering just a bit of AI fatigue. So we're not going to spend a lot of time on AI itself today. But it's hard not to acknowledge that we probably stand at the cusp of what may be the biggest revolution in technology of our lifetimes. In terms of what it means for Lam, AI is driving tremendous WFE investment. And if we look just at data centers, we see the capacity build-out requiring something in the range of $200 billion of WFE over the next five years.
What is most important for us is that the performance requirements of the AI era are driving major technology changes at the leading edge. These changes are highly beneficial to Lam. AI data centers require high-performance accelerators, high-bandwidth memory, and fast, dense enterprise SSDs. To build these devices, our customers are racing to implement new architectures and new integration approaches. These include things like gate all-around, advanced packaging, backside power distribution, and high-density NAND.
Each of these changes drives increased performance, but also increased manufacturing complexity. Over the past five years, WFE spending per leading edge wafer grew about 20%. With the technology inflections that are underway and those that are on the horizon, we see WFE intensity accelerating at the leading edge, increasing by over 50% per leading edge wafer over the next several years as we move towards the $1 trillion semiconductor industry.
We have focused our product investments to capitalize on the changes that are taking place. Lam is the enabler of 3D architectures. And we see a unique opportunity through these coming inflections to change the status quo on SAM and share leadership in our industry. So let's dive into why we believe Lam is the best place to be in semi-cap. First, as semiconductor industry progresses towards that $1 trillion future, we will expand our SAM faster than WFE. Second, we will gain share in the technology inflections with best-in-class products. And third, we will grow our CSBG business faster than the installed base by reimagining customer value in a more complex world.
Let's begin with SAM expansion, where we see the opportunity to increase our SAM share of WFE from the low 30s to the high 30s% range as deposition and etch intensity grows at the leading edge. In fact, through the rest of this decade, we see a strong technology-driven tailwind for Lam's SAM expansion. And whether you look at Gate-All-Around or Advanced Packaging or Backside Power Distribution or any of the other inflections shown on this chart, these are all multi-billion dollar opportunities for deposition and etch, each enabled by new specialized capabilities in areas where Lam is already a leader. For instance, copper plating and high aspect ratio silicon etch for TSVs, or ALD and selective etch for Gate-All-Around, or moly metallization and Carbon Gap Fill for high-density NAND. Our products are ready, and the market needs are moving towards our strengths.
To illustrate this point, I'd like to start with NAND. Now, as everyone knows, Lam is a leading technology provider for NAND. Unfortunately, as you also know, the NAND business has been a pretty rough place to be for the last few years. Capacity expansion in 2021 and 2022 led to a drop in spending of more than 70% in 2023, with only a slight recovery last year. But technology requirements don't wait for markets to recover, and so Lam continued to invest during the downturn. And today, we have a technology roadmap. We have a product portfolio, and we have the customer engagements to build NAND at 500 layers and more. And as the number of layers grow, so does the need for more deposition and etch, which we see driving our SAM in NAND 1.8 times higher.
Now, we said on a recent earnings call that because of the severe downturn in NAND over the last couple of years, two-thirds of the industry's bit capacity is still made at nodes less than 200 layers. These NAND devices, they rely on several critical technologies from Lam that we've been leading in for more than a decade, things like stack deposition, the channel hole etch, the tungsten ALD. And as we enable scaling to 300, 400, 500 layers or more in NAND, customers are looking to meet the needs of the AI era. We're going to need many more deposition and etch steps to reach those higher density NAND. So let's take a look at how NAND manufacturing itself is evolving.
For 3D NAND, one of the ways to add more layers is to build with a tiered approach using carbon plugs, allowing for a taller stack with more memory layers. Even though these stacks are only tens of microns thick, the stress they put on the wafer can cause it to bow by hundreds of microns. By depositing a stress-compensating material on the backside, the wafer is flattened. To continue to increase bit density, the logic circuitry, or CMOS, is now bonded to the NAND wafer through the formation of additional copper lines and pads.
As the layers become thinner with each node, resistance of the word lines becomes a challenge. This is addressed with a new material, molybdenum, which requires advanced atomic layer deposition technology to uniformly fill the thin word line openings. To add even more layers, the memory arrays can be bonded together, requiring more deposition and etch steps. The result is a structure with effectively more stories than five Empire State Buildings standing on top of each other, and even more in the future.
Beautiful. It's easy to understand how increasing the number of layers is going to require a lot more deposition and etch steps, but hopefully, what that video did was also show you what we've been saying for the last several years, which is the complexity of building those taller NAND stacks requires new types of equipment, and as the industry is currently at this point where two-thirds of the bit capacity is below 200 layers, we see our customers needing to spend over $40 billion over the next several years to upgrade that existing capacity to higher layer node counts. Demand will grow for many of those new tools you saw in that video, like backside deposition for stress management, carbon gap fill for tier stacking, and bevel deposition for array bonding.
These are just some of the examples of tools that do not exist below 200 layers, but will be needed as customers increase their layer count. I talked earlier about the fact that the technology is moving towards our strengths. If you take an example of something like the backside stress deposition, a backside deposition for stress management, we are the process tool of record at all customers running 3D NAND with more than 200 layers. To put it simply, we believe we are the best positioned company for the coming NAND technology upgrade cycle. Now, I've spent 30 years focused on deposition and etch technology. In fact, in a few weeks, I'll be at an event in the Bay Area that Lam hosts every year, celebrating with other Lam employees who are reaching their 20, 30, or 40-year milestones with the company.
And I truly think that one of the things that makes Lam very special is this long tenure of our employees. They've seen the history of the industry. They understand how it works. They understand cycles. But what I can say from all my time in the industry is that rarely do you come across a story like 3D NAND or the game-changing impact that it had on Lam. So what we're very excited about today is what's happening right now in DRAM and Foundry Logic, where it looks like history may be repeating itself. We're seeing 3D transitions begin to take shape in both DRAM and Foundry Logic. And our extensive learning in 3D NAND is giving us a unique jumpstart on the competition.
With the major architectural changes that are ahead in DRAM, we expect a 1.7x increase in Lam's SAM per wafer as we move into the future. Now, today, Lam already plays an important role in DRAM scaling. You probably saw last month that we announced our first dry-resist production win for EUV on 6F squared DRAM. Also, AI is driving tremendous demand for high-bandwidth memory and, in turn, strong growth for our TSV silicon etch and copper plating tools. Today, all leading edge AI-enabling HBM devices are being manufactured using our equipment. Now, stacking DRAM chips on top of DRAM chips is clearly a move to 3D, and Lam's SAM has stepped up as a result, and the road ahead towards 4F squared and 3D DRAM holds even more exciting opportunities for Lam.
4F squared combines both 2D shrink, but also now a vertical scaling element to increase bit density, to improve power efficiency, and also enhance reliability. 4F squared is a marker on the road to 3D DRAM, where the game really changes in favor of deposition and etch companies like Lam. So again, let's take a look at how we see DRAM manufacturing evolving. In the 4F Squared approach, the word lines and bit lines are rearranged into a more efficient layout. This architecture requires high-aspect ratio etch to form channels, which are then filled using atomic layer deposition, or ALD. In addition, a CMOS wafer is bonded to the backside of the array, which requires copper plating.
As memory density continues to increase, the shrinking word line runs into a challenge with resistance, creating the need for the atomic layer deposition of a new material, molybdenum. When 2D reaches its scaling limit, the roadmap calls for 3D scaling. Here, the memory cells are effectively rotated by 90 degrees and stacked upwards. This requires high-aspect ratio etch through microns of alternating material, akin to channel hole etch in NAND, followed by a lateral etch that forms indentations in each layer that subsequently need to be filled using atomic layer deposition. Innovations like these will enable DRAM density that is equivalent to squeezing 500,000,000 people onto the island of Manhattan.
We got just a couple more New York references for you guys next couple of videos. Now, as you saw in the video, DRAM's moved to vertical scaling. It's definitely going to increase our opportunity. If you take just one example that was shown there, the move to build the CMOS logic on a separate wafer and bond it to the memory cells to reduce the total chip area, that change itself is going to more than double the metal wiring layers. Also, in 4F squared, the critical aspect ratio might not have been so obvious from the video, but it increases by about 50%, which requires us to now implement new void-free conformal fill and etch capabilities. And those capabilities themselves have to have unprecedented depth uniformity and profile control, all things that Lam is really good at delivering for our customers. I don't want to steal Seshasayee's thunder.
He's going to come up and give you a lot of detail on products. But our products really have been targeted towards these new challenges. He's going to talk about a new etch tool today that we're launching that we believe is a true breakthrough. The new capability that we'll talk about will also be critical as the industry moves towards 3D DRAM, where the vertical height, as you saw in the video, will now increase to about 40 times what conventional DRAM is today. Sesha will go into those details in just a bit. But before we get to that, of course, we need to take a look at what's happening in Foundry Logic. Node scaling is enabling node to node about a high teens % growth in transistor density. But when you listen to what AI needs and you think about compute growth requirements, it's about 40%.
Very similar to what's happening in DRAM, Foundry Logic is rapidly moving towards 3D structures like gate all around and backside power distribution to achieve greater density scaling, but also to improve performance and power efficiency. This is creating, again, a compelling SAM expansion story for Lam. We're seeing new materials and new deposition and etch technologies being needed to form those complex gate all around and CFET structures. In our estimation, this will help drive a 2x increase in Lam's SAM per wafer in Foundry Logic. Now, earlier, I highlighted that in the case of high bandwidth memory, advanced packaging is playing an important role in DRAM. Foundry Logic is really where advanced packaging started, and it's still where the industry is being led with the most complexity and the greatest need for advanced solutions.
For instance, in wafer bonding, alignment becomes much more critical, which requires new backside deposition steps to flatten the wafers and enable proper alignment. Now, if you've been paying attention, where did you just happen to see that as well? In NAND. And so, again, we're able to take a lot of the learning that we've gotten in 3D NAND, and we're starting to be able to port some of those solutions over to help us extend and solve problems as DRAM and Foundry Logic go 3D. Also, new bevel deposition steps are needed to prevent gaps from forming where the thickness of the two wafers may not quite match.
And so we can help eliminate those to improve yield in Advanced Packaging in Foundry Logic. Now, we said recently that our shipments for Advanced Packaging for Foundry Logic and HBM exceeded $1 billion in 2024 and continue to grow. This is an example of how 3D inflections are having a big impact on our business, and I'd like to take a quick walk through some of the other changes happening in Foundry Logic. Building the Gate-All-Around structure requires a number of specialized atomic layer processes, such as selectively etching between the channels one layer at a time, and using ALD to form the spacer in between.
These processes become even more difficult at the next node because the device gets taller and smaller. In addition to shrinking the transistors, the wires that connect them together also need to scale. When building such small wires, only dozens of atoms across, many things can go wrong. New EUV patterning technologies reduce these errors and defects. Another approach to scaling is to reduce the complexity of the wiring. Traditionally, the power and signal wires are intertwined. Moving the power wires to the wafer's backside reduces congestion and improves performance.
This requires even more copper plating. Back at the transistor level, the CFET inflection will require even taller structures and additional wiring with more deposition and etch, which will enable the logic scaling journey to continue to the point where there are more transistors on a single chip than 50 times the number of pizza slices eaten in New York City in a year.
Okay, that might be the last New York reference. Now, what's interesting is people will often say, "Tim, but other deposition etch companies, people do have other products." They'll be focused on these same opportunities. But what you saw is things like backside power. They're deploying more copper plating, copper plating a traditional strength of Lam. And so again, many of the technology inflections are not only creating greater opportunity for us, but as I said earlier, in areas where Lam is already a market leader. And that's really what is so exciting. Now, you saw a lot in the video. From the perspective of Gate-All-Around, it's really the first true 3D logic transistor structure. And as you saw in the video, it introduces the need for new tools where also Lam is doing well: atomic layer deposition, atomic layer etching.
