Good morning, and welcome everybody to the final day of Needham 's 26th Annual Growth Conference. My name is Quinn Bolton, and I am the semiconductor analyst for Needham & Company. It's my pleasure to host this fireside chat with Vicor Corporation. The company is a leading supplier of high-performance power modules. Vicor designs easy-to-deploy modular power systems for power delivery networks that provide the highest density and efficiency from source to point of load. As a leading supplier of 48 volt solutions, Vicor is well-positioned as data centers and automotive markets increasingly adopt 48 volt power delivery networks. Joining me from the company are Chairman, President, and CEO, Patrizio Vinciarelli; Phil Davies, Corporate Vice President, Global Sales and Marketing; and Jim Schmidt, Corporate Vice President, CFO, Treasurer, and Secretary.
Before we start the fireside chat, I'd like to remind participants listening that you can submit questions through the dialogue box at the bottom of your screen. I'm gonna hand it over to Jim for his disclosures on forward-looking statements, and then we'll get to the Q&A. Jim?
Thanks, Quinn. Our remark, or our remarks that follow, including answers to your questions, may contain statements that we believe to be forward-looking statements within the meaning of the Private Securities Litigation Reform Act. These forward-looking statements are subject to risks that could cause actual results to be materially different from those expressed or implied by such statements. These risks include, among others, matters that we have described in our most recent Form 10-K and Form 10-Q, and other filings we make with the SEC. In addition, Vicor has not yet disclosed our financial and operating results for our fourth quarter ended December 31, 2023. Therefore, we are unable to discuss our financial and operating results for periods after our third quarter ended September 30th, 2023. Thanks, Quinn.
Great, thank you. Guys, I wanted to start off with a couple of questions around just the sort of general business environment. Obviously, the semiconductor industry was affected by inventory digestion for a good part of 2023. Wanted to start out with, you know, just a question, how you're feeling about inventory levels in your various end markets and you know, to the extent there's still inventory, when do you think that inventory levels may clear?
So, Quinn, this is Phil. I think that we really haven't been affected by, as such by inventory levels at our customers or even at our distributors. As you know, we've had a very, very large backlog that we've been working down. We had a large overdue backlog that we've been working down, so that's been the focus. I think that if anything, in terms of any effect on our business in recent years, it's been, I would say, the China market. We'd built the China business up to be quite large, and due to the trade problems that we've had over the last number of years, the
I would say, the China business has really reduced quite a bit, and that's had a bigger impact on our business than anything to do with inventory or anything like that.
Phil, do you see that China business, you know, recovering, or do you think the political situation keeps that, that business depressed?
Yeah, I think the political situation has been tough and, you know, a lot of the markets down there are commoditized, and so we're picking our spots. When we talk maybe a little bit about automotive, we've got some, you know, nice collaborations going on there, but, that's future, future business rather than short-term stuff.
Got it. Okay. You had mentioned the backlog, obviously, through the COVID supply chains, backlog for a number of semiconductor companies got to pretty extended levels. You talked about working that backlog down. As backlog has kind of come back, I think, at this point for the company to more normalized levels, have you started to see the orders environment stabilize and start to pick up? And do you have any thoughts when you might get back to a positive book-to-bill in the business?
So, since we're in our quiet period, Quinn, I would prefer to keep that question for, I think it's February 23rd for our earnings call. We can go to that sort of part of the business then, if you don't mind. I apologize. We're. It's just that we're in our quiet period.
Understood. Understood. Let me ask one more question, just sort of lead times. Again, things got pretty extended over the last 12, 24 months. Do you feel lead times in the business have come back down to normal or sort of pre-COVID levels?
We've kept our lead times out at the extended periods that we pushed them to because we want more visibility. Also, for our supply chain, we buy a lot of components to build our power modules, so that visibility there is very important. With our new factory coming online, I think we can start to reduce our lead times for our advanced products as we move through 2024.
Perfect. I think that's a good segue into my next set of questions is about the Andover facility and the expansion that, that you've undertaken. Can you just give us an update with the electroplating line? Has that now been qualified? Are you starting to manufacture products for customers?
Yes, we are. So the, you're referring to it as the electroplating line, but it's really a lot more than that. There are many processes, unique equipment, components to our fab, our ChiP fab, our Converter housed in Package fab. And we're on. We have essentially curtailed tendencies from outside processes and, you know, former partners.
