Rambus Inc.
Q2 2021 Earnings Call Transcript

Published: 2 Aug 2021

[Operator:]

Welcome to the Rambus Second Quarter and Fiscal Year 2021 Earnings Conference Call. At this time, all participants are in a listen-only mode. At the conclusion of our prepared remarks, we will conduct question-and-answer session. As a reminder, this conference call is being recorded. I would now like to turn the conference over to Desmond Lynch, Vice President of Finance and Investor Relations. You may begin your conference.

[Desmond Lynch:]

Thank you, operator. And welcome to the Rambus second quarter 2021 results conference call. I am Desmond Lynch, VP of Finance and Investor Relations; and on the call with me today is Luc Seraphin, our CEO; Rahul Mathur, our CFO. The press release for the results that we will be discussing today have been file with the SEC on Form 8-K. A replay of this call will be available for the next week at 855-859-2056. You can hear the replay by dialing the toll free number and then entering ID number, 7478989 when you hear the prompt.

[Luc Seraphin:]

Thanks, Des, and good afternoon, everyone. Q2 was an exciting quarter for the company. We complimented our excellent financial performance with a number of strategic advancements to capture the next wave of semiconductor growth focused on next-generation data center architectures. This quarter, we delivered $84.9 million in quarterly revenue and $51.6 million in cash from operations, beating expectations for the top and the bottom line. In addition, the company launched our CXL Memory Interconnect Initiative announced the acquisitions of PLDA and AnalogX and initiated a $100 million accelerated share repurchase program. Each of these activities grows the size and scale of our chip and IP businesses and puts the company in a great position to capitalize on the next wave of semiconductor growth in data centers. As we have discussed before, Rambus see the tremendous opportunity to lead the market and accelerate the adoption of CXL Interconnect in next-generation data centers. CXL is going to be a critical enabler for memory expansion and pooling in disaggregated and composable architectures. Through our CXL Memory Interconnect Initiative, Rambus is leveraging our unique combination of technology experience and expertise to develop breakthrough solutions to these new high-performance use cases. The acquisitions of PLDA and AnalogX strengthen this initiative by providing critical building blocks for high bandwidth chip and subsystem interconnect solutions. The complimentary CXL and PAM4 based interface solutions augment our interface IP offerings. And when combined with our existing interface and security IP, we have all of the necessary ingredients to be an industry leader in CXL and PCIe IP as well as offer differentiated CXL based interconnect products. Turning back to this quarter’s performance, we continue to consistently meet or beat financial targets. Our balanced portfolio of chips and Silicon IP continues to scale and will continue to drive growth in the coming quarters. Product revenue from memory interface chip remained solid with another quarter of a $30 million in revenue. And we expect record product revenue in Q3. Silicon IP revenue was up 10% quarter-over-quarter bolstered by a number of design wins as the industry continues to build solutions for data center and AI. Demand for our memory interface chips continues to be robust. We have sustained momentum in DDR4 and the strong market position in DDR5. We have received orders and have started to deliver production DDR5 memory interface chips to the market.

[Rahul Mathur:]

Thanks, Luc. I’d like to begin with a summary of financial results for the second quarter on Slide 8. Once again, we delivered a solid quarter with financial results at the high end of expectations and generated $51.6 million in cash from operations. Now, let me talk you through some financial highlights on Slide 9. We’ve consistently realized profitable growth over the past many years. This has enabled us to invest in strategic initiatives, returned capital to investors and improve cash from operations and free cash flow. We’ve built a strong foundation for future growth. Let me walk you through our non-GAAP income statement on Slide 10. Revenue for the second quarter was $84.9 million above expectations. Royalty revenue was $41.9 million, while license and billings was $65.2 million. The difference between license and billings and royalty revenue primarily relates to timing as we don’t always recognize revenue in the same quarter as we bill our customers. Product revenue was $31.2 million consisting primarily of the buffer chip business. Buffer chip revenue was slightly below the midpoint of expectations and would have demonstrated growth if not a shipment delay at the very end of the quarter. Contracts and other revenue was $11.8 million consisting primarily of the silicon IP business. As Luc noted, we were delighted to report quarterly records for the digital controller business we acquired from Northwest Logic and security IP. Our execution bodes well for the integration of the two acquisitions we announced in June. Overall, the silicon IP business grew over 10% quarter-over-quarter. Total operating expenses, including COGS for the quarter came in at $56.1 million. Operating expenses of $43.7 million were lower than expectations due to the timing of certain discreet R&D expenses. We expect to grow investment in our product roadmap, including through acquisitions in the coming quarters to drive for long-term growth. We ended the quarter with head count at 592, roughly flat from the previous quarter. I expect head count in Q3 to increase as we integrate the strong engineering talent from AnalogX and PLDA. Under ASC 606, we recorded $2.4 million of interest income related to the financing component of fixed fee licensing arrangements for which we recognize revenue, but not yet received payment. We incurred $0.8 million of interest expense primarily associated with our convertible note. This was offset by incremental interest income related to the return on cash and investment portfolio. After adjusting for non-cash interest expense on the convertible note, this resulted in non-GAAP interest and other expense for the quarter of $1.6 million, excluding the financing interest income related to ASC 606 this would have been $0.8 million of interest and other expense. Assuming a flat rate of 24% for non-GAAP pre-tax income, non-GAAP net income for the quarter was $23.1 million. With continued focus on cost and disciplined execution, we delivered profit that was nicely above expectations.

