Rambus Inc. (RMBS) Earnings Call Transcript & Summary

Good morning. Why don't we -- good afternoon, excuse me. Why don't we go ahead and get started? I'd like to welcome everyone to the afternoon session on day 3 of our 24th Annual Technology Conference. It's my distinct pleasure to introduce the management team of Rambus. Luc Seraphin, the President and Chief Executive Officer; and also Rahul Mathur, the Senior Vice President and Chief Financial Officer of the company. We've got about 30 minutes in this forum to go through a fireside chat. [Operator Instructions] But first and foremost, Luc, Rahul, I wanted to thank you for supporting the conference and welcome you today. Luc, I always find it helpful in this sort of setting to make the first question somewhat open-ended. So I was wondering if you could just spend a few minutes kind of positioning Rambus for the investment community, talking about your core IP, the markets you exploit with that IP, kind of the strategy, and especially given last year's Analyst Day where there was a real kind of refocus on the company to move from sort of licensing to products. Maybe you can get upon that strategic change.

Yes. Thank you, John, and good morning, good afternoon, everyone. So as you said, John, we started about 30 years ago by developing the fundamental interface technology that is now used in all DRAM interfaces on the memory side and on the SoC side. Our business model originally was to license these patents to the industry. It was a patent licensing model. The patent licensing remains an integral part of our business today. It's very predictable, it's profitable. It's a strong cash generator. But over the years, we have evolved our business model and diversified our offering. We've invested the cash generated by this licensing program in the development of interface IP and semiconductor products. We also recognize the growing importance of semiconductor embedded security. So we developed the technologies to address that trend as well. So today, Rambus delivers these critical technologies in 3 forms: patent licensing, which was the original business model; IP cores that we sell to semiconductor companies and they build those IP cores into their semiconductor products; and the third leg of that stool is semiconductor product themselves. More recently -- over the last 2 years, we have recentered the company around its core competencies as you indicated. We have refined our perimeter. The reason we did that is that memory subsystems have reemerged as being critical to the performance of all emerging applications that require fast access to large amounts of data with low latency. So data centers, networking, 5G, IoT, ADAS, all of these applications require fast access to large amounts of data. At the same time as we were doing this, embedded security has also become pervasive in these applications for semiconductors. So in order to capitalize on those trends over the last few years, we closed our lighting business. We sold our payment and ticketing business to Visa. Both businesses were not central to our strategy and core competencies. At the same time, we bought INSIDE Secure from Verimatrix and Northwest Logic, both reinforcing our position in embedded security and memory subsystems. Internally, we have redirected our R&D dollars to reinforcing our technology leadership in those deals. So as we go into that transformation, despite the contractual step-downs of some of our licensing agreements over the past years and despite the refocus of our perimeter, our top line grew very modestly, thanks to the exceptional growth of our product business. As we indicated in the last earning calls, our buffer chip business only grew from $36 million 2 years ago to $70 million last year to about $110 million this year. At the same time, we were very disciplined on operations, and that allowed us to continue to improve our operating margins and ability to generate cash from our operations. So now in 2020, this is a real pivotal year for Rambus because we have, I think, built a strong foundation for growth. Number one, 2020 for us is a crossover year where our patent licensing billing will be roughly equivalent to what we bill from silicon IP and products. Secondly, with the extension of our license agreement with Micron and the contractual step-downs behind us, we expect our licensing business to stabilize going forward. And thirdly, our product business will continue to grow at a fast rate, pulled by the memory demand from the new applications I talked about earlier and the need for security. So going forward, it would provide an absolute growth, both on the top and the bottom line. So really, we are in this pivotal year with great expectations for the years to come.

Now that's a great sort of setup. You touched upon in your prepared comments, Luc, some of the top-down kind of secular drivers influencing the business: data center, networking, AI, sort of ADAS, IoT, edge. I'm wondering if you can help kind of frame what do you think your total addressable market and served addressable market looks like versus kind of our calendar year '21 revenue estimate that we have out there of around $260 million.

Yes, and Rahul will comment on those numbers as well. On the licensing side, patent licensing side, we've literally licensed the whole industry because those technology are so fundamental to the interface that the whole industry is licensed. So we've kind of maxed out on covering the sand there. On the buffer chip, this is a market that is approximately $600 million. We said this year, we're going to be at $110 million dollars. So we have room to grow. But we are -- we have that momentum behind us. As I said, we grew from $36 million to $70 million to $110 million, and we may talk about this a bit later. I think given our design win footprint and the fact that we invested very early in DDR5, we expect to continue to grow our share in a market that is growing. On the IP core and security, these markets are more difficult to size. This is a market that grow about 8% per year, 8% to 10% per year. And we're going to grow at that rate or a little faster. So in total, we are looking at markets that are in excess of $1 billion in terms of SAM for the current portfolio that we have.

Now that's a great backdrop. You mentioned that the big success story this calendar year was the outsized growth that you saw in your memory buffer business, the product business. I think the overall market only grew about 5% this year. I think you guys are on track to grow nearly 50% this year despite some digestion that you're seeing in the December quarter, which I want to get into. But I'm wondering if you could talk about what's driving that outsized growth when you think about your market share and design funnel. What are you doing better than everyone else right now?

Yes. Thank you. Yes, we grew 50% on the 5% growth market. There are several factors for that. The first one is, as I explained earlier, we refocused the company. And we've been working on high-speed interface, low latency for the last 30 years. And as this market evolves, the speed required between the memory and the process keeps increasing. So these interfaces are harder and harder to make, and that plays right at the center of our core competence. That's one reason. The second reason is that over the last 2 years, we put a great emphasis on something that might sound boring, but quality and reliability. That's absolutely critical in that market. The end customers are the large cloud companies. They don't want to have any disruption in the supply chain. And we've had absolutely high-quality and reliable products over the last couple of years. And thirdly, as we refocused the company, we made sure that every time there was an opportunity for a design win, either when a processor company introduces a new processor in the market or when a memory company introduces a new memory, we have the opportunity to win more designs. So the combination of high-quality, reliability, good supply chain, design win and focus is what explain our growth in that market.

There's a couple of important sort of transitions going on at the market level, one to sort of DDR5, and the other will be the introduction of Ice Lake hopefully early next year with the ramp in second half of Q1, Q2 where the memory channels are going to go from 6 to 8. I'm wondering if you could talk a little bit about how those will drive your business. What's your relative share in DDR4 versus expected DDR5? And how do you think the Ice Lake product cycle might help you?

Right. Thank you. Yes, so Ice Lake is still going to be a processor in the DDR4 generation of memory, just higher speed. So as I explained earlier, because this is a new generation of processor at a higher speed, all of these systems will have to be validated in the ecosystem. So that gives us a -- that gave us an opportunity for a better design win footprint, and this is what happened. So that design win footprint improvement will translate into continued revenue growth when Ice Lake ramps in the market. The second thing as you said is Ice Lake is going to feature 8 memory channels potentially as opposed to 6 memory channels in the previous generation. So to the extent that our customers' customers want to improve their capacity in terms of memory, that gives the opportunity to have more buffers per processor, because each memory channel requires a buffer chip on each one of the modules. So potentially, this factor of 6 to 8 could translate into a factor of 6 to 8 if people decide to use the additional channels. When it comes to DDR5, this is going to be a great potential uptick for us for several reasons. One is we were the first one to sample the market about 2 years ago with a very, very first sample. And if history repeats itself, the people who have sampled the market first tend to come in the highest share. This was the case for Inphi in DDR3, IDT for DDR4 and Renesas. We were the first to sample the market in DDR5. We're actually today receiving orders for sample that we're shipping for DDR5 to all memory vendors. The second thing with -- so we expect to have a high share just because of that. The second thing is typically when there's new generation, there's an uptick in pricing that, over time, decays but there's going to be an uptick in pricing. The third thing is that JEDEC, who is the standardization body, has changed the definition of DDR5 module in the sense that there's a potential for additional companion chips that are going to be built on the module. And as part of our R&D strategy, we are investing in the development of those chips. So the content, the silicon content on each one of the modules on DDR5 is going to be higher than what it is on DDR4. So if you combine those 3, we believe we have an opportunity to command the first position in DDR5 when DDR5 comes to full production in the market.

A couple of questions around that. So where do you think your share can go from today through the DDR5 cycle? And when should we expect to see that ramp of DDR5?