Patrizio, on the capacity that you've brought on, obviously, you know, I mentioned the electroplating stuff, but the idea is to be vertically integrated. Can you say, do you have the, you know, the amount of equipment you would need to now be effectively fully vertically integrated at your current, you know, approximately $100 million quarter run rate? Or is there additional equipment you need to take delivery of, to be able to be, you know, to fully integrate, vertically integrate your current, you know, revenue level?
So we have, through the course of 2023, received and installed, all of the equipment that had been planned for, that timeframe. That's not the end of the line, no pun intended, with respect to additional equipment. There will be some additional equipment coming in middle of this year. But the bulk of the equipment is in place, up and running. Generally speaking, in terms of capacity, with the equipment we have, subject to some relatively minor incremental additions, we will have the capacity for up to around $1 billion a year in revenues. That's based on current expectations with respect to not just the number of panels, per day or period that we can manufacture in a vertically integrated facility, but also average sell price per panel.
Got it. So it sounds like you've got the equipment to you know certainly support the current business run rate and significantly more than that at this point?
Yep. Yes.
Good. Next question on, on, the facility. You know, as, as you've been ramping the vertically- integrated process, how are you feeling about yields on, on the new line? I know you had outsourced, you know, some of these steps, to an outsourced partner historically. Do you feel like your yields on the internal process are, you know, kind of at the level of the outsourced partner or perhaps even better?
So the first measure of vertical integration, in terms of its impact on manufacturing efficiencies, is cycle time. We've already seen major reduction in cycle time. And for the ones among us with a key interest on manufacturability, manufacturing processes, the impact that cycle time has on, among other things, yields, we've seen already the positive impact of processes that take longer to be completed, that reduce the amount of width that is necessary in order to achieve a certain capacity level. So we're getting better yields and greater profitability with respect to vertically- integrated facility.
That's great, great to hear. Lastly, on, on Andover, how are your, your customer qualifications progressing? I know you had, on the third quarter call, talked about expectations for sort of increasing activity in the year-end and, here early in 2024. But is there any update you can, provide on, on the customer qualifications?
So just within the last few weeks, we've had one of the multi customer come through the facility. We had an aerospace customer come through the facility. And generally speaking, every customer quality group that's come for qualification, including to the multi standards in aerospace recently, has been impressed by the capabilities we've been able to accomplish, and the process, the equipment, the unique capabilities that we put in place. You know, comments to paraphrase one recent customer, "Never seen anything like this." Very impressive. So generally speaking, customers get a very good sense, both the ones that come for joint certification and the ones that send teams to audit, you know, the equipment, the processes as to our capabilities.
Great. Wanted to move next to your vertical power delivery and the fifth-generation architecture. The, you know, vertical power delivery, I think, is something that you invented and have heavily patented. But maybe first just start quick, you know, a quick overview of the benefits of vertical power delivery.
Yes, to your point, we, we pioneered vertical power delivery. I think, the best evidence of, our effort with respect to what we call first gen VPD, which is what we pioneered years ago, is, with wafer scale applications, where as far as I know, we have been the only supplier, and fundamentally enabled something that would otherwise not be feasible. Because as in every application where the current density and the ability to provide a power distribution network is constrained with respect to physical dimensions, VPD is an enabling technology without which you can't move ahead to the next generation of products. But, you know, to be clear, and we pioneered what we call first gen VPD. We, we also see its limitations.
We've been seeing its limitations now for years, and that's what catalyzed fundamental development and enablement of what we call a second generation of VPD, which we're beginning to engage on with customers in AI, in particular. And fundamentally, what second generation VPD brings about is a level of scalability, flexibility, and cost effectiveness, which, as a pioneering technology, first generation VPD not enable. Now, to your point, we have IP with respect to first gen VPD. As importantly, or perhaps more importantly, we have IP with respect to our 5G or fifth generation factorized power system technology, which is part of second generation VPD. And these advances are, I think, on the clear path of further advances with respect to AI data center high-end computing.
What portion of the AI accelerator market do you think over time will require Vertical Power Delivery? Is there sort of a threshold, either on a current or power consumption level, that you think Vertical Power Delivery becomes, you know, mandatory?
So just to give you a sense of the opportunity at the level of, you know, in round numbers, 1,000 amps, just transitioning from lateral to vertical without going to full vertical, which becomes a necessity above 1,000 amps. What we've seen and validated is much as a 150-watt reduction in total loss in the system. Now, some of that is in the PDN, meaning in the copper associated with the substrate, delivering current, high current at the level of a 1,000-amp in the aggregate laterally. About 80-90 watts of that is in the PDN. There is another 50-60 watts, which is in the silicon itself.