The contact etch for gate all around, it needs 100 to 1 selectivity, which is five times higher than what sorry, which is roughly 20 times higher than the 5 to 1 selectivity that's required today. And so you're really seeing not just continuous improvement, but true innovation and true disruption being needed to enable these new, much more complicated 3D devices. You also see the increasing need for things like surface preparation, surface engineering, as these nanosheets become so small that there's only a few remaining charges left in that transistor. If you don't prepare that surface correctly, you damage the surface performance of that transistor and ultimately yield suffer results. So we're seeing greater demand for some of our tools which help with surface cleaning, surface preparation. And I think that's something that will continue to grow as these 3D structures in Foundry Logic continue to shrink.
So beyond Gate-All-Around, the industry is going to move to CFET. And there, the transistor stacking, as you saw in the videos, is expected to double the depth and aspect ratio requirements, again, creating longer deposition steps, longer etch steps, and ultimately more served market expansion for us. So there's a lot on SAM expansion. Hopefully, you've seen that across all device types, there is a growing need for deposition and etch. We can confidently say that we're gaining traction with our new products at multiple customers.
And as we look forward, our target is to capture more than 50% share of the incremental SAM that's being created by the inflections. It's a big number. But as I said, many of these inflections are occurring in places where Lam has already demonstrated market strength, whether through 3D NAND or other applications. And so we have confidence in that number. And Sesha will also talk about why the products are specifically designed to ensure that we capture that share of that incremental SAM. So with that, I'd like to bring up Sesha Varadarajan, head of global products at Lam. Sesha.
Good morning, ladies and gentlemen. A real pleasure being here today. The good news is I don't have any New York references in my presentation. So with that, I think as Tim said, he walked you through how Lam's share of WFE is growing in the future inflections in NAND and DRAM. I'm here to talk about some of the products that underpin our confidence that we will outgrow that growing fraction of WFE. It starts by developing differentiated products that uniquely address scaling challenges for the three device segments we participate in: NAND, DRAM, and Foundry Logic. One thing to note is we don't call advanced packaging out separately. At Lam, we believe scaling for all three device types requires advanced packaging crucially to achieve those. So our product plans for advanced packaging are baked into each of these device segments. Let me first start with NAND.
As Tim explained, the challenges to scale NAND are getting harder, and with our unique strength in etch and deposition, we have a vantage point to help address these scaling challenges. We have a twin-pronged strategy for outperformance in NAND. One of them is around developing upgrade packages for technology scaling and productivity for our installed base. The other piece of it is as you scale NAND, we run into critical roadblocks that require unique new products to address, and we are developing many of those industry-first technologies as well, as I will go through. This combination is very important. Our customers have challenging economics for NAND. So this dual roadmap investment becomes the most efficient way for our customers to continue scaling their device roadmaps, and scaling device roadmaps for our customers are critical.
It's critical because as the NAND market shifts between different segments, enterprise, storage, or consumer, the form factor of the memories and the performance requirements are different. So they do have to upgrade. And upgrading also reduces their bit costs. So that, even in the face of the challenges of NAND industry, continues. So let me jump in and walk you through our product offerings here in NAND. We've been, as Tim mentioned as well, the leaders in foundational technologies for NAND for over a decade now. It starts with Strata, which is our plasma CVD tool that deposits the O and N layers. Flex and Kiyo are critical etchers that do the mask open and the channel hole etch itself. And the Altus tungsten product does the word line or replacement gate formation in all the NAND devices.
The only reason I mention these are obviously they are the installed base that we are focused on moving forward. But those foundational positions also give us a unique vantage point to anticipate future challenges and how best to address them. I'm excited with our NAND portfolio today. It's the richest and most comprehensive it's ever been. And they comprise a number of industry-first technologies, or in some cases, like carbon gap fill that I'll cover today. It's literally an industry-first product. I can't cover every product here, but there are two products I'll dig into to show in here. One is Vantex, which is our high aspect ratio etcher. I'll walk you through what the differentiation is and how it addresses customers' needs. And then I'll also cover the carbon product that Tim mentioned, both in the earnings and also in his talk this morning.
So let's look at our high aspect ratio etcher. Obviously, well known that we have been in the industry leader for 10 plus years. A rich history of innovation, lots of firsts. We were the first to bring cryogenic etching into high volume manufacturing in 2019. Its technology by itself is not the value add. It's the productive technology implementation that really separates Lam. So over the last several years, we have been focused on how to build on that strength. Recently, a few months ago, we introduced our latest cryogenic etch technology, which we call Cryo 3.0. I wanted to spend some time here to peel that back and explain what makes that go. As always, it always starts with hardware, a differentiated piece of hardware. So at your left, there is a cross-section of our latest generation etcher, a couple of unique capabilities in that.
Generally, this is a capacitively coupled reactor. So a wafer sits on a chuck. There is a counter electrode at the top. We strike a plasma between them. For those who are familiar with our industry, for 20 plus years now, the chucks on which the wafer sits, they tend to have multiple zones of control for temperature. And it's done just to achieve uniformity. What we have done in this product, actually, is to take the upper electrode and segment it and independently control RF. And that's the first time anybody has done that in the industry. And the reason why that's important has to do with a rather esoteric metric called tilt. When you etch channel holes, you want it to be vertical. The reality is, for many number of reasons, there's fractions of variation, fractions of a degree variation in them that we characterize as tilt.
Tilt is bad because effectively it takes up 2D space. So you can think of it if something leans this way, even though the CD is small, the footprint on the wafer is very big. So you don't want tilt. As the NAND scales taller, that same fractions of a degree, which was okay two years ago, not okay now, because it's taller, so the same amount of tilt takes up way more space. So by having this multi-zone upper electrode, we can actually control tilt very tightly across the wafer. And that's something that we are very proud of. Another one called out here is the industry's highest voltage. We are driving down ions into a tall hole. More voltage means ions go deeper. Seems pretty obvious. Like everything else, the devil is in the details.
The amount of voltage we are talking about here, the power density in our latest etchers, the one here that I'm talking about here, it's in the order of magnitude of a nuclear power reactor. It's obviously not continuous. It operates in time segments, but still, it illustrates the material challenges associated with bringing something like this to the market that we have basically overcome, so these types of differentiated hardware features combined with our latest process that we call Cryo 3.0, this package is with our leading customers. And the feedback so far has been very positive. Some of the benefits they're seeing, profile deviation reduced by 80%. Just profile deviation means as you look at a channel hole, there is variation in CD for different reasons, and you want that to be as low as possible.
Several of you understand one of the big drivers for NAND right now is around enterprise storage. They are built on programming QLC and things like that that cannot deal with channel width variation because it induces programming errors. So something like this becomes very attractive to our customers. We have done this with also a 90% reduction in the carbon emissions footprint for the product. So altogether, we're very pleased with the trajectory thus far for the adoption of this. I talked to you about a twin-pronged strategy. So the other aspect of it is the installed base itself. Over the last four or five years, the NAND wafer starts have maybe somewhat decreased. But our cryogenic etch installed base, predominantly based on conversions of our installed base, although there are some new tools as well, has actually been growing every year.
For the most part, this is based on our first and more recently our second generation cryogenic etch technologies. I think we disclosed that by end of 2024, we have almost 1,500 chambers worth of cryogenic etch in the market, by far, by far the leader in terms of production expertise there. As we look into the future and we expect to port our third generation technology over to our installed base, this trend of cryogenic etch installed base growth will continue. We anticipate almost a 3x growth from today within the next three to five years. That's part of what Tim mentioned as this huge conversion WFE that we anticipate for NAND. I want to switch over to another product that showed up in the video. I think you might have noticed this carbon plug, which is used for stacking.
It might have not been obvious there, but this carbon plug is actually a sacrificial film. It's there so that when you do the second tier and etch, it has to stop. It has some other functions as well in terms of holding the mechanical form factor and such. But let me walk you through how this product works. Before I start, a common theme that you will see is a lot of plasma and RF innovation. This is a relatively niche field. Only our industry uses it. What goes as RF more generally in the world is predominantly communication. It's at gigahertz. This tends to run at a way lower frequency. But this is core to our differentiation. At Lam, we have thousands of man-years' worth of expertise built in.
And we view our ability to bring complex and continual RF innovation to market as one of our building blocks for differentiation in the market. So in this case, the product has two RF innovations. One is a multi-frequency plasma. It's combined with a chuck that has multiple zones at very high temperature. Rather than going into details of that, what I want to guide you towards is how we use this. In this case, we exploit this differentiated hardware to set up simultaneously two different species, two different profiles on the wafer. So it's kind of illustrated with that golden and green lines. So passivation species there, think of it as something that doesn't deposit as much or prevents deposition as much. And green line is active deposition. And you can see there we have set up the plasma in such a way that you have two different profiles.
The gap between the two profiles is what we exploit to achieve carbon gap fill. This product, Tim mentioned in the last earnings call, it's something that we are seeing traction and revenue even starting in 2025, and very strong reception from our customers. Our NAND customers are reporting maybe 50% reduction in process steps compared to legacy approaches like CVD plugs or even spin-on carbons, and also seeing other benefits like reduced CD loss. That actually turns out to be quite important. After you finish processing, you have to exhume this plug, and when you do that, you don't want to open up the CD because it's effectively losing 2D space. Because it's carbon, it's very benign to de-plug, to use that term, without losing CD. That's some big driver for adoption of this technology. That's a NAND story.
We see use cases for this in DRAM and logic as well, and something we are very excited about to continue building our leadership on. Let me switch now to DRAM. While we have a very rich NAND portfolio, as I showed you, we've been working for many years on building a disruptive set of products for DRAM as well. Evidence of early success there, our HBM leadership position that Tim talked about as well. Syndion, which is our critical silicon TSV etcher, market leader for HBM. In fact, perhaps every HBM chip out there or nearly there runs through our Syndion etchers. And the gap fill for that is done on the Sabre plating tool as well. Our strategy for DRAM is kind of built on top of that leadership on HBM by introducing new products to enable the future 4F squared and 3D DRAM.
As Tim mentioned, they look a lot like the 3D that we are very strong on. Let's see how that product portfolio looks. Our foundational products, our strengths are in patterning and metallization. I also talked to you about HBM already. Looking into the future, we have an exciting set of products that will help the industry transition from 6F squared to 4F squared. 4F squared itself takes the gate and makes it vertical, just the gate portion, and then takes the CMOS array and makes that vertical by stacking it on top. Both kind of leaning on some of the things that we have developed with NAND. You'll see products like high aspect ratio etch and ALD fill here. But at 4F squared, you still have 2D shrink in DRAM.
So our dry photoresist that I'll get into with more detail today, again, it's a very exciting technology in being able to do unique layouts for DRAM in the future. So in the case of DRAM, I want to maybe peel back on two products. One is our Aether dry patterning system. And the other one is Striker ALD, which is our advanced gap fill product. DRAM actually has a lot of critical gap fill steps. And traditionally, gap fills and spacers, you deposit. And to get the electrical quantity you want, you do things like annealing. And those types of approaches have run into a limitation. The actual width of some of these features are so thin, the mere act of annealing something and the shrinkage that it induces ends up toppling features.
So there's a pressing need in the industry for better approaches that gap fill electrical quality films without needing any pre or post-processing. And that's what we have done here. Traditionally, ALD, rather simple to understand, there's a precursor. You purge it, you convert it, and you purge it. A four-step approach. In our reactor, we've introduced some additional degrees of freedom here. One of them is an ability to introduce something that we call as an ICE agent. It's an inhibitor. And we can control, not truly arbitrary, relatively arbitrary profiles on wafer for how that inhibitor is distributed. And then we use that to serve as a counter template for how we want growth to happen and have bottom-up fill. Sounds somewhat similar to the carbon product.
There are a lot of nuances here on why you need a much more complicated approach here, primarily because we're talking about electrical quality films, but it's our reactor design that has this immense flexibility to do multi-cycle ALD, the deposition interwoven with etch cycles or densification cycles, these types of very, very complicated cycles that helps enable this type of approaches. This product already established a leadership position in 6F squared, where we've been able to do very high-quality low-K films using this type of approach. We are also testing this product at 4F squared and 3D DRAM as well, where the number of applications for us is increasing, and of course, the challenges are more. The aspect ratio goes up, as shown here, but with that initial success that we have had, we are building incremental capabilities there to address the future needs as well.
The feedback from the customers is pretty positive on this measure as well. The other thing I wanted to talk to you about was Aether. Several of you probably understand this quite well in terms of how this has a potential to revolutionize resist for EUV scaling. Maybe something not that appreciated is it's actually a combination of tools. It's not just one tool. The first step in switching to a dry resist system is something called a Nimbus underlayer. So this goes down before the photoresist comes down. What we found is, rather amazingly, the quality of the patterning, whether it's measured by the efficiency of the dose or defects, is influenced by what's below the photoresist because the attenuation behavior of the photons that leak through the photoresist itself has kind of a memory effect back on the feature.
This Nimbus underlayer is critical for exploiting the full potential of the dry resist. After this comes the anchor leg, if I can use that term, which is the actual dry photoresist deposition itself. The photon absorber concentration, composition, distribution, the control that you get by a CVD process, way more degrees of freedom, lets us get an on-wafer result, which is very differentiated. Once we do the photoresist, it goes to scanner, gets exposed, comes out. We have replaced traditional solvent-based wet develop with a dry plasma develop scheme. Introducing plasma into develop was a rather counterintuitive finding. We've been able to get excellent results on collapse minimization by adopting technologies like this. Altogether, optimizing all three products, a lot more complicated, has taken a lot more time than we at Lam imagined when we first entered this market.
We are at a very positive junction. As Tim mentioned, we announced a leading memory customer has implemented this in production. This implementation validates a lot of our hypothesis behind introducing dry resist. And we are still in the early stages of this concept being adopted. Obviously, DRAM, many logic customers also testing it. And over the next five years, we are targeting cumulatively $1.5 billion in revenue for this product. Actually, now that I mentioned logic, maybe a good segue into my next segment, which is on why we think we'll win in logic and foundry. The push for advanced logic device formation driving the need for innovative architectures that favor our portfolio. We are leveraging our expertise in patterning, where we are market leaders with our conductor etch family of products and metallization to deliver cutting-edge solutions for these new architectures.
So let's take a look at what those products are. One thing maybe not easily appreciated, I think people understand the depth of our portfolio, how critical our HR is. And through today's talk, maybe we want to convey a sense of how the breadth of our portfolio has improved to address these various inflections across the three device segments. Over here, based on our foundational products, several new products that address the transitions that Tim talked about in logic. For example, Gate-All-Around. Gate-All-Around, you think of it as something, of course, the scale is different, but something similar to NAND. Effectively, you deposit a mold, you exhume, you replace. So in this case, it's silicon, silicon- germanium, epitaxial stacks. You have to exhume the silicon germanium out and gap fill that.
That lateral processing, Prevos, it may be a product you haven't heard of that much, is our selective etcher that does that silicon germanium exhume. And one of the challenges with exhuming silicon germanium is you have selectivity, but there is an unintended consequence. You can't leave behind the byproduct of the exhume somewhere in the wafer because obviously it's the channel. So in our case, we have developed what we call as an industry's first salt-free process. And it has to do on a rather unique hardware. Our wafer sits on a chuck that can actually vary temperature very rapidly, even within a single processing step. So at different temperatures, we activate different reaction mechanisms. And that's how we achieve our salt-free process. And then ALD gap fill, I think I already covered that in DRAM as well. So not necessarily.
But in this section, I want to focus on Sabre. A product near and dear to my heart. I started as an engineer at Lam working on this product. And it's the industry standard for electroplating. Over the last few years, whether it's because there are increased number of wiring layers or because of new architectures in packaging, our overall opportunity for this product is more than doubled. We have over 10,000 chambers now in the market and firm leadership position. And that leadership position is based on a number of breakthrough technologies, some of which I'll cover today. If I can focus your attention on your left, talking about chamber design. When it comes to plating, broadly, there are two ways you can do plating. You can spin a wafer and have fluid impinge on it. Technically, it's called a rotating disk electrode.
Or you can hold a wafer and run the fluid across. And the various terms for it, it's a paddle plater, not a rotating disk electrode, effectively. And there are lots of intricacies on why you would choose one or the other. But in our case, we built the industry's first hybrid plater. It is an RDE. It takes all the benefits of RDE in terms of uniformity of the limiting current across the wafer. But the TurboCell concept, the crossflow element of it, lets us improve our mass transfer and increase the limiting current. Lots of things, perhaps not that interesting. But what it means to customers, maybe easier to appreciate that. Logic structures are rarely one CDs. There are a range of CDs. There are small features, big features.
And when you bump something and you're trying to join them together, it's obvious that you want them to be at the same height. Otherwise, you're not going to get the connection that you're looking for. So this type of high convective process with very high limiting current lets you achieve very good coplanarity with very good productivity. And that's been a key driver of why we've established a leadership position here. A couple of technologies which are more forward-looking. I think many of you know HBM DRAM. It takes a lot more silicon for the same bits. And there are two reasons why HBM tiles are bigger. One of them has to do with the fact that each tile has to be addressed. So there's circuitry there. So leave that aside. The other reason the HBM tiles are bigger is because you need a TSV.
There has to be a keep-out zone around TSV for a couple of reasons. One, if the TSV is scaled in such a way, its CD is reduced, the keep-out zone can be managed better. That's obviously value-add for our customers. As you scale TSVs, challenges with both critical edge and gap fill. Over here, we are talking about a new approach called electrochemical ALD that lets us bypass a requirement for a conductive seed, grow directly on a liner, a conformal process, and then set up a gap fill. It's something that we are actively researching and developing with our customers for future generations of HBM. To your right, we talk about grain engineering. This is tied to hybrid bonding.
You can think of hybrid bonding as a direct metal-to-metal connection or direct metal-to-metal in this case, and effectively avoiding a medium in between like a bump. Because when you have that, it's a lateral spread. So you want it to directly connect. When you do that, the quality of interface is crucial. So we've been able to control the metallurgy of what we deposit. In this case, something we call fine-grain copper. What we found is when we do it in this particular way, the quality of the interface and the wiring is the best. And what it means to our customers is they're reporting over 2x increase in the bonding strength. And that's critical for the reliability of the packages itself.
Across a range of opportunities for plating, lots of breakthrough technologies and more to come, as I showed you a preview of a couple of new technologies here. With that, I'm going to switch over to the main portion of my presentation, which is around two new products that we're going to be launching today. Throughout the course of Tim's presentation and mine, these two products, one is on etch, the other deposition, have shown up as a critical technology that is needed for scaling all three device types, something that's applicable across the fab industry. Let me start with the deposition product itself. It's about ALD Moly. We are the leader in tungsten ALD deposition and have been for two decades. Believe it or not, we started developing a replacement for tungsten maybe seven, eight years ago.
So we knew eventually Tungsten is going to run out of steam. And we focused on Molybdenum. And with our early investment, we have ended up with a very differentiated reactor and market position for Moly. There's a rather interesting story behind how you ended up with Moly. Everybody has a periodic table. I have one in my office too. And you can pick any number of materials that have low resistivity. But resistivity itself is not what you choose a new material for. Industries take a long time to make a transition to a new material, seven, eight years. It takes a lot of ecosystem changes to enable it. So when we pick new material, we need to consider a lot of other factors about how it's integratable. In this case, Moly is very favorable for being relatively edge-friendly.
Both from a dry and wet perspective, its removal characteristics can be optimized somewhat analogous to how Tungsten has handled itself. So yeah, there's a lot more that goes into choosing a new material. But let's take a look at what this groundbreaking system looks like. So we have a short video for you. Hopefully, that'll give you an idea of that. More than 20 years ago, Lam Research pioneered Tungsten atomic layer deposition in high-volume semiconductor manufacturing. Since then, we've continued to push the industry forward, leveraging our deep expertise in ALD hardware and process requirements to enable the transition from 2D to 3D NAND and developing our Altus system into the industry benchmark for Tungsten-filled productivity.
Now, as the demand for memory capacity and computational capabilities surges with the growth of AI, scaling requirements on the path to 1,000 layers are beginning to exceed the limits of Tungsten-based metalization schemes. Across NAND, DRAM, and Foundry Logic, chip makers are transitioning to Molybdenum or Moly interconnects due to the material's ability to provide low resistance and barrierless integration. At Lam, we're working to drive this industry-wide shift with Altus Halo, the next generation of our trusted Altus line of products and our latest advancement in atomic layer deposition technology. Built from years of experience and research, Altus Halo will empower chip makers with capabilities to continue innovating by providing solutions to address the unique challenges of Moly implementation across the leading-edge integrated circuit types. Altus Halo delivers the most precise and advanced deposition of Molybdenum in the semiconductor industry.
The Altus Halo tool series is optimized for a range of metallization needs, with the ability to deposit conformally or selectively with bottom-up feature fill using chemistry and thermal flexibility, as well as plasma for temperature-sensitive applications. At Lam, we believe Molybdenum is the next frontier in semiconductor metallization and a key component for powering the AI era, and with Altus Halo, we're bringing all its advantages to today's chip makers, proving what's possible for our industry and the world. Yeah, so pretty exciting. I've actually personally been involved in this program right from the inception, so I'm quite satisfied with how far along we have come. Our early investment in this has led us to develop a product that's quite differentiated, so let me actually walk you through one. In the video, if you noticed, we talk about barrier-free integration.
So one of the ways Moly achieves lower resistance beyond its intrinsic properties is the fact that the metal cross-section in the lines is bigger. So specifically, Tungsten needs titanium nitride barriers underneath. For example, Moly doesn't. So the cross-sectional area of the conductor is increased. So that's the claim. But the reality was, while it's true that Moly can be deposited directly on dielectric, the quality of the interface wasn't good enough. So very early on, we discovered a unique sacrificial liner that you could do. The idea is you do a few monolayers of this sacrificial layer. And subsequently, when you switch to the actual ALD metal deposition, right during the process, this sacrificial liner gets converted to pure metal. And that's how we achieve this true barrier-free integration.
That's a patented technology that's tied to our architecture that we showed, where we do different steps in each portion of our reactor. The barrier-free deposition itself, depending on the device, is getting almost 50% reduction in resistance. People will do no changes for 10%. 50 is like a generational leap or beyond. We also talked about this wide process window. Although the video doesn't truly explain it, the non-DRAM logic, same metal, but the form factor and the envelope in which the process has to fit is extremely varied. We're very proud of the fact that the process window or reactor is large enough to deal with all of this. The transition to this new metal has already begun. I think Tim mentioned that during the earnings call last year, last quarter as well.
With the increased number of applications across the three device segments, our opportunity is two times what it was during the Tungsten era. We are targeting over $2 billion in shipments over the next three to five years for this product, and something that we are very excited about. Here's an example of a customer, in this case, Micron, that took our leading Moly capability and integrated it around a new product that they're doing. They leveraged it basically for increased input-output bandwidth and also storage capacity associated with their high terabit packages that they're developing for their customers. This is an example of NAND. Obviously, we are continuing to work with customers and logic and in the future DRAM as well to realize similar value propositions. Moly was exciting. We talked about the two-decade leadership and then making the transition.
Let's talk about another one, in this case, etch. And in this case, we are talking about our conductor etch product. We've been industry-leading for 20 years, all the way back to early stages of 300-millimeter, where we were the first to adopt Waferless AutoClean. And most recently, we engineered ALD deposition into our conductor etchers to improve profile characteristics of what we etch. A long history of innovation. But as Tim showed, the demands on conductor etch are actually increasing and accelerating, looking into the future, whether it's 4F squared or 3D DRAM, a number of high-aspect-ratio silicon etches. And that challenge led us to develop this groundbreaking new product. So let's roll this video. Future advanced logic and memory technologies demand extreme etch precision for devices with smaller geometries, increasing aspect ratios, and more demanding profile requirements.
To shape these critical features, the etch must prevent profile failures, such as bowing, tapering, and irregular edge depth loading. This performance must be delivered with the highest levels of production efficiency, wafer to wafer, and run to run. To address these challenges, a generational leap in conductor etch capability is needed. And we're delivering this with a revolutionary product that extends our history of etch excellence. Introducing Argos, a new conductor etch tool for Lam's Sensei platform. Key to this innovation is Lam's Direct Drive solid-state plasma source, an industry-first technology that is 100 times more responsive than existing technologies. Featuring Tempo, an ultra-fast plasma pulsing technology that generates ultra-high etch selectivity. And Snap, a breakthrough ion energy control technology that shapes etch profiles with atomic precision. These advanced capabilities enable Argos to meet the critical etch challenges of current and future devices.
With Argos, we're eliminating the bottleneck of slow plasma response by leveraging solid-state technology to enable precision, repeatability, and productivity. Feature to feature, wafer to wafer, and run to run. The result is an industry-leading solution that ensures product matching with angstrom-level repeatability. Integration of Argos with the Sensei platform also enables automated replacement of consumables for lights-out fabrication and minimizes defects for higher yield rates. We're talking about a revolutionary step forward for Lam and our customers, one that builds on the proven legacy of our industry-leading conductor etch products and raises the bar to all new heights. With Argos, we're helping customers achieve results that were previously unthinkable. Yeah, we're very excited about the breakthrough technologies that are baked into Argos.
Some callouts from what we showed in the video, Direct Drive, which breaks the, I don't know, 30-year paradigm of using a match between a generator and a chamber and Tempo. Maybe one thing to elaborate on from the video is when you think of reactive energy chains, you think of it as, yeah, I sent down ions that remove material. But the reality is there are three distinct things that happen in the etching process. One is the actual removal of etching. There is also passivation, which is effectively protecting where you don't want to etch. And the third is the removal of byproducts. And traditionally, all three of them happen all the time. And etch recipes are super complex because of the fact that you are dealing with three different phenomena here.
With our new hardware and Tempo pulsing, we're talking about controlling plasmas in microseconds that actually lets our engineers visualize all three steps independently and optimize for every one of them. And since it's happening in microseconds, it almost looks continuous. It is continuous in terms of how the profile evolves in the wafer. So really exciting. The feedback from our customers for this product is outstanding. And bringing technologies like this is always in conjunction with our customers because we have to work with them to understand the future challenges and also work with them to exploit the features that we are bringing to market. Argos, already several DTOR, PTOR wins across leading DRAM and logic and moving into volume shipments this year itself.
And as we look further into future scaling needs, whether it's CFET or even 3D DRAM, we've built so much extendability into our Argos product that it's going to take us well into the era past these transitions for the next 15-plus years, at least. So with that, I ran over a bit, but perhaps because of my enthusiasm for what we are building. But regardless, it gives you an overview of our differentiated products. And then I asked Tim to come back on stage to talk about how we are going to extract differentiated value from our installed base. Great. Thank you, Sesha. A couple of very exciting products, a whole lot of great work being done by our product teams.
And as Sesha said, really, one thing I've always talked about with many of you is that our R&D is somewhat deterministic, meaning we engage closely with our customers. We understand their challenges. And what you might have noticed on some of Sesha's charts, I hadn't noticed it before we kind of did the run-through. You see a lot of things in the new category. And then in the future, they're actually almost every time there were fewer. And I think what that indicates is we just don't yet know what those next challenges are going to be. But we know there will be some. And customers, as they work through these new technologies, will continue to identify new innovation opportunities for Lam. Now, I have the opportunity here to wrap up a little bit. But we've talked a lot about our SAM expansion.
Hopefully, you've seen that this move to 3D is a fantastic opportunity for Lam. Sesha talked about our share opportunities, I think, with these new products and the way we've developed them closely with customers making these transitions. Fantastic opportunity. But of course, we can't have an investor day without talking about our customer support business group, or CSBG. Our installed base business is an incredibly important part of Lam's story, and we've grown our installed base. We said on our last earnings call to about 96,000 chambers, and this number of chambers increases every year. By its nature, the installed base business is more immune to industry cycles, one of the reasons why we really like it. Since 2013, our annual CSBG revenues have actually only been down once, despite multiple WFE down cycles. CSBG revenues over that period of time have also increased 5x.
So as we look forward, we expect to, again, more than double our CSBG revenues in that trillion-dollar industry. Now, as we look forward, one of the most important areas of value creation and growth will be in upgrades. I talked earlier about this upcoming NAND upgrade cycle that we envision. The magic of upgrades is that upgrades enable our customers to cost-effectively scale their technology on the Lam equipment that they already have installed in their fabs. It's also the fastest way to get that technology up to the next node because it takes a lot less time for us to perform an upgrade on an existing system than to facilitate and install a new tool. Now, in some cases, customers don't want to upgrade that existing tool for that same application.
So instead, we'll engage them on an upgrade that allows them to repurpose that tool for a different application. And in that case, that repurposing creates a share gain opportunity for Lam. So whether it's the technology upgrade or it's this repurposing, one thing that's true is Lam tools rarely get retired. Now, Sesha talked about how kind of he started on Sabre. I actually started on Altus Tungsten. That was the first tool I worked on, first tool I installed 30 years ago. It's likely still out there running today. Our tools just don't find themselves obsoleted. And that's because of the power of upgrades. And it's about the fact that semiconductors are used over such a broad array of applications that those tools continue to be used. They continue to generate parts and service business for us for decades and decades.
Now, there's one additional type of upgrade that we don't spend a lot of time talking about, but it's becoming increasingly important, and that's the type of upgrade that boosts productivity or improves sustainability. And if you're a fab director, fab leader, any fab in the world, you're continuously under pressure to improve the cost of your operations and, more recently, to really focus on the sustainability of the fab, the green aspect of the semiconductor manufacturing process, and so we work with customers to help achieve both higher productivity and better sustainability with upgrades that can be put onto their existing tools throughout the product lifecycle. Now, semiconductors, as I've said several times today, it's a great place to be, but our business and our customers' business is becoming a lot more complex.
Whether it was the COVID pandemic or the supply chain shortage and crisis, the world recognized the criticality of semiconductors to just about everything that we use in our daily lives. And governments around the world took notice and began to pledge tremendous financial support for regional semiconductor manufacturing. At last count, I think we were up to somewhere in the range of about $300 billion of government spending that had been announced around the globe. Now, frankly, as an equipment supplier, we hear numbers like $300 billion, we get pretty excited. We're happy to see such strong interest in semiconductor manufacturing. But it is creating a number of challenges both for ourselves and our customers.
One, with all of that regional interest, there are going to be a lot of fabs built in different parts of the world, many of which will not have had semiconductor fabs before or certainly not for a very long time. And this is expected to create a significant shortage of skilled workers, some estimates as much as a million workers needed by 2030 to support all of that growth. And on top of this, you can imagine that the leading-edge fabs now cost tens of billions of dollars to build. And so those customers making that investment really require very rapid ramp of those fabs, shortened time to achieve yield so that they can deliver ROI from that investment. I think you've gotten the message today at Lam. We see all of these challenges always as an opportunity for us.
In this case, we see the opportunity to completely transform how we think about customer support and, in doing so, accelerate growth in our services business going forward. Our efforts are really focused in two areas. First, we're using the deep domain knowledge that we've gained over all these years and over all those systems out in the field to develop machine learning algorithms with AI to improve tool performance and solve problems on our tools faster. For example, in one case, we're using signals of metal species that are coming through the chamber seasoning process to make a data-driven decision when we can switch from the seasoning process into production. By using that data, we can actually shorten the time to get that tool back up into production. Again, higher productivity for our customers.
In other cases, we're using data to predict when a part might need to be serviced or replaced, and that helps fabs avoid unscheduled downtime, which is quite costly for them, and in other cases, we're using AI to match chambers, and often, that's chambers matching that's occurring during startup within one fab, but now, with this global footprint and customers building fabs across the world, that matching can also be helped to ensure that they achieve the same results in fabs that may be located in different continents. Second area of focus is on next-level service precision. In this case, we're focused on the use of collaborative robots or cobots. We announced Dextro, our new collaborative robot, about two months ago as the semiconductor industry's first maintenance cobot.
Now, just like in the AI and machine learning case, cobots rely heavily on the data that's coming off of our tools and the data that's being collected while they perform this maintenance. And they actually become better. They learn and adapt and continuously improve so that the maintenance they're performing really is achieving results that are beyond what a human engineer alone could be achieving on that same tool. And this precision service and precision maintenance by the cobot is driving predictable, repeatable performance on our tools. And so I want to show you a video of Dextro in action. As technology advances in semiconductor manufacturing, our customers require us to deliver precision and accuracy at a level that has never been achieved at scale before. Collaborative robots, or cobots, can support fab engineers by automating repetitive and sensitive operations. Here are some examples.
For etch chamber cleaning, our cobots increase the safety of our employees by taking on the time-consuming task of cleaning the polymer buildup off the chamber sidewall. After cleaning, the cobot replaces consumable parts with a level of precision beyond what a human can do, leading to enhanced performance. And finally, for top plate installation, the cobot ensures that the bolts are installed in the right order with the right torque and tension every time. Lam's Dextro cobots provide a superior level of precision, repeatability, and first-time right, delivering better outcomes for results. And this is just one way that Lam is at the forefront of making the autonomous fab a reality. Great. And you can probably imagine who's the happiest about these cobots. It's our engineers. Nobody likes going in and having to do that manual work of cleaning those chambers.
Our customers as well because, again, it eliminates a lot of the downtime and improves the performance. Whether it's the use of Equipment Intelligence or it's the use of the Dextro cobots, we're seeing tremendous value being created. We're doing this through improving first-time right maintenance, again, back to that use of data and the precision of the robotic maintenance. We're shortening the overall maintenance time, again, using data and robots. We're also reducing unscheduled maintenance events. When you add these all up, it's a significant impact and savings on our customers' running costs. Just thinking about what we've shown you today, new technology, inflections growing our served market, the new products that are going to help us gain share, both of these are success in those areas.
They feed into faster growth of our installed base and therefore, again, an opportunity for our installed base to grow even faster. When you add on top of that the growth acceleration that's coming from the investments we're making in upgrades and advanced services, we believe we're on a path to more than double our CSBG revenues in a trillion-dollar semiconductor market. And so I really would like to allow Doug to show you how all of these elements come together: SAM expansion, share gains, installed base growth acceleration, and how they come together in our long-term financial model. But we're going to make you just wait a little bit longer for that. We built in time. If you take a little bit of a break, get some coffee, please, next door, you can go see some of the displays. And in about 15 minutes, we'll come back.
We'll bring Doug up, and he'll share with you the financial outlook for the company. Thank you. Please welcome Executive Vice President and Chief Financial Officer, Doug Bettinger. All right, everybody, find your seats. I know you just want the model, but you're going to have to wait for just a little while. I'm going to go through a little bit of history, but if you'll allow me a couple of minutes, I'm just reflective. You know, it's been almost five years to the day since we stood up in front of you at an Investor Day, and as Tim went through, so much has changed. So many different things in the world, in our environment, and frankly, at our company. I thought it'd be interesting. I went back and looked, actually, what was our revenue five years ago?
The March quarter we reported right after the last Investor Day. It was $2.5 billion. If you move forward, the next couple of years actually were phenomenal for our business. We peaked in the December 2022 quarter at the all-time record level of revenue for the company at $5.3 billion, and then that memory downturn showed up. Two quarters later, we were back down to $3.2 billion, and then every quarter since then, business has gradually built such that the most recent quarter we just guided was $4.65 billion, so we're feeling good about where we're headed in the short term, but that's not what we're here to talk about, right? We just talked about some of the most amazing product announcements we've had in the history of the company, frankly. We've never had a stronger product portfolio than we have today.
So we're going to get to the numbers, but if you'll allow me, I'm going to go back over a little bit of history of the company. But I'm always reflective when I get up at this point in an Investor Day. Tim had all these amazing videos, so did Sesha, some great customer testimonials. I got nothing cool like that. But what I think I do have is a review of a phenomenal track record of what this company has been able to do. So if you'll allow me a little bit, I'm going to go back over a little bit of history. My objective with the next four or five slides is to demonstrate to you what an amazing company this is and what an amazing management team this is, and give you confidence in how we're going to move this forward into the future.
So I'm going to go back, and the next several slides are going to start in 2013. Why am I starting in 2013? That was such a long time ago. I'm starting in 2013 because it was right after what I believe is actually the best transaction in semiconductor capital equipment history, the merger of Lam and Novellus. That occurred in June of 2012. So 2013 was our first full year as a combined company. So let's look back. What's happened since then? One, the semiconductor industry is a phenomenal place to be. Semis have grown at 2.9 times the rate of GDP. We live in an amazing neighborhood, is how I like to describe it. Now, if you think about breaking that down a little further, what has Semicap done? Semicap has grown at 1.7 times semiconductors. There's a multiplier effect showing up here.
Our SAM, etch and deposition, because of the evolution of 3D device architectures, has grown even faster than overall WFE, 1.2 times. These trends are going to continue in the future. What I love about the business is what Tim just talked about. The installed base business, CSBG, service, spares, upgrades, the Reliant product line, that has grown 180 times since 2013. Pretty amazing stuff, quite frankly. I'm going to walk you down our P&L here in a minute. The multiplier effects are only just beginning to show up in our environment. I think one of the things when you look back over a couple of five-year periods, which those pie charts on the right of the slide show you, is everybody's always historically thought about Lam as the memory company, and true enough, we love our enablement of what's going on in memory architectures.
But honestly, if you look at more recent history, our footprint in foundry and logic has meaningfully grown. 3D device architectures are showing up everywhere. Advanced packaging. You're beginning to see gate-all-around. Patterning process flows. Our broadening out in terms of the distribution of our revenue has expanded quite a good amount over the last five years, such that almost half of our revenue actually today, if you look back, has come from foundry and logic. So don't lose sight of this. This is kind of an exciting trend, and this was purposeful. A lot of those new product announcements, historically anyway, that come out of our R&D organization are winning business in foundry and logic. So that's pretty cool. Second, so WFE has outgrown semis. What I'm showing you on the left of the slide is kind of where has WFE trended?
In the 2024 timeframe, I'm adjusting out the amount of WFE that is inaccessible to us any longer due to restrictions in China. So WFE has been a great neighborhood, right? WFE has tripled. Pretty amazing. You can see where the growth has been. What the chart in the middle of the slide shows you is, okay, so that's great. WFE is up 3.3 times. DRAM has been great. Foundry and logic and NAND has also grown. In this timeframe, our revenue has quadrupled. So outperforming our industry. You can see where it's come from. CSBG is almost five times what it was a dozen years ago or so. Systems business has tripled, but that doesn't show you the Reliant product line. So systems, if you actually include Reliant, has done more than that. How's this happened? We've invested in technology.
We've brought great service offerings out. We've won incremental new applications and the business has grown. The multiplier effect keeps going. Remember, WFE outgrew Semis. Our revenue has outgrown WFE because of the growth in SAM as well as share positions. So that's pretty cool. But I really love this chart, actually. I just showed you that our revenue has outgrown WFE. We've been able to actually compound the impact of that revenue growth by expanding our gross margin. And if you look at it, it looks like it's accelerating because it is. You know, in early 2023, when business turned down, we began talking to you about re-architecting the company. I don't know if I'd call it restructuring, but there was a little bit of a pivot to enable the growth to show up close to where our customers were. We talked about that Asia factory strategy.
We talked about what's going on in Malaysia. In 2021, we opened that Malaysia facility. It's now the largest factory in our global network. I'm going to show you more about that as I go through some of my slides. But that's been important in terms of that acceleration in gross margin that you see. Admittedly, we've had some favorable customer mix, but a good portion and a sustainable portion of that growth in gross margin has been how we've pivoted the factory footprint. And like I said, I'm going to show you more about that. But this is only part of the story. And actually, when you look at it, this is actually pretty amazing. In calendar 2013, we had an operating margin of 15%. You can see it on the slide. By 2019, that ticked up to the mid-20% level.
In dollar terms, that was an expansion of 4.2 times. Let me remind you, today we are a 44-year-old company. You don't see performance like this from companies that have been around as long as we have. I can't think of too many other examples. But it kept going. In 2024, because of that gross margin expansion and our management of kind of the spending profile, we delivered north of 30% operating margin. That's 8.3 times in dollar terms of where it was in 2013. This is pretty impressive to me, all right? I'm a numbers geek, a finance guy. This is awesome. I love this stuff, and I hope you do too. In addition to that more efficient operating structure that's close to our customers, CSBG has done amazing. I love the CSBG business. It is less exposed to the ups and downs, the undulation of WFE.
We've also been able to pivot our spending profile and allocate more of that total spending pie to R&D. I'll talk a little bit more about that. And I just want to point something out. This didn't happen by itself. There was an enormous focus from the management team of this company to make this happen. More to come. But what's even more impressive, to me anyway, is what we've been able to do with earnings per share. So on top of delivering this compelling expansion of profitability, we've delivered even more in earnings. Operating income grew at a 21% CAGR. EPS grew at a 24% CAGR. It's a 10-bagger in terms of growth in earnings per share. And again, let me remind you, we are a 44-year-old company. You don't see growth like this from 44-year-old companies. This is what you see from super successful startup companies.
Feel great about what we're doing. So the reason I walked through some history here is this is a management team that has delivered, has focused on shareholder returns, has been extraordinarily focused on making sure we were growing the top line in a way that enabled an expansion in profitability. That combined with disciplined capital return has enabled us to really leverage the value that we're delivering for shareholders. So now that you've seen what we've done, let's look at how we've been able to do this. Again, I'm building confidence in what I'm going to show you is what the future is going to look like. Since the last Investor Day we had in 2020, we're a sustainably stronger business today. Let me walk you through how we think about how we've been able to do this. So first, it's all about revenue growth.
It's all about the right products. It's all about the right services. It's all about innovating, delivering what customers need, solving customers' challenges, expanding from our NAND leadership position, growing the business, diversifying the business out in foundry and logic as well as DRAM, and growing that annuity stream from the installed base, like Tim said. Our equipment really never goes away. It runs for decades. And the opportunity to continue to upgrade, sell spares, service, innovate with things like Equipment Intelligence and cobots, there's a large installed base out there. So that's an important piece of what we've been able to do and what we're going to continue to do. So revenue growth. Second, the global footprint. We've expanded our global footprint. And I'm going to show you a couple of slides about this. But this has been important.
We've prioritized investments in global infrastructure to be close to where our customers are. This has come through expansion of the manufacturing footprint. It's also, perhaps, we haven't talked quite as much about it. We've expanded how we do research and development. We've expanded our lab footprint. That's enabled a close-to-customer strategy. It helps us turn information quicker. It helps us win incremental business. Proximity to the customer is critically important, and actually, when you look at how we do this, this is a differentiator for us, I believe, so having that global footprint, investing in that global footprint has been an important characteristic of the sustainably stronger business, so revenue growth, global footprint. Third, operational scaling. This has been important, and the way I think about this is a couple of ways, the way we think about this. This is a growth business, but there's also a cycle here.
And so you have to have variability in the cost structure. You have to be flexible. I'm going to show you a slide that I really love in a couple of slides that demonstrates our current lows and a downturn are higher than our previous highs have been. We've been able to continue to expand the profitability profile of the company by maintaining this flexibility and investing in system infrastructure. I'm going to share with you and provide a little more visibility into what we call digital transformation at Lam. This is the next leg that's going to deliver expansion of profitability as we go forward. More to come on that. And then finally, investment in R&D. These new products don't come by magic. They come by a lot of engineering effort, and that takes investment.
One of the things I think we're pretty proud of is when you look at the percent of OpEx that we spend today, it's approaching 70% of total spending allocated to R&D. In 2013, that was in the mid-50s. So not only have we delivered leverage to the bottom line, we've done that by also being able to allocate more and more to R&D. When I share with you some of the stuff around digital transformation, we think that'll keep that engine moving forward. And so when you look at the investments we're making, it's pretty clear this has been very positive. We've got a sustainably stronger business today than we had a dozen years ago, than we had five years ago. This is going to enable us to continue to build on the business model. So let me dive in a little bit to each of these four pillars.
So, strong revenue growth. I chose to give a little bit of visibility into the revenue growth of CSBG. I love this business. We love this business. It's a business that, frankly, should grow almost every single year and has grown every single year. We give you the number of chambers in the field at the end of every calendar year. We'll always continue to do that. That defines an opportunity that continues to expand. If you look at the installed base, over this timeframe, it's grown at a cumulative average growth rate of 10%. That's pretty cool. Like Tim said, our equipment really never goes away. It just gets repurposed, redeployed, moved from one application to another. It's a great part of the business. When you look at, okay, so the installed base grew great. Dollars have grown even more rapidly, 700 basis points higher. So that's pretty good.
That's what we told you we were going to do in 2020, and that's what we've done over this timeframe. But a lot of you guys and gals point out to us all the time, "Hey, yeah, but you've got that Reliant product line in there," right? Give us a little visibility into what would have happened without that. So I'm doing that. It still would have grown, I don't know, 300 basis points higher if you adjust for the growth of the Reliant product line. So this is pretty cool. We told you we were going to do that. We've been talking about this. This is a business that, frankly, should grow nearly every single year. So we've been talking more and more about CSBG. So what is this going to do into the future? Tim already showed you, and I'm going to show you again.
We expect and will continue to drive CSBG revenue to grow faster than the installed base. Like I said, I have a hard time envisioning that this business doesn't grow nearly every single year, nearly every single year. I'm going to build very soon and show you a 2028 financial model and maybe something else too, but as part of that, you're going to see this growth from CSBG. I love this business. We love this business. 1.5 times by 2028. I know one of the sell-side guys has already told me, "Hey, this isn't good enough." It's pretty good. We think this doubles by the time the semiconductor industry reaches that trillion-dollar mark, so let me pause for a minute and share a little bit about what we've done with the global footprint of the company over the last several years.