[Operator:]

Your first question comes from Gary Mobley with Wells Fargo Securities. Your line is open.

[Gary Mobley:]

Good afternoon guys. Thanks for taking the question. I want to start out by asking about CXL in your approach there. First question is, when would you expect to productize CXL or how should we think about the different phases of the product innovation of CXL, whether it relates to IP, licensable IP and then eventually chip-related revenue? And then specific to two acquisitions, which help form a portion of the basis for the CXL initiative. How should we think about the revenue contribution from each of those? Maybe not so much in the near term, but when you don't have the purchase of accounting headwind and then as well, the related expenses to those two. Thank you.

[Luc Seraphin:]

Gary, this is Luc. Great questions. So for us, the CXL, with CXL, we kind of kill three birds with one stone. First, it improves our IP offering for memory expansion and desegregation for anyone who wants to build chips for that market. So that revenue from IP is going to start almost immediately post-acquisition. People will buy IP, integrate that IP into their products and address the CXL market with those products. So that's the first bird, if you wish. The second thing is, we are going to be able to use that IP in building our own semiconductor products. And that's critical because these CXL products are complex, especially in terms of latency. And having the IP in-house is critical for the successful development of those products, those products that would take a few years to develop. And we hit the market in about three years from now and run from there. So there's going to be a second wave of revenue coming from the CXL products themselves based on their IP. And the third aspect of which is not related to revenue growth is that in these times of war of talents, the acquisitions of these two company allows us to bring very talented engineering teams into Rambus. So it shows three objectives for us bringing talents on board, increasing our IP revenue in the short run and increasing our product revenue in the longer run. And we expect these two acquisitions combined to bring about $20 million of royalty revenue in the first year.

[Gary Mobley:]

Okay. I appreciate all the color, Luc. And it's my follow-up question, I'm interested to hear I'm sure others as well, what sort of alarm jam there was on the shifts that business, the bumper chip business late in the quarter and what steps you took to fix it, and to what degree do your Q3 buffer chip revenue, to what degree does that contemplate any additional supply chain constraints? Thank you.

[Luc Seraphin:]

So, first of all, I would say that we've been able to manage our supply chain quite well over the last few years, by building redundancy in the supply chain. We grew the revenue, as you know from about $36 million in 2018 to $114 million last year. So we seen the supply chain constraints later than some of our competitors. And now what we see, we see some pockets of supply constraints across different products, and we working very actively and we have great support from our suppliers to make sure that we address each one of these potential challenges immediately, and that we don't disrupt adequately serve the markets and the demand, which continues to be very high. But we do see across the supply chain people increasing their lead times or some pockets of shortages. And as I said, it affects different products in different ways, but at this point in time we are working very actively with our suppliers to make sure it does not disrupt the demand that we see ahead of us.

[Gary Mobley:]

Thank you.

[Operator:]

Your next question comes from John Pitzer with Credit Suisse. Your line is open.

[John Pitzer:]

Yes. We’ve just a follow-up on Gary's question on the supply side, I guess, how much better could your product revenues have been in September quarter? And I guess importantly, as you look into December, do you see some incremental supply coming on such that it's pretty clear that you'll show sequential growth on the product side in the December quarter as well?

[Rahul Mathur:]

I'll start, yes, great question. I think for the third quarter, we are able to manage the situation. This is short term and we have great visibility of what we can do. So we feel comfortable with a third quarter with respect to the fourth quarter. We just have a lack of visibility. I think we continue to improve the situation with our suppliers but between the pockets of challenges with suppliers and the transition between DDR4 and DDR5, it's difficult to predict what Q4 is going to be. But what I would say is, that our design win momentum continues. Our backlog for DDR5 continues to grow very fast, demand for the products continues to work in, and the support from our suppliers is there. It's just difficult to predict what Q4 is exactly going to be at this point in time, but we feel quite confident with our Q3 number.