That's often not understood, but it's a derivative power loss stemming from the fact that in a lateral PDN, you not only dissipate power through so-called I-square-R losses in delivering the current, but by virtue of the voltage drop that takes place through that current delivery, you also create a playing field for the silicon, the large multiplicity of pins that interconnect the silicon GPU, whatever you upper bowl to the PDN. And that, in effect, playing field, that is not as flat as it would ideally be, but shaped to create a substantial voltage drop away from the current sources on the sides, in a lateral power system, creates a derivative loss in the silicon itself.
So, the combination of additional silicon losses, which are often not appreciated, and the PDN losses themselves, which everybody is focused on, can be for a 700-800 watt system, as much as 20% or 150 watts.
So it's certainly significant losses on the lateral solution. So it sounds like that 1,000-watt or, sorry, 1,000 amps or about 800, 700-800 watts is where you really think, you know, that transition point hits.
Well, to be sure, at least to a lateral vertical, but getting beyond the 1,000 amps, we're involved now... You know, my trip earlier this week had to do with the 2,000-amp application.
Yeah.
When you get to those levels, even lateral vertical will not work. You need the fully vertical system.
Okay.
Now, to be clear, as with everything, it's not that you can't make anything happen without the benefit of first-gen or second-gen VPD or lateral vertical or vertical systems, but the fact of the matter is that you're substantially handicapped with respect to the low performance that you can achieve, and you have to throttle back trans and compute capability, you know, without the benefit of more advanced power components and a more advanced power distribution network.
You'd mentioned, sort of you've developed the second generation of vertical power. You're also developing your fifth generation architecture, I think, to enable that technology. Can you just give us, you know, a brief update on, you know, the advantages in terms of current density, power density with your fifth generation technology? And, for the investors listening, are you still on track to deliver samples of the fifth generation products to customers this quarter?
Okay, let's start with the benefits. So I think as mentioned in the past, the primary benefit of our 5G or future generation factorized power system is that we're raising the bar in terms of current density from 1 amp per square millimeter up to 3 amps per square millimeter. And that threefold increase in current density capability is at the heart of enabling what we call second-gen VPD. Those first-gen VPD with reduced current density capability is fundamentally challenged by the need to create stacks that make for more complex, heavy, thermally handicapped assembly. So the much higher current density of our 5G chipset enables a much simpler form of vertical power delivery that doesn't require the stacking.
Regarding the second part to your question, so this quarter we're bringing up our 5G demo system for internal characterization and validation. We are scheduling the second quarter to deliver samples to two customers, and we are on track to accomplish that.
Okay. So volume production, I think you've previously stated probably 2025, then for the fifth generation architecture.
Certainly in terms of volume production, there may be some de minimis power production for one customer later this year. But in terms of moving the needle with respect to revenues, this is a 2025 event.
Okay. Last question on the fifth generation. Do you intend to offer solutions of this, you know, technology to some of the lower power markets, maybe even down to the standard CPU server? Are there benefits that that higher current density can bring even to the lower power segment of the market?
So we have developed this generation of technology, learning from past experiences, to expand the sweet spot of the technology. Needless to say, key to success is continuing to raise the bar on current density, and general performance capability, to enable leading-edge solutions. But part of the focus in our development was to achieve, generally speaking, greater scalability, ability to support loads down to a few hundred amps, very cost effectively and with the same general traits and benefits. So if you want to look at it in terms of the range of current capability, where 5G Factorized Power Systems are applicable, they range all the way up, let's say, in a Wafer Scale Engine, to 50,000 amperes, you know, for one wafer, and all the way down to, you know, a few hundred amps.
Obviously, these are dramatically different types of applications. But the inherent attributes of our new factorized power system is that it can address those extremes very cost effectively, simply, more scalably, with lots of benefits for customers.
Got it. Thank you. Wanted to move now to some key questions that we get from investors, and I realize that there may be some customer sensitivities, you know, that I'm asking about, but I'm hoping, you know, you can address as best as you can these questions without divulging customer-sensitive details. The first question we get is, you know, on your June quarter conference call, you had announced a design win on a new AI platform, and wondering if you can share any update, you know, on that program. Is that program still moving ahead?
So, to your point, there are sensitivities here, but, you know, we are, generally speaking, on track with respect to shipping backlog. