Rambus Inc. (RMBS) Earnings Call Transcript & Summary

Welcome to day one of Citi Virtual Technology Conference. My name is Atif Malik. I cover U.S. semiconductor and equipment stocks here at Citi. It's my pleasure to welcome Luc Seraphin, CEO of Rambus; as well as Rahul Mathur, CFO. The format of our discussion is a fireside chat. Luc is going to walk us through a few background slides, and then I'll open it up with my questions. If you have any questions for the Rambus management team, please e-mail it to me, that's atif.malik@citi.com, atif.malik@citi.com, and I'll ask those questions at the end. Luc, over to you.

Good morning, Atif. Good morning, everyone. First of all, you can find a copy of our presentation on our website. I'll start with the safe harbor -- obligatory safe harbor, and I encourage everyone to read our SEC filing where you can find tons of very good information about the company. But let me give you a brief overview of Rambus. We started the company 30 years ago, this year is our 30-year anniversary, as a patent licensing company initially. We were developing the fundamental technology that is used in DRAM interface and processor interface to the DRAMs. The patent licensing company remains the backbone of our business. It is a good business, it's predictable, it's profitable and it's stable. And it's a very strong cash generator for our business. Several years ago, however, we evolved our business model. We invested the cash generated by our patent licensing programs into developing IP cores and products that are important and relevant to the road maps of our customers and partners in the industry. We also recognized the growing importance of semiconductor-embedded security. So we also invested in developing solutions to address this trend in the market. So if you look at Rambus today, we delivered these critical technologies of high-speed interface and embedded security in 3 forms. In addition to the architecture license or patent licensing programs, we have an IP core business and a semiconductor product business. For the silicon IP business or IP core business, we develop complex, high-speed interface and security cores that we typically sell to semiconductor companies. These semiconductor companies then build chips, integrate those cores into those chips and sell those chips to their own customers. For the product business, for us, it's really buffer chip business. We design and sell the semiconductor product that goes into DRAM DIMM modules. Last year, we further focused the companies by redefining the perimeter of Rambus. We sold the activities that were not central to what we do. We sold our payment and ticketing business to Visa. We also bought 2 companies that reinforce our play in semiconductors. We bought Northwest Logic, which is a memory controller company that complements our memory IP business. And we also bought INSIDE Secure from Verimatrix, which reinforces our play in embedded security. So with this redefined perimeter and focus, we continue to support our strategy, which is based on 3 pillars: One is to continue to focus and strengthen our position in our key interface and security technologies for key markets like data centers, telecommunications and AI; the second is to improve our operational efficiency by leveraging the synergies we have between these different businesses; and the third one is to continue to generate cash and invest this cash in M&A and programs to fuel our growth. I'll let Rahul talk a little bit about our financials. But what you will note with our business is that we have this ability to generate very, very strong cash from our operations quarter after quarter based on the strategy that we have put in place over the last 2 years with the company. So in a nutshell, this is what Rambus does. And maybe Rahul, you want to say a few words about the financials.

Yes. Absolutely. So Atif, thank you again for having us. Always enjoy being at your conference. So as Luc mentioned, because of the strategic choices we've made and our operational execution, we've done a fantastic job of generating cash flow for our company. Our last quarter, in Q2, we generated $62 million of cash from operations, which was our highest quarterly cash flow generation in over 10 years. That compares very favorably. If you look at the first half of the year, we've done almost $100 million of cash from operations, which compares very favorably to $128.5 million for all of 2019. So very strong cash generation or execution, both in terms of the product growth as well as discipline from an investment perspective. I think one of the things that's a little bit confusing about Rambus is the impact of ASC 606, the new revenue recognition standard that was implemented a couple of years ago. Under ASC 606, the -- many of the long-term predictable licensing arrangements that we have with our partners across the semiconductor industry can no longer be considered revenue. So under the previous accounting standard, ASC 605, the way we would recognize revenue for these licensing agreements is that if we had a contract with a customer that was, call it, a $400 million contract, so $100 million a year or $25 million a quarter. Under ASC 605, we would recognize $25 million a quarter of revenue as we bill and collect from our partners. ASC 606 says that if there's no performance obligation in the contract, then you take the entire value of the contract in the period in which you sign it. So we would have recognized $400 million of revenue in that first quarter and then put the remaining balance on our balance sheet as an unbilled contract asset. So if we have payment for the first quarter, we'd recognize $400 million of revenue and then put $375 million on our balance sheet as an unbilled contract asset. Then every quarter, as we bill and collect from our partners, we would simply collect the cash and reduce the unbilled contract asset. So what many of our analysts have done is not necessarily looked at our GAAP revenue, but looked at components and operational components to understand what's the real cash flow of the company. So instead of looking at royalty revenue, they look at licensing billings. And many of our analysts will add licensing billings to contract and other revenue and product revenue to get a number that's more akin to what we used to publish under ASC 605. The easy way to look at it is the product revenue that we reported is for the buffer chip program that Luc talked about, and the contract and other revenue is related to the silicon IP business we have. As I mentioned on our last earnings call, this year, we expect about $35 million of licensing billing that's really related to our silicon IP business. So that, in a nutshell, tells you a little bit of our financials, very strong balance sheet with limited debt. We have a convertible debt for $172.5 million that comes through in Q1 of 2023. That's the only debt that we have. So we absolutely have the ability to continue to invest in our strategic choices both organically as well as inorganically. With that, I'll hand it back to Atif to begin questions.

Thank you, guys. Luc, the first one for you. Over the last year, there've been a lot of changes at Rambus from the change in CEO to the sale of payments businesses. Rahul talked about conversion to ASC 606. I feel like the Rambus story continues to improve over the last 5, 6, 7 years from kind of the memory focus and the litigation, to there was a change in management team. And now under your leadership, there's a higher focus on the silicon core expertise. So as I call this, Rambus 3.0 to my investors, how should we think about this Rambus 3.0 strategically moving forward?

Thank you, Atif. You're right. We did a lot over the last few years in redefining Rambus. As I said earlier, patent licensing remains important to us as it's a strong cash generator, it's predictable, and it's long-term contracts. By the way, you probably noted that last week we signed an extension with Micron with -- for the patent licensing agreement for another 4 years. So it's a very predictable, strong cash generator. And what we did over the last few years is really refocus the companies onto our core strengths in memory interface and high-speed interface and embedded security. So we really separated from these activities that were not central to this and refocused on to those technologies. And the reason we refocused onto those technologies is that we believe there is high demand coming from the communication market and from the data center markets. It's been compounded by something we were not expecting with COVID-19, the fact that people have accelerated the move towards working remotely, using games remotely, learning remotely. So all of this has shown the importance of data centers and communication, and this is right at the center of what we've decided to do. So again, what we've done over the last 2 or 3 years is refocused the company on to those end markets and those technologies that we've been building for the last 30 years is to make peace with the industry on our licensing program and use the cash generated by our patent licensing programs to reinvest in those technologies that we believe are going to be critical in the years to come. So that's the direction we're taking. I'll say a few words about the operational efficiency of the company. That's another thing that Rahul and I and the rest of the team have worked on. As we were transitioning our business from a pure patent licensing company towards a patent licensing and IP core and product company, we've continued to improve the operational efficiency of Rambus and our ability to generate cash. And the last thing I would say is that the company has dealt really, really well with the sudden transition to working from home. And I'll just say a few words about how proud I am of the employee base as we did that over the last few months.

Great. Luc, the last time I hosted, you mentioned that your cloud exposed sales in your memory buffer market have something like doubled last year or so. Can you just talk about the dynamics in the memory buffer market? How accelerating cloud CapEx investments are impacting that business?

Yes, sure. The nice thing about this business from a demand standpoint is that there's a very strong ecosystem supporting that business. The buffer chip business -- for people who are not familiar with that, a buffer chip is a chip that sits on a memory module. And a memory module sits in a data center, basically in a server. So that ecosystem is supported by the processor companies that send processes to these servers, namely Intel and AMD -- mostly Intel and AMD are strong players in that ecosystem. The memory players in that ecosystem are also very strong companies, Samsung, SK Hynix and Micron. So what you have here is you have a set of very large companies that support that ecosystem, and we are right at the center of that ecosystem. And every time one of these companies on the processor side, whether it's Intel or AMD, or on the memory side, introduces a new memory or a new speed grade, then there's an opportunity for us to introduce or to be -- to have a new design win with our buffer chip. So that's what's creating that dynamic in that market. So what has happened to Rambus is that when Intel, in particular, introduced their latest generation of processor, our footprint for design wins was much stronger than what it was in the previous generation of processors, and that explains the growth we've had. What you see on that slide is, in 2019, our revenue for products was $73 million, and mostly, this was buffer chip. What you see is in the second quarter of 2020, for that same product line, we generated $31.7 million. And we also had more than $30 million in the first quarter. So if you take the run rate of our buffer chip business and you extend that run rate towards the second half of 2020, then you could say that this business could be $120 million when it was north of $70 million last year. So we've been on a very, very high-growth trajectory that is based on the strength of the ecosystem, the strength of the end market and also our ability to increase our design win footprint. Going forward, the processor companies are going to continue to introduce faster processors that will continue to give us the opportunity to have more design wins. So we expect to continue this business to grow and to be very strong in the long run. In the short run, I think Rahul said it earlier, this business may be flattish quarter-over-quarter. Because of COVID-19 in the first half of the year, the ecosystem built a little bit of inventory to protect themselves against potential disruption in the value chain, and they may be bleeding some of that in the second half of the year. But the fundamentals of this business are very, very strong, and our position in that business is very, very strong. The last thing I would say, Atif, from this question is there are only 3 suppliers to this market for buffer chip. It's us, it's Renesas, who used to be IDT, and it's Montage. And as you note, we are the only U.S. supplier left in an ecosystem, which in the current environment also gives us some strength.

I see. Luc, the question I have, it's more like a clarification. You are exposed to the big 3 memory makers, Samsung, Hynix and Micron. How closely do you work with the processor companies, whether it's AMD, Intel or NVIDIA from the other side? Or you're pretty much approaching it from the memory side of the interface?

A very good question, Atif. We actually approach this from both sides. The way it works is that the buffer chip has to be qualified by the memory vendors because it sits physically on their memory modules. So they need to make sure that their memory modules are reliable and of high quality. So we have a very strong partnership with the memory vendors. But because this memory module interfaces directly to the processors, we also have to work very closely with the processor vendors, Intel and AMD. So as much as we have a very strong and intimate qualification process with the memory vendors, we have the same with the processor vendors. We have to work in close partnership with them to make sure that their processor interfaces well to the memory modules.

Great. And then, Rahul, switching to you on the model. Can you talk about your long-term revenue growth, operating margin targets? And what type of leverage you have in OpEx to achieve these targets?

Sure. Absolutely, Atif. So at our Analyst Day last year, I had provided a little bit of guidance in terms of how we think the different pieces of our business will go. I think what I had said at our Analyst Day was that we would expect licensing billings of about $220 million to $240 million this year, and that included about $20 million from silicon IP. I think because of the timing of the acquisitions we did and the structure of the accounting and contracts and acquisition accounting, instead of about $20 million this year, it's going to be closer to about $35 million. But what it comes back to is a pet licensing billings of, call it, about $200 million, $220 million a year or about $210 million at the midpoint. And what we expect is we expect that to be roughly flat for the next several years. As Luc mentioned earlier, one thing that we announced last week was the extension of our Micron agreement for another 4 years. So I think Micron had signed their original agreement somewhere in 2013, and they've built in perpetual extensions into their contracts. I think one thing that was interesting to note for us is that they actually asked us to extend the contract for 4 years instead of 3, which was their option in the agreement. But what that means is that Micron's license then has extended to Q4 of 2024, which they will have another option to extend after that. Samsung technically comes up for renewal in the middle of 2023. And then Hynix also comes up for an extension in, I think, the middle of 2024. So for many years, we have clear line of sight in these 3 contracts. And out of the $210 million or so of licensing billings, these 3 contracts on the typical year make up about almost $150 million. So a lot of predictability in terms of what's there. I think out of the remaining, call it, $60 million or so, it's split between security companies and SoC companies. We've licensed virtually everyone in the industry. So we're perpetually in the state of renewals, and that's why you see a little bit of variability quarter-to-quarter. But that's why we have comfort that, that licensing billings number will be roughly flat for the next couple of years. Then what you see is growth from our product initiatives. And this is growth, both from our silicon IP initiatives, either in memory or security, or on the product side, so specifically in terms of buffer chip. In silicon IP, we've seen very nice growth as well. And I think that certainly has the opportunity to continue to be a low double-digit grower for us. And our chips business has also grown very nicely. As Luc mentioned earlier, I think 2 years ago, it was about $36 million for us. Last year, it was about $70 million. And then if you look at where we are just from a run rate perspective, could we get $120 million this year? Of course, as Luc mentioned, it's just going to be interesting to see what happens for the remainder of this quarter with all the headlines and what's happening with inventory and mostly concerns with supply chain in China. So we'll see what happens over the next several weeks. But nevertheless, a very nice growth on the chip side. Now from an overall market perspective, we think that market size on the buffer chip side is roughly about $650 million, and that's split up by about $450 million on the RCD and about $200 million on the data buffer. As Luc mentioned, as we continue to get design wins, then you see that growth from a product revenue perspective. Now -- so Atif, that was a little bit of commentary on the top line. I think from a spend perspective, we've done a very nice job. And this is something that Luc and I had partnered on even before he became CEO, of taking costs out of our company and taking costs out in terms of programs that were far afield from the core semiconductor base that we have and also taking cost out from an infrastructure perspective. If you look at kind of our pro forma SG&A spend, you've seen it's come down very nicely over the last several years. I expect our spend to stay roughly flat because we have the investment in the programs. Any given quarter, you'll see spend be a little higher or a little bit lower just on timing of the programs. And also, we've had a little bit of benefit of less travel. We're doing this virtually instead of being in the New York with you, Atif. I think in future years, you will see R&D spend increase as our programs continue to grow. I think from a use of cash perspective, organic growth is #1 for us to continue to invent -- invest in that patent portfolio and then these product programs as well. So under ASC 605, what we talked about from a profitability perspective is operating margins between 37% and 40%. And I think analysts like you who look at our numbers, substituting licensing billings for royalty revenue, see that we're coming through -- closer to the bottom end of that range. Now here's what I expect to happen is that with licensing billings roughly flat and contract and other growing modestly and nice growth in product, we should see leverage in our model. From a margin perspective, our licensing program is almost 100% gross margin. Our contract and other is, call it, 90% gross margin. And our product margins have been in the 60% range, a little bit higher. And I expect that they'll stay roughly in the 60% range. So what I expect is that as you see growth from the products perspective, it maybe a little dilutive from a gross margin perspective, but it should improve our operating margin because of the fall through. Predominantly, our investment in that business is there. Now as we grow, we'll continue to invest more. But that 60% gross margin should be much more than 40% incremental operating margin. And I think that's where you'll see the leverage in our company from a profitability perspective. And you're already seeing it coming -- shining through from a cash perspective. So those are a few of the moving pieces, right? I think licensing billings stays roughly flat. You see modest growth in the silicon IP contract and other revenue. And you continue to see nice growth in our product revenue. We'll reinvest some into R&D, but you should see fall through more than 40% as you see incremental growth, and that will help our overall operating margins. Atif, does that answer your question?

Yes. Rahul, just staying on the cash topic. You guys have done a nice job in cash generation last few quarters. Last couple of years, you've been fairly acquisitive, Northwest. Before that, Inphi. How important are acquisitions to drive growth? Or is it just part of the bigger semiconductor consolidation trend?

Atif, what I'll tell you is that from a use of capital perspective, as I mentioned, first is organic and then second is inorganic. I think the transactions that we did last year were very well received, both inside and outside the company because it really showed that focus back into our core semiconductor business. So selling the payments and ticketing business, software security business that was going further and further away from our semiconductor base. And then as you mentioned, acquiring a memory IP business and security IP business as well to support our existing programs were very well received. I think we continue to be very interested in growing inorganically. You should assume that we talk to everybody in terms of different opportunities. What you've mentioned, Atif, I think, is exactly right, is that I expect that there's continue to be consolidation in our industry. And Atif, what that means is that there's not just whole companies available, whether they're public or private, but there's also portfolios that become available. Often, what I've seen and experienced is that if you have 2 companies, each of which that has 5 programs, and when they combine, they may not want to keep all 10, right? So we look at not just whole company opportunities but portfolios as well. And I expect it will continue to be acquisitive going forward as a use of cash. The third thing that we've done from a use of cash perspective is shareholder return. So I think in the last several years, we've returned $100 million back to our shareholders through an accelerated share repurchase program. So that's something that we've also considered in the past. But I think if you look at -- in our history, we've returned roughly 40% of free cash flow back to our shareholders historically. But the priorities are organic investment, inorganic and then shareholder return.

Great. Luc, coming back to you. You guys have a very extensive deep portfolio of IP around memory and technologies. You are the de facto on DDR5 standard. Citi has written about DDR5 being an important inflection for the market. When do you see your DDR5-related sales to improve? And also, can you just talk about some of the other technologies that are coming beyond the standard DDR5, DRAM technologies, like resistive RAM and other areas? What excites you longer term?

Yes. Thank you, Atif. So we continue to invest in our R&D programs on all sorts of memory interface technologies. One of the interesting aspect of the patent licensing business, to start there, is that you find the value of your patents years down the road. And what we've done with Rahul and the rest of the team over the last 2 years is we really refocused our R&D efforts on to memory interfaces because whatever memory interface standard is going to emerge, we want to be a player there. So in the long run, this is a strategy that we have taken over the last 2 years. Now when it comes to products, the DDR5 buffer chip business is an opportunity to have a positive step function in terms of market share. Every time there's a new generation of DRAM in the market, there's an opportunity if you're first in the market to commend the #1 market share position. So for example, for the DDR3 generation, Inphi got the lion's share of the market. For DDR4, it was IDT. For DDR5, we expect we're going to have a very strong share because we were the first ones 2 years ago to introduce the first DDR5 centers to the ecosystem we talked about earlier in this call. So as the standard evolved, we continued to produce samples and provide those samples to the ecosystem. Last quarter, JEDEC froze the spec of DDR5, so this is a sign that this is going to come to market. We believe the first shipments of DDR5 buffer chips are going to happen late next year in modest volume, and that is going to start ramping. And the crossover between DDR4 and DDR5, because these technologies take time to phase in and phase out, is probably going to be in 2023. The other interesting aspect for us with DDR5 is not only were we the first of introducing samples to the market and therefore we expect to have a strong market share, but the DDR5 memory modules have been defined in a slightly different way than DDR4. So there's potential for more silicon content per module, which is really, really good for us because we are also investing R&D money into developing those companion chips that are going to be required on the DDR5 modules. So in summary, the first shipments are going to happen late next year. The crossover between DDR4 and DDR5 is going to be probably in 2023. But between now and then, the DDR5 share is going to continue to grow. And the silicon content on these DDR5 modules is going to increase as well. So it's all good for us. To your second question about other technologies, we believe the DDR types of technologies are going to continue to be prevalent in that market. But what's going to happen is that the architecture around the processor and the memory at the system level may change. So the vast majority of the market is going to be -- to continue to be on the current architecture, where you have a processor and a memory interface that interface directly to the memory. But some of the cloud providers who have a need for very high bandwidth and very high speed, especially when they use AI type of applications, these cloud companies are thinking about other architectures where the interface between the processor and the memory may change over time and may not be a parallel interface but a serial interface. So that will open up a whole range of new types of potential chip development that will make an extensive use of high-speed SerDes interface, which we have as part of our silicon IP offering, and memory interface on the other side. So again, the demand for high bandwidth and high speed in applications such as AI may actually create an additional vector of growth for DDR type of technologies, and we're quite excited about that.

Great. That was very helpful, Luc. A couple of things on that. Firstly, can you kind of level set what your share is in the memory buffer market right now? Is it 1/3, 1/3, 1/3 across the 3 competitors? And maybe just to kind of feel your aspirations, expectations when we move to DDR5. And then on your last point of architectures working more kind of closely with memory, does this mean that you're always trying to work with some of the cloud companies or that is yet to come?