I'm going to break this into two pieces, and the world map shows you every piece, but first, let me talk a little bit about what we've done with R&D. Several years ago, we came to the view that it was critically important to have labs close to where the customers were so that you could collaborate more closely, so that you could share information more quickly, get quicker information turns, and we've done that, so let me talk about that. In 2016, we expanded that lab in Fremont, California, where the headquarters of the company is. We call it the Rick Gottscho Lab, or RGL for short, so that was 2016. You've probably heard announcements we've made about the expansion of the lab footprint in Oregon. In 2023, we started talking about that.
In 2019, we began to invest in a lab in Korea, where a couple of our largest customers are. And more recently, you may have seen some headlines last week coming out of our intention to invest in a lab in India. Not only has this made us more efficient and accelerated our ability to deliver capabilities for our customers, it's given us access to great talent. Global talent availability, again, I think is a real differentiator for Lam. So that's a little bit about the lab footprint. Second, then manufacturing. So we've always had manufacturing in the United States, California, and Oregon, and we continue to have that, obviously. That's kind of where we've always been. In 2019, we added a factory in Springfield, Ohio, to support our silicon fabrication operations. So great spot in Ohio.
So the U.S. is critically important in terms of how we do what we do for manufacturing footprint. We've got a small manufacturing capability, and I should have said R&D capability in Austria. That's where our wet clean business is located. But importantly, and what has been enabling us to deliver some of that expansion and gross margin has been the expansion of our footprint in the Asia region. Like I said, in 2021, we announced that location in Malaysia. Biggest factory in our network has enabled a lot of that expansion and gross margin, but it's not just Malaysia. We've got factory presence in Korea as well as in Taiwan. So increasingly, as business is on this growth path that I described to you, the incremental volume is being manufactured in those Asia factories. Why is that important? Well, first, it's close to where the customers are.
On the left side of the slide, it shows you the geographic footprint of where our business resides. 80% of it is Asia-based. That's where the fabs in the world are located today. And yes, there's government money coming all around the world that's trying to expand the industry. That likely will continue to, but as things sit and for the foreseeable future, a good concentration of our revenue is in Asia. So what has that enabled us to do? Those two charts in the middle and on the right are important because this is what has enabled a lot of that expansion and gross margin. First, we've localized a lot of the supply chain. As we've built out this Asia footprint, we wanted the supply chain to be close to where we were, to be close to where our customers are.
This enables efficiency and effectiveness, shorter distances, a lower cost structure. So you can see the growth in the localization of the supply chain. It's showing you the percent of purchases that we make close to where the factories are. So that's important. The most important chart on the slide, though, is the one on the far right. Essentially, what this shows you is our operational cost as a percent of revenue. Those of you that listen to all of our earnings call will remember a couple of years ago we were talking about the Malaysia factory being a headwind, a little bit of a headwind, and you can see it on the chart. From 2020 through 2022, we built the factory, but we hadn't ramped it yet. Since then, we've been growing business, like I described to you, right?
Business peaked at the end of 2022, turned on for a couple of quarters, and has been growing ever since. The incremental volume, as that growth in the top line has shown up, has increasingly come from that Asia factory footprint. That trend will continue, so that's important. Again, I'm building a little bit of confidence that you're going to have when I show you that financial model. This is going to be a key aspect of how we are going to continue to deliver improved profitability, so this isn't the only investment at the company. In 2023, we also embarked on what I would describe as a transformation of the business processes at the company. We call it internally digital transformation, or DT for short. We haven't made investments in this scale and scope and magnitude in over two decades at our company.
We're still running on an ERP that was put in place more than 20 years ago. At that point in time, Lam was a single product company in a single factory location, and so we've managed to keep the business running in an environment that was somewhat constraining since then, so it's time to update things, and when business turned on, we took an opportunity to focus on this transformation, so this is important. This is a multi-year journey. We're in the investment phase today, so today, this is a negative in terms of the profitability of the company, and yet, we're still delivering pretty impressive profitability. I'll remind you, we just guided the March quarter to 32% operating margin with the investments we're making here, so important, you'll hear us talking more and more on earnings calls about what's going on here.
Like I said, this is a once-in-a-quarter-century investment to update business process, to provide AI hooks into the system so that we can take advantage of capabilities that are known today, but also emerging in the future. Business has gotten a lot more complex than it was when we were that single product, single factory company, and so these investments are important relative to scaling where we think this business is going to go. Why am I rambling on about this? Well, because we are in investment phase today, but this is going to transform systems, processes, how people do work at the company, so what this chart shows you is the investment phase we're in, right? It's right now a little bit of a drag on operating profit. It will eventually become accretive to operating profit.
So when I get to these financial models that I know you all came to hear us talk about, this is an important thing to understand. Today, again, we're north of 30% operating margin and at the same time making these significant investments. In a few years' time, this is going to begin to deliver positive operating margin profile. So anyway, more on this as we go forward in the next several earnings calls. We'll keep giving you visibility on what's going on here. So let me kind of wander around a couple of items relative to differentiation. To deliver the products that you just heard Sesha talk about requires investment. It requires investment in labs. It requires investment in that global footprint.
What this chart is showing you is R&D and capital by year that we have invested in this business: $2.5 billion last year, and it's going to grow this year, and it's going to grow into the future, and it's delivered some amazing new products, right? Just a couple: Altus, the Sensei platform that we first introduced to you five years ago, and have now talked about two new products, Vantex and Argos, that have come from that selective etch. You heard Sesha talk a lot about that. These are just examples of investments that we've made in the business to grow the top line, so this is important, but what I'm always just completely impressed with is our ability of this business to grow the investments that's needed to keep this going, and yet deliver absolutely world-class return on invested capital. This is a purposeful result.
I showed you kind of how we drive expansion and margin. This is also a pretty CapEx-like business: 4%-5% of revenue reinvested in CapEx. So the end result of driving and holding the business accountable for expanding gross margin and managing a growing investment, but also a fairly reasonable level of investment, enables us to deliver ROICs at five times the cost of capital. Finance 101, that's the definition of value creation. So super excited about this. But what I also love about our business is it is extraordinarily cash-generative. We invest first and foremost in what the business requires to grow it into the future. R&D, CapEx, critically important. But at the same time, when you look at the cash generation of the business, it's actually quite impressive. At least I think it's quite impressive.
And it's enabled us to return over $28 billion of cash to shareholders over this timeframe. Since 2013, we've returned over $6 billion in dividends. We've grown it pretty much every single year since we first introduced it in, I think it was 2014. And during that timeframe, we've also purchased nearly $22 billion in share buyback. That's enabled us, on a split-adjusted basis, to retire 840 million shares. If you look at the earnings that that has generated since 2013, it's $6.50 in cumulative earnings by share count reduction. In 2024 alone, it generated an incremental $1.32 in earnings per share. That was 40% of the income we generated per share in 2024. And it's 65% of our current shares outstanding. So if you look at what we've done, this has been 99% of free cash flow. This is a key aspect, obviously, of creating impressive total shareholder return.
So our plans going forward, we're going to continue to return a lot of cash to shareholders. Our previous statement was at least 75%. We're now telling you at least 85%, and pretty pleased with how we've been able to deliver very impressive total shareholder return over this timeframe, so what I really love, and this, frankly, five years ago, this was my favorite chart in my deck. This is also my favorite chart in the deck. What this slide is showing you is our quarterly revenue, as well as the dotted line showing you kind of average operating income levels over the timeframes you're looking at, and again, this goes back to early 2013, March of 2013. Business has evolved from kind of that mid-teens, mid-high teens operating income level up to the high 20s, up to the low 30s.
The lows of the business actually are higher than the previous highs used to be. Really proud. We are really proud of the fact that we've been able to achieve these results. And again, I hope I'm building a little bit of confidence in that this is a leadership team that knows how to deliver improving profitability, so let me pause just for one minute before I get into what I know everybody wants to see, which is that financial model, so how have we done this turning proven leadership into profitability? We live in a good neighborhood. Semis have been amazing. WFE has been even better. Etch and Deposition have been even better. Operational scaling and increasingly the digital transformation from the business, proactively managing the spending, being purposeful about returning cash to shareholders, and growing that installed base.
I kind of think of this as the virtuous cycle, the wheel, so to speak, of how we do what we do. So let's just kind of look at what this means for the future. Where are we today? What are we coming off of? Here's the 2024 financial results from the company: north of $16 billion in revenue, 16.2 to be very precise. The all-time record level of gross margin, 48.2% since we brought the two companies together in June of 2012, enabled us to deliver a little bit north of 30% operating margin and earnings of $3.36. So what does this then look like into the future? Here it is. So 2028, how do we think about what the business is going to look like? I think revenue is going to be between $25 and $27 billion.
This is because of the great product announcements you just heard from us. It's because the industry is going to grow. It's because the intensity of etch and deposition is going to go from the low 30s into the high 30s, and we're going to gain 50% or more share of that growth. The expansion of gross margin still has gas in the tank, guys and gals. This is the first time you've seen us put gross margin up with a five handle. So we think by 2028, we can approach 50% gross margin. And the enabler of that is that factory strategy that we talked about, the Asia footprint, being closer to customers, having a better cost structure because of the localization of the supply chain, lower freight and logistics spending because we're closer to the customers.
When you do the math on operating margin, 34%-35%, and we think we can approach earnings between $6 and $7 in EPS. So pretty impressive, I think. I thought, just for fun, maybe we'd show you what we think this looks like by the time the industry approaches that trillion-dollar level. Tim's already talked about, you know, we think we'll double the top line of the company. Let's call it $30 billion plus. Just plus. I'm not going to give you precise numbers quite yet. Then 50% plus on gross margin. We think as we look at where the business is going, we can continue to expand gross margin. Then call it mid-high 30s operating margin, above 35%.
We believe in this timeframe, assuming the industry does reach that trillion-dollar level, we can be at $8 in earnings, $8 plus, and 30% plus free cash flow. I always like to, like full disclosure, we don't always get this exactly right. There's a lot of assumptions built into how we've put these numbers together about WFE continuing to expand, Etch and Dep growing, so forth and so on. You can read what's on the slide. We're excited about the future. We're excited about all the new products that we shared with you today. Hopefully, you're excited about this financial model. I know I am. With that, we're almost to the end. For those of you that have been to our investor days before, we always like to have a close with just one more thing: a haiku. I always like doing the haiku.
One proven leader, two critical enablers, Lam outperformance. So with that, just summarizing the key messages, we're targeting high 30% SAM share at WFE, that incremental growth from the low 30s to the high 30s, an intention of gaining more than 50% share of that, and we're going to double the size of CSBG in that trillion-dollar industry. So with that, Ram, I am precisely on time according to the clock in front of me. With that, I'll invite Tim and Sesha and Ram up on stage, and we'll be happy to do a little bit of Q&A with you. I think the way we're going to do this, Ram, you'll kind of direct to where we're going.
Yeah, can I? So I'll direct the questions. I think there are people with mics around. So yeah, try to stick to one question and one follow-up. Also, if you're directing questions to Sesha, you have to say the full name, including his last name. If you get it right, you get an extra NDR or a one-on-one depending on whether you're an analyst or an investor, okay? Tim.
Thanks a lot. So if I strip off the service from 2028, so I get systems of like $16 billion roughly. So what WFE are you assuming out in 2028? I'm just trying to back into what your WFE share is. It seems like maybe you're assuming $130 billion.
