Rambus Inc. (RMBS) Earnings Call Transcript & Summary

Good morning, and good afternoon, everyone. I am Sidney Ho. I cover semiconductor, semi-cap equipment and IT hardware at Deutsche Bank. The next company we have is Rambus. So Rambus is a leading intellectual property licensing company in the semiconductor market. In recent years, they have also expanded into memory chipset product market through an acquisition, targeting service and data centers. Today, we're very excited to have Rambus newly appointed CFO, Desmond Lynch with us. Welcome, Des.

Thank you, Sidney.

Before we get into Q&A, Des, I know, you have a few slides you want to go through. Why don't I hand it over to you. Go ahead.

That's great. Good morning, everyone, and thank you, Sidney, for hosting us here today. It's a great pleasure for us to be here at the conference. Before we begin on the short presentation, I would like to remind everyone on the safe harbor forward-looking statements. And I would encourage everyone to read the documents on file with the SEC, these contain a lot greater information on the company than I can cover in the short presentation today. Okay. Starting off with Rambus at a glance. Rambus has been a pioneer within the semiconductor industry for the last 30-plus years. The company was founded based upon foundational IP associated with memory interfaces, which can be found in many of today's modern computer systems. The company is headquartered in San Jose, California, and we have over 750 employees worldwide. The bedrock of the company has been our patent licensing business. This has provided long-term stable and predictable cash generation for the company. Last year, our cash from operations was $209 million, and this strong cash generation has enabled us to invest both organically and inorganically in our product programs. Our product revenue from chip and silicon IP last year is mainly focused around the data center and edge, with over 75% of our revenue coming from these end markets. This next slide is a great representation of how we go to market with our solutions as a company. Internally, we refer to our business as having 3 pillars: our patent business, our silicon IP business and our chip business. On the patent side, this is a foundational IP associated with memory interfaces. There is no physical transfer of IP here as companies pay us to license our technology. In total, we view this as a $200 million to $220 million billings opportunity for the company. So using the midpoint of the range of $210 million, $150 million of this comes from the top 3 DRAM companies, and the remaining $60 million comes from a variety of memory, SoC and FPGA companies. These have been long-term stable, predictable arrangements for the company. Moving on to our silicon IP business. These solutions are mainly associated with high-speed memory interfaces, controllers and security IP. There is a physical transfer of IP on this business as companies take our IP and integrate it into a larger ASIC or SoC solution. During the last earnings call, we were delighted to announce that this is $120 million to $130 million run rate business, which is up 20% to 30% versus the prior year. In this business, we have a much more diverse range of customers, which range from large well-known semiconductor companies to small and medium start-up companies. On to our chip business. This is where we sell memory interface chips to the DRAM companies who integrate this into the DIMM modules. The growth in this business has been phenomenal. In 2018, we had revenue of $38 million, and last year, 2021, with revenue of $144 million. And the growth has continued into fiscal year '22 with our first half '22 revenue being up over 60% versus the first half of '21. With the DDR5 server platform expected to ramp later this year and continue into 2023, we remain excited about the growth opportunities of this business going forward. On to the next slide. This shows the new memory architectures, and really driven by the megatrends of high-performance compute AI and ML, the data center architecture continues to evolve, which offers an amplified opportunity for Rambus. The transition to DDR5 is happening just now is the desire for higher memory bandwidth capacity continues to grow, and we have a great opportunity to build on the success of DDR4. With the addition of the coherent serial links, architectures like CXL offers further growth opportunities for Rambus in the midterm. Last year, we were delighted to announce our CXL product initiative, which intends to have products available to market in late 2023, start of 2024 with the launch of CXL 2.0. In the longer term, rack-level disaggregation opportunities offers further growth and expansion for Rambus. With the new memory architectures, the performance and the speed of the interface will continue to play a critical role in all of these architectures, and Rambus is well positioned in this market. Let's look at our financial highlights. The growth in the product revenue has been great, as our chip and silicon IP revenue last year with $192 million, which was up $60 million versus 2019. In addition to the top line growth, we have done a great job in managing the middle part of the P&L. Through disciplined operational focus and execution, we have reduced our operating expenses while funding the high-growth opportunities ahead of us. With the combination of higher product revenue and more operating expenses, this has led to superior cash generation. In 2021, our cash from operations was $209 million, which yielded $1.66 in free cash flow per share. And you can see this being reflected in our ending cash balances. We've had a consistent approach to capital allocation across the years. And I think 2021 was a great example of our capital allocation strategy in full force. Organically, we launched our CXL initiative and we invested in product programs. Inorganically, we completed 2 strategic M&A acquisitions to support our silicon IP business. And we returned $100 million of free cash flow to our shareholders in the form of an ASR, which equated to 55% of our free cash flow. Let me end with Rambus investment summary. The backbone of the company has been a patent licensing business, and this has enabled us to make investments in both our chip and silicon IP businesses. We've really focused the company around the high-growth areas of the data center, and through operational discipline, execution and focus, we have delivered strong financial results, which has yielded strong cash generation. We have a consistent policy also of returning cash to our shareholders. I think this takes us through the formal part of the presentation, Sidney. I'll now hand it over to you.

Great. Thank you for the presentation, very informative. Maybe I'll just start kicking off with some of the near-term questions. Most people here do care about the demand side of things. I think we are kind of all aware that the data points around inventory corrections in the consumer electronics market, like PCs and smartphones, but your focus, to your point in the slide, is that you are most in the server and data center market. Based on your conversations with customers, how do you see the demand trends in the second half of the year and even do you have visibility on this in next year? And can you comment separately on cloud spending versus enterprise spending?

Good question, Sidney. And really, our demand for our memory interface chips remain strong. We delivered quarterly product revenue in the first half, which was record revenue for us by working closely with both our customers and/or suppliers. Our demand in the first half of the year was significantly above what we could deliver from a supply adjusted revenue from there. We continue to work with our customers, we continue to receive new orders and we've had no cancellations of orders from customers. But overall, our demand remains strong. I think the second part of your question relates to the visibility into the cloud versus enterprise. Our visibility in this area is limited. But in aggregate, I would say that our demand for the data center and memory interface chips remains strong.

There are some comments from earlier speakers that between -- even within cloud, there is a bifurcation of U.S. cloud versus the non-U.S. cloud [ meaning ] Chinese cloud. Are you seeing any certain -- any main difference between the 2 categories?

Not really. We continue to work with all the sort of cloud providers, and we continue to see strong demand, irrespective of where these cloud providers are based just there.

Okay. Got it. Now maybe switching to the product side of things. Your product business has been tracking pretty well. I think it was up 40% year-over-year, and that's despite some delays that we're seeing in new server platforms, while you also have going through the supply constraints as well. So can you help us decouple the various dynamics within this bucket of revenue, DDR4, DDR5? And how sustainable do you believe easing dynamics are given the delays in key -- like I said, in the service platforms, as well as the certain macro hubs into the first half 2023?

Great question, Sidney. We've been delighted with our product revenue growth. In 2021, it was $144 million. And I believe this year, you have this modeled around $209 million. So that's up 45% year-over-year in a supply-constrained environment. In DDR4, last year, our overall share was just over 20%, which was up from almost 0%. We started in 2018 with $38 million of revenue, which equated to 0. So with each generation of DDR4, we've been able to grow our qualification footprint, grow our revenue and increase market share by delivering quality reliable products to the market there. In DDR5, we have invested early into the technology here. And our qualification footprint in DDR5 is stronger than any generation in DDR4. The transition to DDR5 will happen. I think we've seen some announcements that will be later this year for production and sort of rollout with a ramp into 2023. From a Rambus perspective, we do expect a crossover, meaning DDR5 will be the predominant unit shipment in first half of 2024. So this is a nice inflection growth opportunity for us as a company. I think we've done a great job overall on the chip side by excellent execution and continuing to grow, and we remain excited about the growth opportunities ahead.

Got it. So just to follow up on that. You said DDR4, you're in the 20% of kind of market share. I know it's been growing for every generation even within DDR4. But where do you think your DDR5, based on design wins, what kind of market share are you getting to?

Our qualification footprint is strong. And if you went back to earlier generations such as DDR3, DDR4, there was a leader in the market who probably got 40%, 50% sort of market share. So that is the goal for the sort of team of making sure that we maintain our leadership position within DDR5 and get to that sort of market share. Our qualification footprint remains strong as well.

Got it. Which leads me to the next kind of the follow-up question on DDR5. You talked about the timing, the crossover in the first half of next year. In the short term, you have been clear about this lumpiness that's tied to the adoption of DDR5. Can you help us understand this lumpiness? At what point does this lumpiness go away?

Yes, it's a great question, Sidney. The transition to DDR5 will happen. It certainly has pushed a little bit to the right. And as I commented on the crossover point being sort of first half of '24, we've been in earnings call sort of talking to investors about the lumpiness of DDR5 demand given the sort of delay in the rollout, as well as the combination of the supply challenges that we've been facing. I think we've done a great job of managing that by producing quarterly record revenue. I think as we get into sort of next year, and we'll watch carefully the ramp of DDR5. But that's the sort of time frame, maybe this lumpiness of sort of demand once the server platform has went into full production ramp, that should go away.

So just to be clear, I might have misspoke, so the crossover is first half of '24 not '23?

'24, yes. So with the Intel platform releasing towards the end of this year, we expect it to ramp in '23, and we expect a unit crossover, meaning DDR5, is a predominant unit shipping in first half of '24.

Should we assume the ASP for DDR5 is going to be higher than the DDR4 sort of maybe revenue crossover will be earlier?

That's fair. With the new generation of DDR5, we do have higher ASPs, so the revenue crossover could happen in advance of the unit crossover from...

Okay. I do want to touch on the supply constraints you just mentioned there. So that reminds me, I should ask a question there. You continue to face pretty fast capacity. Can you give us an update on the supply chain situation? Is there a way to think about how much you are impacted by these constraints, whether that's from volume or from margins?

Like many others within the semiconductor industry, we have faced a supply challenged environment for the last 2-plus years. Our visibility is improving into supply, but I would say it's still limited at the 90-day level from there. I think as a company, we've done a fantastic job of managing the supply chain as I talked about quarterly record revenue from there. But we still remain in an environment where our demand is greater than our supply from there. So we continue to work with our customers and suppliers on a sort of go-forward basis. But even though with small incremental improvements in the visibility, we still remain in a tight supply environment.

Okay. Sorry for going back and forth. But I'm going back to DDR5. Because in addition to the chipset, the DDR5 chipset, you recently talked about the opportunity of companion chips, right? So there are -- I think they launched 2 products. But can you help us just size the incremental opportunity relative to these -- to your base?

Yes. The transition to DDR5 offered an adjacent opportunity for us on the chip side is the companion chip for on the memory module as opposed to the motherboard. From a size perspective, this offers maybe $400 million TAM expansion to us by 2025. So a significant opportunity for us. Earlier this quarter, we were delighted to announce the temperature sensor in SBD hub, which launched within qualifications with the customers just now, and we remain excited about the future opportunities that this offers to us.

So the -- just to be clear, the 2 products that you announced today is not covering all the $400 million. There's probably more chips coming in the future, is that what you're talking about?

From a companion chip perspective, there was really 3 chips. One was the temperature center in SBD hub, which we released. The last chip is the power management chip. We're in development with that just now, and we anticipate 2023 data will be from there.

Okay. Perfect. So what are the customer feedback so far on those companion chips? How should we think about the "attach rate" the chipset business? And what is the competitiveness for companion chips?

Great question. The feedback from customers has been positive thus far within the qualification phase just now from there. From an attach rate, these are independent chips. So it's not a 1:1 attach rate with our RCD solution. But if you look at the module, especially on the temperature sensor, there's 2 chips there in 1 FBD hub. These are independent chips. And ultimately, we would like to have a 1 sort of product sale for the customers from there. But this is something we'll continue updating and watching sort of going forward. We're just in the qualification phase, as I mentioned just now. And I think from a competitive landscape with the second part of your question, we'll see our traditional competitors such as Montage, Renaissance, and especially on the power management chip, you will see more competitors out there, [ PI ], PS and the variety of other customers.

Okay. Got it. Well, another area within products that seems to be quite exciting, and I think you mentioned that on last year with the CXL standard targeting data centers. Can you give us an overview of this CXL technology and maybe perhaps the timing of the ramp? And how -- what do you think the incremental revenue opportunity for you in that area could be?

Yes. We are very excited with the CXL opportunity, Sidney. This is the new memory interface technology, really to address the data-intensive workloads of high-performance compute, AI and ML. It's really the interface standards, which will allow connections between different ICs and solutions within the data center. Our primary play just now in the next 12 to 18 months will be selling IP into this market. And this is really our CXL controller and set the solution from there. Last year, we were delighted to announce our product play into this market, which is really targeting having solutions available in time for CXL 2.0, which will be late 2023, start of 2024. The market is developing with many different sort of solutions from expansion to pulling devices to retimers to switches, and we've seen very large estimates out there on market size and opportunity. And we're very excited about this. This is an ecosystem that we know well. we've invested early into this, and we are excited about the growth opportunities going forward.