[John Pitzer:]

And then, Luc, on the DDR5, I think Micron on their earnings call, talk about DDR5 perhaps ramping a little bit later than they thought next year. I’d be curious to kind of get your thoughts on the DDR5 ramp. And I guess, importantly, what kind of revenue add or should that be to you either on ASP and our market share. And do you have a sense of how much the market could shift the DDR5 as an exit trajectory at the end of calendar year 2022?

[Luc Seraphin:]

So yes, we heard about the potential delay of DDR5, but what’s happening in the ecosystem is that people are building systems ahead of the launch. So we do see very solid orders from all memory customers that products that they ship to their own customers, because the ecosystem despite the potential delay is building the systems to be able to run the products. So it has had very little impact on our order book on our backlog. And that’s a good thing. We expect to see a price increase when we move to the downside. We see these as part of our backlog. And as we said in earlier calls, our footprint on DDR5 is quite strong, because we started early. So when it ramps in the market, we should actually see our share continue to grow. So we’ve consistently grown our share in DDR4 over the past few years, but we’re going to see a step increase in our share when DDR5 come to – comes to production. So at this point in time, we have a strong backlog. We receive bills from the three of them. Their customers are building systems in anticipation of the ramp of DDR5. So from that standpoint, all looks good. We’ll see when the actual launches is going to happen.

[John Pitzer:]

Great. Thank you.

[Operator:]

Your next question comes from Sidney Ho with Deutsche Bank. Your line is open.

[Sidney Ho:]

Great. Thanks for taking my question. So a couple of questions on the memory buffer chip side. I know you don’t want to talk about Q4 guidance yet, but what kind of regular expectations do you expect from DDR5 chipsets. And second question, I know you talk a lot – talk about three waves of memory TAM expansion, potentially doubling the TAM by 2025. Can you maybe talk about what are the sizes and timing of these various waves that that’s going to happen over the next few years?

[Luc Seraphin:]

Sure. So as on the first question about the DDR4 hyper mix, the first question on DDR4 versus DDR5, it’s – as I said, we have little visibility over Q4, but what I can say is despite the talks about potentially delay of DDR5. We see our mix between DDR4 and DDR5 in a favorable way in favor of DDR5. So we’re watching the mix as well. But as it looks today, the mix looks favorable to DDR5, but I think it’s due to the fact that the ecosystem, despite the potential few months of delay is building products for that market. And as I said, for Q4, we had watched the mix and we’ll work with our suppliers to make sure that we don’t disrupt any shipments to our customers. With respect to the product growth, we see the time doubling by 2025. This is, if you add buffer ship with the other interface product based on CXL, so that’s the combination of these two. And as I said, the CXL ramp is going to start in 2023, 2024, type of timing. So that’s where the market today is around $600 million, $620 million in size, so that market is going to double from that number by 2223, 2024. And there’s a reason for that as we explained in earlier calls, this is the need from the cloud service providers to have access to more capacity in terms of memory with lower latency. So that creates the need for either memory expansion use cases, which is going to be the target of our first product in that product line or for memory disaggregation and memory cooling which is going to be the second wave in that. So we do have a roadmap of products. So these are CXL initiative we starting with the first product, which is going to be the CXL memory expander, which would address the need for more memory per processor.

[Sidney Ho:]

Okay, great. Maybe a follow-up question is on the licensing and billings. If I’m not mistaken, I think typically, your third quarter is a strongest, usually up about 5%. Curious, why licensing billings and your guidance will be lower this year, given the strong utilization in the industry. I understand you don’t recognize any revenue from AnalogX, because of purchase accounting and you probably don’t have anything in PLDA is in the guidance. But just curious, is it the typical for Q3 for you guys to have to see a decline in billings and licensing billings?

[Rahul Mathur:]

Hi, good question. What I’ll tell you is that any given quarter, we’re perpetually in the process of renewing partners. And so what ends up happening is that just based on how we structure those contracts, you can see licensing billings fluctuate from a quarter-to-quarter perspective. And that’s often done kind of at the end of the deal negotiation just in terms of as our partners are working with us in terms of how they want to structure the financials. The other piece of it is that from a licensing billings perspective, we’ve seen more of our Silicon IP revenue show up in licensing billings this year than we’d anticipated. I think coming into the year, I thought it’d be closer to $20 million. I think now it’s probably going to be closer to $40 million. And so as we print that number, you could see some fluctuation just depending on how our Silicon IP contracts are structured as well. The base of our business, I think is just as wrong as we always thought. We still have that patent licensing business that continues to be very strong, very predictable. We’ve had some great renewals in the first half of the year that continue to support our belief, that will stay somewhere between a $200 million to $220 million business for the next several years. Our Silicon IP business also is growing very nicely. And I think as we mentioned earlier, it grew 10% quarter-over-quarter, and that shows up in a combination of contract and other as well as sometimes in royalty revenue or licensing billings. So I think that’s the way I look at it, I wouldn’t read too much into the quarter-over-quarter change.

[Sidney Ho:]

Okay. Very helpful. Thanks.

[Operator:]

Your next question comes from Mehdi Hosseini with SIG. Your line is open.

[Mehdi Hosseini:]

Yes. So thanks for taking my question. Want to go back to the growth opportunities a bigger picture. It seems to me that you would need to have DDR5 implementation for other opportunities like CXL to materialize. But once there’s a change in architecture, then there will be several different growth driver. Am I thinking about destroyed or CXL could stand on its own. In other words, can – is there anything with Silicon IP and a growth driver that could be viewed as independent of DDR5?

[Luc Seraphin:]

Great question, Mehdi. At a high level, a CXL memory expander has a CXL interface in the processor side and a DDR interface on the memory side. So it will come as a compliment to buffet chip business. Our buffet chip business, we’ll sit on the standard memory interface on the processor and once this is fully populated and the CSPs need more memory, then you will attach more memory to the CXL bus. So the CXL bus would attach itself to the CXL product. And on the other side of that product, you’re going to have a standard DDR interface. And that’s why it was strategically important for us to bring those IP on board, because we do have a complete set of critical IP that allows us to both develop the next generations of standard buffet chips, but also to develop those memory expanders that will help processes address more memories and they can do on the standard bus.

[Mehdi Hosseini:]

I guess, what I’m trying to understand is, let’s say, if the CPU company is late with their CPU upgrade. Can other CPU – GPU based vendor or ARM-based CPU chips that are coming to the market enable adoption of a new memory architecture, like, either DDR5 or CXL or both? I think there is some confusion around Sapphire Rapid and I’m just trying to understand how your growth opportunities are tied to introduction of Sapphire Rapid.

[Luc Seraphin:]

Right. So I start to the end, with Sapphire Rapids. Sapphire Rapids is going to feature DDR5 standard memory interfaces? So that bodes well with our DDR5 buffer chip offering, and this is what driving the growth that we’re talking about for the quarters to come. It will also show some PCI Gen 5 and CXL 2.0 interfaces. So those interfaces are going to be available in ecosystem, if you wish both from Intel and AMD. So the main stream is processors are going to pool the markets to use potentially those interfaces. And then I’ll follow with the first part of your question with people developing GPU’s or specialized CPU’s based on their cores. And what they’re trying to do is, they’re trying to address a portion of the market that with very data demanding use cases. And what these people could do is they could align to the same PCI Gen 5 and CXL 2.0 types of interfaces, which would be good for them, because if they do that, they know they can interface to all the products developing ecosystem. So what they could also do is they could accelerate that roadmap and move faster to CXL 3.0 or PCI Gen 6. And this is another reason why we did these acquisitions of AnalogX and PLDA, because they are advanced in the development of this next-generation of CXL, CXL 3.0. And they are advanced in the development of next-generation 5 – like the PCI Gen 6, 5. So by acquiring these companies, we will be able to address the needs of people who develop specialized processors for data intensive use cases in addition to the mainstream business.

[Mehdi Hosseini:]

Right. And just a final comment, I guess, given what you just laid out perhaps a year from now, your revenue volume would be different than what we’re dealing today. Like a year from now, there will be a relatively more broader diversification because of what you just said. Would that be a fair statement?

[Rahul Mathur:]

Maybe that’s exactly how I look at it. And as I was talking about earlier, we have a very stable, predictable licensing business because of fantastic cash flow and as I mentioned earlier, allows us to invest organically, inorganically and continue to capital return. You see wonderful growth opportunities for our silicon IP businesses. So as we mentioned in Q2, we had records for the controller business, we bought from Northwest Logic, as well as for our security IP business. And then you also see that the growth in the product opportunity, not just in the buffer chip, but as Luc mentioned in the future, it’s going to be in CXL. So certainly in three years from now you’ll see more on the product sides. I think next year you’ll probably see also growth in IP in particular. So I think we’re set up pretty nicely to continue to grow in across our portfolio in the coming years.

[Mehdi Hosseini:]

Got it. Thank you.

[Operator:]

At this time, there are no further questions. This concludes the question-and-answer session. I would now like to turn the conference back over to Luc Seraphin.

[Luc Seraphin:]

Thank you, everyone who has joined us today for your continued interest and time. We hope each of you stay safe and healthy. And we look forward to speaking with you again soon. Have a great day. Thank you.

[Operator:]

Thank you. This now concludes today’s conference.