Yeah, Tim, we got ranges around all this. I think as we look into the future, I think WFE grows into the mid-high single digits kind of on an annual basis. So you can get to where it is with that. Okay, thanks.
No follow-up, Tim? We'll give you a follow-up if you want it.
Take one. Yes, please. The question is on cost in NAND. A lot of what was presented was that there's a stacking with bunches of layers and even some logic. It seems like you're not going to get the leverage from the higher aspect ratios. Does that mean that costs in NAND begin to scale more with layers? And what does all this mean for cost per bit in NAND? Because the producers would argue that costs are continuing to come down. What you're presenting would argue that it's going to be more about scaling layers.
Yeah, I think there's some combination of both. In general, as you know, technology has been the answer for bit cost, whether it be in NAND or DRAM over many, many, many years. We still see that being the case. And in fact, some of the products that we talked about, Sesha talked about groundbreaking results in Cryo 3.0. It's also a productivity improver for our customers. And so look, when we deliver technology, it has to come with the right cost or it doesn't get adopted.
But I think that one thing that we're certain of is that technologies that were introduced back five, six, seven years ago won't be the low-cost option as we move forward toward 2028. So that's why our strongest conviction is that the two-thirds of the industry bit capacity that still exists below 200 layers has to be upgraded for performance, yes, to access the high-performance demands of AI and other applications, but also to get better bit density and lower cost.
Thank you. Srini Pajjuri from Raymond James. Thank you for the great presentations. I have two questions, both for Tim. Tim, you talked about $200 billion WFE from AI over the next five years, and we have tried to do the same analysis, and we got $50 billion in 2027 alone. So one struggle for us was how to figure out how much of that is incremental. I mean, I guess it's a big number, but just trying to figure out how you think about what is incremental, what is already invested. That's one question I keep getting from investors. Any thoughts on that?
Yeah, I mean, it's a tough question. I mean, obviously, what we're trying to give a sense of is that, and maybe the most important takeaway is not the number itself, but it's that the driver in that $200 billion is all leading edge. And for us, our story of SAM expansion and share gains with new products comes at the leading edge. Inflections, we use that term so many times.
I mean, inflections are a point of change. And in that case, we see customers being willing to change to new technologies that deliver the performance needed by AI. So $200 billion, I mean, it's our best estimate, but I think most importantly, $200 billion of leading edge WFE where LAM is an opportunity to introduce all of the new products, whether it's Argos or whether it's Halo, whether it's ALD, ALE. And so that's what's exciting about it for us.
Great. And then my next question is on the transitions that you talked about on the foundry and logic side. You kind of quantified the GAA opportunity and advanced packaging opportunity. If you can talk about how you see the backside power over the nex couple of years? And also, if you can talk about where we are in this transition in terms of GAA and also maybe touch upon how you kind of view CFET transition. Do you think the opportunity is somewhat similar to what we're seeing with GAA and backside power? Do you think that's going to be even better than the current transitions?
Thank you. Sure. I mean, first thing I would say is that likely every customer that you talk to has a slightly different insertion point for each of these technology inflections, whether it's three, two, one point four, one nanometer. So in some ways, we don't get hung up too much on the exact timing until Doug tries to do his financial forecast. Most important thing we want to make sure is we have the best product, the right technology that delivers for that need.
It will occur at different times at different customers. But we're seeing all of them moving forward. We work with customers, especially in foundry logic. You're working five, six, seven years with that customer ahead of the insertion point. So it's not as though we've just identified these inflections and are starting work today. So I feel very good about where we are in that process. Backside power, I made a comment in my talk that what's most exciting about the SAM expansion is some of these. You don't want to say anything's a layup because as Tim pointed out in his question, you have to deliver technology and cost. But the expansion is coming in things that we're really good at. Backside power uses a lot more copper plating.
I think that is just a case where if that's the type of work that's needed, Lam is already ready to deliver that with the products we have in place. And so I think where there are some new technologies being introduced, we still need to see those reach full production maturity. I mean, we said in CFET, the aspect ratios get even taller. Gate all around, there's things like the lateral etch that are pretty challenging technically. But we're working through those with our customers. And I would say in general, both sides end up being highly motivated to make them work in production. And so we feel really good about the transitions.
Thanks for the question.
Harlan here.
Yeah, good morning. Thank you for hosting this presentation. Harlan Sur with J.P. Morgan. Great to see all the new products and the innovation. When we talk to the production engineers and the fab managers at your customers who are driving the high-volume manufacturing at the mega fabs, they have a totally different set of metrics that they're focused on, right? Uptime, throughput, uniformity, defectivity, smallest footprint, consumables usage, etc., and they tell us that's the reason why they choose LAM, right, so as you develop these new systems, how does the team integrate the manufacturability requirements of your customers as a part of the development process? How does the systems development team also work with the CSBG team?
Well, Harlan, couldn't have been a better question for us. I mean, that's a fantastic place to build on. I mean, if you think about it, if you have a strong installed base, you have the support of the production customers, they see your tools and your solutions as reliable and dependable. I mean, that's a great place to build on the new technology insertion. So we build, as we're developing new tools, we're really thinking about that high-volume application. Because ultimately, if we get that wrong, I mean, the customer suffers, we suffer, and ultimately, you don't get the next chance. And so we spend a lot of time on that.
The Sensei platform that we first introduced for the Vantex tool and now is also the basis for the Argos system, we called that our first truly enabled equipment intelligent platform. Tremendous number of new sensors all feeding into those Equipment Intelligence solutions that I talked about enabling faster troubleshooting, more predictive analysis of parts lifetime and maintenance schedules.
And so at the point you get to introduce a new platform is really where you get to change the paradigm on how service is done, what the reliability of the tools are going to look like. And so I would just say we're engaged very closely. Our customers also have started to move a lot of that production engagement up into their R&D because they recognize, I talked about, if you're spending tens of billions of dollars on these fabs, they know they need to ramp them fast. And so they can't have a clumsy handover from R&D into production and a re-engineering. So I think we're all of the same mind. Do R&D that is already ready for production. And that's what we think of. And I think that's well aligned with our customers' objectives.
I appreciate the response. At the same time, this is a question for Doug. So at the same time that you've been growing revenues and scaling more of those revenues into Malaysia, we've seen the positive gross margin impact. You've also consolidated your legacy high-cost manufacturing base. Can you just give us a sense on the revenue capacity of Malaysia factory as you drive towards a strong growth profile over the next few years? I've heard Malaysia currently has about $6-$8 billion in annual revenue capability. Do you have the real estate and footprint to continue to expand Malaysia as you continue to grow the business?
Yeah, Harlan, no, that's a great question. Listen, we've got plenty of capacity right now to support the near-term needs of the business. And frankly, if we really are going to double the business, we'll need more. And we're always thinking about that and planning for the lead time at which we might need to make those decisions. So no problem. Thanks, Harlan.
Oh, yeah. Thank you. Maybe Mehdi Hosseini, Susquehanna, two follow-ups. Going back to 28 target, what is the underlying assumption for China and to what extent that includes the downside risk from some of the Chinese e-commerce vendors scaling their operation? And then as a follow-up, how should we think about the mix of your top customers? Would these targets, especially with the SAM expansion, have a dramatic impact on the mix of the 10% plus customers that you have highlighted in your recent 10Ks?
You know, Mehdi, I think about over time and the things we just talked about, the expansion of SAM and whatnot, it seems to me that China becomes a smaller percentage as we go forward. W e've already talked about that in the current year. Likely China ticks down a little bit. And I think that you'll continue to see some of that. It's not going away, though, by any means. China will be an important geography. But over time, certainly as we evolve to this trillion-dollar industry, China will just be another region like every place else that's got fabs. Sorry, in the second part of your question?
The second question has to do with, as I think about these SAM expansions for different applications, NAND, DRAM, and foundry logic, how does your SAM expansion and new products, new material would change the mix of the 10% plus customers that you have been highlighting in the past few years, 10Ks?
You know, Mehdi, this always, I don't know, in recent history, has been a pretty consolidated set of customers. I think given dynamics you see in the industry, that's going to continue. It's a consolidated set of suppliers. It's a consolidated set of customers. I don't think that's really going to change all that much as we go forward.
Anything you want to add, Tim or Sesha?
No, I think just on the China question, I mean, I think to just reiterate what Doug said, everything we showed you today about SAM expansion technology moving forward, those are things that demonstrate that Lam's growth path really is at the leading edge. Mature node is not going away. It's an important part of our business. But really, we're investing in new products to capture the growing etch and deposition intensity that exists in the future as higher performance is needed. And you've got to serve the needs of the AI applications. And so there'll be some element there, but we anticipate the faster growth coming at leading edge sort of starting to minimize the impact of some of these other businesses, including probably the China mature node region.
Thanks, Mehdi. Vijay had a question here.
Hi, thanks. Vijay from Mizuho. Just a quick question on the high aspect ratio etch. Obviously, with CryoEtch, I guess you showed the number of chambers growing 3x. Do you see more upside to that with 3D going into DRAM and NAND as well? And then a follow-up.
I think CryoEtch technology adoption for NAND is clear. It's already happened years ago. In DRAM, we also use dielectric etch for capacitor. And the scaling of that is also challenging. I mentioned that in passing. We do see that the scalability of those can also be improved by adoption of CryoEtching technology for that. Whether there is upside on that 3x number, a lot of assumptions, which WFE, who's investing. But broadly, I'm confident that Cryo is going to get adopted for NAND. Already happened and then will do so for DRAM in the future as well.
And as you look at the memory side, NAND and DRAM, that mix has obviously come down quite a bit for you. Do you see that pendulum swinging back up? And how do you see that playing out? Or as you see high aspect ratio in NAND, DRAM, and other MALI, et cetera, coming in, how do you see that mix playing out, I guess, over the next couple of years?
Thanks. Yeah, maybe I'll take it. And Tim, you can add on. Listen, the last time we gave you a model, I think the memory mix was north of 40%. Today, it's a third. I don't know that it's going to change all that much from where it is. It's ticking up a little bit as NAND is coming back to kind of the upgrade cycle that we described. But it's a third-ish.
And I think what's most important, kind of the message we delivered today about what's coming in NAND as WFE starts to return is it will be very upgrade-driven. And that's where essentially customers, they won't spend, the headline WFE number won't be as high. But Lam's SAM capture of that SAM is extremely high. You said it's about two-thirds. And so therefore, I don't anticipate that WFE will outpace kind of the demand, but the benefit for Lam will grow at a faster rate.
Yeah, it'll be more than two-thirds. More than two.
Yeah. Yeah, thanks for taking the question. I appreciate all the information, guys. Craig Ellis, B. Riley Securities. The first question is more on one of the technical evolutions that you showed. As we go from more planar to vertical DRAM, to what extent are we going to pick up efficiencies in R&D and product development across the memory portfolio? Do products in DRAM wind up looking a lot like what we see now and in the future with NAND? And to what extent is that in the mid-30s% operating margin that you put up in the target model, Doug?
You want to take that?
Yeah, I think across Tim's presentation and I, as we mentioned, the challenge is when you go to 4F squared, the gate goes vertical or we're going to do CMOS bonding. So there is, in some cases, and some I get the percentage wrong, maybe a third of the cases, literally a product that carries over from segment to segment. And there are scaling advantages there. In other cases, the underpinning concept will carry over. So similarly, maybe a lesser degree, but there is some R&D efficiency. And in other cases, just because the dimensions are so different, it may look conceptually the same, but may require additional investment. So a blend of use cases. There is a factor. We have made some assumptions of that. That's baked into what Doug showed as his financial model.
Yeah, it's all self-consistent to the best of our ability to estimate it.