Yes, a couple of questions. So today our share, we estimate the market to be around $600 million, $650 million. And as we said this year, we're going to be at about $110 million. Analysts see us at around $140 million next year, so that gives you an idea about our share today. We believe that the one who commands the highest share in the market in a given generation can go up to 50 million -- 50% of the share -- of market share. So we believe this is a target that we set for ourselves in the DDR5 generation of products. Now DDR5 is going to start to ramp at the end of next year, and we expect the crossover between DDR4 and DDR5 to happen sometimes in 2023.

When we step back and look at the longer-term growth prospects in this market, one of the things that we've written about relative to our coverage of Micron is just this idea that the world is just becoming more memory-intensive at the device level and at the application level. And I think one of the analyses that we did is if you look at the average consumer device and you look at the ratio of logic to memory, it's about 1:1. If you look at an enterprise server, it's about 1:6. If you look at a cloud server, it's like 1:12. And if you look at an AI server, it's like 1:35, to try to give sort of degrees of magnitude about how much more memory-intensive the world is becoming. I'm assuming you guys should be a direct beneficiary if that thesis is right. And if so, how do you think about the long-term growth of the TAM here vis-à-vis that $600 million you've talked about?

That's a great question. We see this actually. Our customers' customers, the cloud companies, the Facebook and Googles and Oracle of the world is they are looking at that and say, "Well, I need to use that amount of memory." And the ratio between memory and processor is not the right one today in the current architecture. So they are thinking themselves, all of these cloud companies, about new architectures that dissociate the number of processes from the size of memory they have to address. And for that, they will continue -- they will have to develop architectures where high-speed, low-latency interfaces will be required, both on the processor side and on the memory side. And that's a great opportunity for us. That's a great opportunity for us because that's going to be another vector of growth for our core technology. And we believe that down the road, I think it's going to take years for these chips to develop, but down the road, that could potentially double the size of the TAM. I think we're going to continue to see a standard architecture for applications that are not that data-hungry. And in parallel to that, we'll see a trend towards developing more complex chip that change the memory subsystem architecture. And that will create an additional TAM, which it potentially over time could double the size of the current market.

That's helpful. Even the best growth markets can have some speed bumps or go through some periods of digestion. And clearly when you guys announced your third quarter -- your September quarter earnings and gave guidance for December quarter, you talked a little bit about this business going through a little bit of digestion to the tune of about $9 million sequentially in the December quarter. Help us understand what's going on in the business in the near term. And you've said on the call, you're fairly confident that the brunt of any inventory digestion would be over by the end of December. What gives you that confidence as you look into the new year?

So you're correct. What happened this year is that the first half of the year was very, very good, very strong because people built some safety inventory out of fear of what the potential impact of COVID-19 on the supply chains could be. Those fears has faded away, so we are going through a digestion of that inventory in the channels. And what we said is that we expect this to recover in the early next year. So we still have that view. And the way we track that is we just track inventories in the channels that we have access to. We look at our backlogs. We look at our bookings and these kind of things. So we believe that early in 2021, we should be back to some normal behavior in the market. That's our view today.

And then, Rahul, I think the other complexity that's going on with the numbers in the December quarter is just the re-upping of the Micron agreement. I wonder if you could just spend a little bit of time as to how licensing is sort of tracking in December, what the expectation should be as we go into the March '21 quarter. and I guess importantly, now that Micron is behind you, when is the next big sort of customer licensing side re-up that needs to happen?

Sure, John, and thank you again for having us. Our licensing business is a fantastic asset for our company. And as Luc mentioned, what we've seen over the last couple of years is step-downs that were structurally designed in some of the agreements that we signed in 2013, '14, '15, '16, '17, so not surprises. And the phenomenal growth that we've seen on the product side has essentially offset those step-downs as well as the businesses that we divested or shut down. As you noted, what we announced last quarter is that we we're able to extend our license agreement with Micron for another 4 years. So on a billings basis, we typically now expect somewhere between $200 million and $220 million a year from our patent licensing partners, so call that about $210 million at the midpoint. Out of that $210 million, $150 million comes specifically from our DRAM partners. So on a typical year, we'll get $60 million from Samsung, $48 million from Hynix and that $40 million from Micron. The nature of the Micron agreement was that the last quarter of the previous agreement had a step-down from $10 million a quarter to $4.5 million in this Q4. And with the extension, that steps back up to $10 million a quarter starting in Q1 '21 all the way through Q4 '24. I think one nuance, John, is that as you mentioned, Crédit Suisse models us on the ASC 606 revenue standard. I think internally and most -- many of our other analysts and investors models as though we were still on 605. And then what we do is substitute licensing billings for what we report for royalty revenue. So I think one of the things that's exciting to me is that next year, you'll see absolute growth on the product side, right? Because as Luc mentioned, many analysts have us going from $110 million or so this year to $140 million, $145 million next year, so you see top line growth there. But also with the nature of the extension of Micron, we'll now be able to recognize that as a variable royalty agreement. And so even though the billing will be almost exactly the same other than the one-quarter step-down, we'll now recognize that as revenue. So you'll see a double stacking of rev rec and ASC 606, both for Micron as well as on the buffer chips side. The good news about the licensing program is that this is stable, predictable, hundreds of millions of dollars of cash flow. And it really supports our investments, both organically, inorganically and also our capital return to shareholders. So Micron, as I mentioned, now comes up from their next extension in Q4 '24. Hynix comes up for their extension in the middle of 2024. And then Samsung comes up for renewal in the middle of 2023. So we have long range of sight for these licensing agreements for hundreds of millions of dollars for multiple years.

Now that's a very helpful information. And then Luc, earlier, you talked about the IP business and you brought up the acquisitions of Verimatrix and Northwest Logic. I'm kind of curious, given the assets you have in that portfolio today, how should we think about growth? And more importantly, how should we think about M&A as augmenting the organic growth going forward?

We're always looking at M&A opportunities. If you look at the structure of our businesses, we said in the past couple of years, we had step-downs in our licensing, but our product growth was organically quite impressive. And I think it would be hard or I will always do that, it would be hard to ask my team to grow even faster than that. So we will complement that with inorganic growth. And the 2 acquisitions we made were great for us. They both meet their targets after a year with us. They strengthened our position in the market. So we will continue to look at these type of assets, but we'll also be looking at assets, not only on the IP side, but on the product side. Because the nice thing about the semiconductor product business is that it scales, and there's much more leverage there. So we are in a constant process of looking at M&A opportunities. We have the firepower to do that because of our strong cash flows and our strong balance sheet. But at the same time, we want to make sure that we pick the opportunities that we -- that are going to add value to the company operationally, financially and strategically. And that was the case with the 2 last acquisitions we've made. So we say no to a lot of potential acquisitions. But as soon as we see one, we have the capability of acting very, very fast.

Luc, I have a question from the audience, and I want to incorporate that here. I apologize for jumping back to the memory business. But the question is, "Can you please elaborate on the new market opportunities in the coming years from the custom buffer market for the cloud companies, the Facebooks, the Googles, the Oracles of the world? Does it relate to the CXL interconnect?" And as a follow-on, "Can you elaborate on the extra real estate on DDR5 DIMMs, i.e., PMICs and the possibility of more TAM?"

Correct, yes. So 2 very good questions. Yes, one of the architectural choice that cloud companies in general have in front of them is instead of using a direct parallel interface between the processor and the memory, they could use a serial interface that would go into a chip and fan out then accesses to banks of memory. And that would be based on potentially a CXL interface. Absolutely right. So your question is absolutely right on. And we believe there's going to be traction in the market for this CXL-based type of applications. And CXL by the way on the CXL side, it's based on a 32-gig SerDes. And if you remember in our IP business, this is one area of focus that we've had over the past few years to develop 32-gig SerDes. On the memory side, it's either DDR4 and DDR5, which is also areas that we focused on over the last 2 years. So it's a great potential for us. On the DDR5 question, yes on the module, there's going to be a 3-companion chip. One is a PMIC and what we've decided to do as part of our strategic plan on the R&D side is to start the development of those chips, so that we are ready not only with the standard buffers on the module, but also with these companion chips when those companion chips are going to be required. And as I said, that's going to add a couple of dollars, a few dollars of TAM on each one of the modules as it comes to fruition.

Now, that's a great answer. I think one of the big misperceptions that I hear from investors, when you hear of licensing, when you hear IP, especially against the backdrop of U.S.-China trade tensions, there's always a concern about what exposure you have to those tensions. Can you walk through a little bit how we should handicap regulatory concerns vis-à-vis your business?

Sure. So we're watching all of these regulatory requirements, of course. Our exposure is not very high in relative terms to China. A patent licensing business is a pure legal agreement. The right to use some technology. So there's no technology transfer. There is no support, there's no communication once the agreement is done. That's the case with CXMT. So for us, when the Chinese market grows in the DRAM industry, if and when, then we're going to see a revenue stream to us, but there is absolutely no interaction with them. Actually, it's a good thing. I think if people want to build those type of products, they should pay for the IP that they use. On the buffer chip, our buffer chip customers are not directly Chinese customers. Their customers may be in China. So the impact on us was limited. And on the IP side, technology IP side, China has a lot of start up companies, especially in the field of AI. So we've seen a lot of opportunities for IP in China, but have not really been impacted by the current trade tensions. So we watch this, but we've been quite immune to what has happened in China so far.

Rahul, you and I talked about this in the past. One of the way we try to square the circle between 606 and 605 is to focus more on free cash flow and the free cash flow potential of the model. I'm wondering if you could talk about and give us some of the targets of how we should think about free cash flow margins over time. And as you answer the question, this was an especially strong CapEx year for the company. How do we think about a more normalized CapEx run rate?

Sure, John. And just to answer that question first, this year what we talked about was CapEx of about $35 million. And our typical CapEx should be closer to about $10 million or $12 million. The reason this year was $35 million is simply because we moved our headquarters facility. So that's something that I don't anticipate doing, except for maybe once every decade or so. So starting next year, I think we'll come back to a $10 million to $12 million of CapEx. And predominantly, that's IT equipment or test equipment. We're fabless, and so we don't have a lot of the other machinery that some of our competitors may have to deal with. I think from a free cash flow perspective, you're exactly right. One thing I've been delighted with is the performance on cash from operations. One of the phrases I use is, "P&L is opinion, but cash is truth." And the truth is we have a very strong business. Through the first 3 quarters of the year, we've generated $143 million of cash from operations. And that's a fantastic performance given our top line, and I expect for us to continue to have leverage out in the future. The delta between that and what people model under 606 is again, the difference between what we bill our partners and licensing billings, versus what we record as royalty revenue. But I expect our free cash flow to continue to be very strong. And what we've done is been very straightforward in terms of our capital allocation, right? So first is to continue to invest in these organic programs that are growing very nicely. The second, as we talked about, was potential M&A. And I think the 2 deals that we did last year establishes us as a good home and a good buyer for portfolios going forward. And the third is return to shareholders. And a month ago when we did our earnings call for Q3, we announced a 20 million share repurchase authorization. And then a week or so after we filed our 10-Q announced a $50 million ASR. And that actually does a little bit better than just offsetting dilution. And what we've targeted is returning 40% to 50% of our free cash flow back to our shareholders in terms of shareholder return. I think we absolutely have the cash and cash flow to continue to do organic investment, inorganic investment in terms of M&A and also continue to return capital to our shareholders.

Rahul, pushing on the buyback, I mean the authorization was roughly 20% of the shares outstanding. And when you adjust for the lower CapEx run rate next year off of next year's street numbers, the cash -- the stock just looks particularly cheap relative to free cash flow. Why not be even more aggressive on the ASR front?

We certainly could be out into the future. I think part of it is being even-handed in terms of looking at opportunities inorganically as well as from a capital return perspective. When we looked at 20 million shares, I think this is something that could last us 3 to 4 years depending on how active we are from an M&A perspective. But it's certainly something we'll look at again into the next year.

Perfect. With that, we've ended our time in this session, but I want to thank both Luc and Rahul, especially the people on the line. And I want to express our wishes that both yourselves, your immediate families and the larger Rambus family stay safe and healthy in what's been a very trying and challenging 2020. Everyone, I think, is anticipating being able to turn the page forward to 2021 and get a vaccine widely distributed and get back to some sense of normality. I really appreciate you guys for participating in the virtual conference. Hope to see you in Scottsdale in person in '21.

Thank you, John. Thank you, everyone.

Thank you, John. Stay safe, and thank you again for having us.