Great. And then the follow-up question, I thought the cobot capability was very impressive. The question is, to what extent is that being deployed either in R&D lines or in production? To what extent is that contributing to the growth rates that you presented, Doug? I won't knock the growth rates. I'll say they were pretty attractive. Thanks, guys.
Yeah, it's yours, Tim.
Yeah, sure. As I said, we launched it just a couple of months ago. Obviously, that launch came on the back of quite a bit of work with customers. It is in production. It is available today on one type of tool for Lam. Given the success that we're seeing in that early production adoption, we now are working to expand the number of use cases across other platforms. It is something that I think right now it's hard for us to tell just how much it will affect growth rate. Clearly, the initial use is being met with a lot of positive customer reception.
And as I sort of joke, reception from our engineers as well. They really like having a lot of that work. I mean, you can imagine the precision tightening of all of those bolts on the top plate isn't exactly what you come in in the morning hoping to. So being able to sort of let that be done by the cobot. And the cobot can collect the torque data and put that away. And basically, it goes into both our Equipment Intelligence database and our customers. I mean, that's just a real big step forward for how maintenance gets done in the fab.
All right, there's a question there. And then we'll come back here to the front.
Thanks, Charles Shi from Needham. I have a question about ALD moly. And maybe this is a question for Sesha. And thanks for the incremental information you provided. A little bit more details, quite interesting. A couple of things. I look at that this is an ALD process. It's a solid precursor. And there's a lot of gap to fill. It's a 3D NAND. I mean, the word line, there's a lot of gap to fill. Very good opportunity. But when I look at the chamber setup, single wafer versus four wafers in your previous Tungsten product, four chambers per system, it looks to me from a cost of ownership standpoint, probably is a big step up from Tungsten.
So my question is this. From a penetration standpoint, if you can actually penetrate all the NAND word line layers, this is probably a massive step up. But at the same time, probably a little bit of a cost of ownership challenge for customers. So questions this. First, how do you convince your customer this is the way to go given the cost challenges? Probably. But secondly, looks like Micron may be an early adopter of that technology. But do you expect more of a uniform adoption of moly across all your NAND customers at a certain node, like two access layers? I'm just throwing it out there. Or do you see more of a staggered approach? So this is a two-part question. But the gist of it is really about the cost of ownership. Thanks.
I think moly intrinsically is a more complicated technology. So it is challenging to address costs. But that's what we've been focused on for many years. As I pointed out, moly actually replaces two different tools. It replaces the TiN barrier tool. And it replaces the tungsten. So when you look at overall cost ownership, that's one thing to consider. Obviously, the other thing to consider is how much investment we do in making the technology productive, how much investment we do in managing gas consumption, and things like that.
Net of all of this, NAND is probably the most cost-sensitive device type there is. So if a customer is choosing to make the transition, obviously, that means we have come a long way in addressing the cost challenges with bringing this technology to market. As to do we expect a wide adoption of moly for NAND, I think over the course of three to five years, we expect the majority of the industry to make the transition. I think the specific node in which it gets adopted will vary by customers. There are many other factors in terms of how customers adopt and exploit technologies like this.
Did you have a follow-up, Charles?
Good. Next question. Here. We're good.
Hey, guys. This is Krish Sankar from TD Cowen. Thanks for hosting us. Question on 3D DRAM. If we look back like eight years ago, the expectation was 3D DRAM would be in 2022, 2023 time frame. That got pushed out to 2025. And now the industry seems to be thinking 2029 would be a good year for 3D DRAM. I wonder, based on conversations you're having with your customers, how do you think about the timeline? Do you think AI demand and HBM result in some pull-in for that roadmap? And second part of that, if we juxtapose what happened with 2D NAND to 3D NAND, your share went from mid-teens to mid-20s%. And the NAND WFE was up 3x. So I wonder if you expect something similar with 3D DRAM.
You want to take the technical part of it, Sesha?
Yeah, I can. Yeah. I think your question was, why was the 3D DRAM assumption delayed? There are a couple of aspects. When you make a device architecture change, the economics of that change are more complicated. For example, entire fab layouts have to look different. So there is like a step function hurdle in making the transition. On the other hand, the technical challenges with 3D DRAM are actually quite immense. DRAM is a very, very high-speed device.
So the access, read-write, the quality of bits that you have to produce in a 3D architecture required a lot more research than what the industry kind of anticipated. The third factor, I would say, is sort of necessity, mother of invention type argument. Meanwhile, 4F squared provided another way to take a baby step as opposed to gutting everything and switching over back to 3D DRAM. So a combination of factors. And we anticipate real volume for 3D DRAM to be most likely outside of this decade. But that means we are hard at work today. So we are engaged with all the major companies and investing in early learnings there.
And given all the complexity Sesha just described and the fact that we don't really know entirely what the process flow is going to look like, too soon for us to give you numbers around the SAM expansion, except to tell you it's going to be good.
Did you have a follow-up?
Maybe just on the target model for 2026. Doug, you mentioned it's 40% of mix would be memory. But I wonder for a third. The previous model was north of 40%, is what I said. If we look at NAND WFE today, it's running at one-third of the levels it did run at two years ago. So I wonder what's implied in terms of NAND WFE in that 2028 model. Thank you.
Yeah, like I said, I think we're evolving to this focus on conversions. I think that's going to continue. It likely doesn't get back to the previous peak WFE levels in 2028. But like I said, memory mix will be about a third of overall WFE. NAND will tick up a little bit. DRAM will continue to be pretty strong.
Within the mix, DRAM, it's more weighted to DRAM. It's a higher percentage to DRAM than NAND.
You got one right here?
Yep.
Oh.
Hi there. Thank you. Brian Chin from Stifel. Appreciate the presentation. I think earlier in the presentation, there was a $40 billion NAND spending figure provided. Can you frame that over a particular horizon, and also comment maybe on your expected share? I think history might suggest it could be 30%. Why could this be higher moving forward? I have a follow-up.
Yeah, so what we framed was that two-thirds of the industry's bits are still below 200 layers. From both a performance perspective and a cost perspective and general competitiveness perspective, we think customers will need to move that bit capacity above 200. That was the framing, and we said to do that, we estimate that they will have to spend more than $40 billion on upgrades, and we said over the next several years. Obviously, the pace at which that happens, I think, depends very much on end market dynamics, but regardless of what the end market dynamics are, customers will need to move those bits forward, both from a cost and performance perspective.
So next several years is about the best we can give you. But even in that case, because of our high capture rate, and we were just saying in a NAND upgrade scenario, we're capturing more than two-thirds of the available SAM. That's a very good outcome for Lam, even if WFE levels don't reach back to the types of highs that they saw through the last cycle.
Maybe for Doug, just the follow-up. Can you just remind us for CSBG what the gross and operating margin is relative to the corporate average and how that factors into the margin expansion targets through 2028?
Yeah, we don't provide specific numbers, except I will tell you CSBG's operating margin is pretty good above the corporate average. But I don't give you a specific number.
Vivek?
Great. Thank you. Vivek Arya from Bank of America Securities. Thanks for hosting the Analyst Day. I wanted to revisit this NAND upgrade question because I think that's a key part of what sets Lam apart. And so in the one-third that has been converted over to 200-plus layers, how much CapEx was spent on that? So just so we can frame it versus the other two-thirds that will be converted, just so we can kind of do an apples-to-apples comparison.
Yeah, I don't think we've actually spent that much.
No, I mean, it's a NAND WFE over the last few years. But it won't matter because it's a step up because it's a higher layer anyway. So it won't be relevant for you to compare. It has no meaning. It's done. But you have to really understand how much is the future going to hold in terms of step up in NAND. So while you can compare, it'll be a smaller number than what it is in the future because every layer brings more complexity. I don't know if you want to add to that.
And the thing that makes it a little bit complex too is not necessarily all of that was put in as upgrades. I mean, there are pilot lines that get built with greenfield tools. There's R&D capacity that's running initially. And so it's a little bit convoluted. But the number going forward, as you see the upgrades, I think that's the analysis that you should use.
Maybe Tim or Doug can add. In terms of peak NAND revenues in the past, that question comes up. The relevant to Lam is maybe along the lines of peak NAND revenues, if you want to add to that.
I guess what I'll say is we don't need to get back to peak NAND WFE to get back to peak NAND revenues because when these conversions happen, our share of the spend is higher. So I think that's obvious, but just to state the obvious.
Got it. And then maybe one on CSBG, which I know Doug is your favorite segment.
I love it.
Yes. So I think this year you suggested it was going to be flattish, right, because of the effect of.
Flattish. What's going to happen this year? Upgrades are going to be up. Reliant is likely down and likely down because of the geographic mix of things that we talked about in the last e arnings call.
And the growth beyond that, is that predicated on this geographic mix staying the same? Or what helps CSBG to start to regrow after this year?
I think the biggest thing to think about is chamber count grows every single year. CSBG grows nearly every single year. Last year in 2024 was a very strong Reliant year. That is going to tail off a little bit this year. That's a unique situation 2024 to 2025 that I think Reliant, as it normalizes, CSBG should grow nearly every single ye ar, Vivek, and finally,
I was just going to add that as I talked about upgrades playing an important role in CSBG, there's a lot of this concern about customers are very conscious about cost. We get that. That's why upgrades are so attractive. They are the most cost-efficient means for the customer to scale to that next technology node. And that's where we have the power of the incumbency with that installed base. I mean, that's a great way for us to add value to the customer, help reduce their WFE spend, and really secure those next-generation technology positions for Lam.
Got it. And then finally, I think you suggested a mid-single-digit WFE growth for the next several years. Is that what?
Mid-high. I said mid-high. Mid-high. And what is China as a proportion? If China is what, about 30-ish% this year?
Likely trending down a little bit, yeah, this year. Certainly from last year is going to trend down. So mid to high with China sort of staying at these? Or what is the assumption around China in that? The fact that all this leading-edge stuff is driving a lot of the growth, China likely ticks down a little bit over the next several years would be my best guess right now.
Okay. Thank you. 25 guidance is in there with 5% for WFE. What Doug and Tim and Sesha said was long-term is mid-high single digits. We're not being that aggressive for 25. We said maybe even the 5% range. China is a lower piece to that. Then you build up.
Maybe one last question, Ram. I think just looking at time.
Yes. One last question. Get the last question there.
Hi. This is KC from Lynx. I understand the need for higher density DRAM in the world of AI. There's a lot of focus here to go to high-density NAND. From a tech perspective, what are your customers saying at a system level? What is the need for higher density NAND in the world of AI? Thank you.
Sure. I think we've talked about it. Our customers have talked about it. Enterprise SSDs, high-density storage for data centers, is clearly one of the key drivers for high-density NAND.
Awesome.
Tim, you want to close any remarks?
Yeah. No, I think just hopefully you took away the key messages we're trying to deliver today. We think it's a fantastic future for the company relative to the way that technology trends and inflections are really moving towards Lam's strengths. We're highly confident in the SAM expansion opportunities that are ahead. We're seeing kind of almost once in a generation changes in materials and technology. And Lam has prepared the tools that are necessary to capture that and secure leadership for the next generation in those steps. And then as customers, again, as we said, are cost-conscious about how they manage their installed base, they're also sensitive to the ROI and the speed with which they can ramp new fabs.
We think our installed base business and CSBG offerings continue to deliver tremendous value. So kind of across all three general aspects of growth, we think the next number of years are going to be quite good for Lam relative to key market and technology trends. So if you took away all those comments, then I think we did a good job delivering them. And we look forward to following up with you throughout this year on our progress and answering any other questions you might have. So thank you very much for attending today.
Thanks for coming.
Thank you, guys. Thank you.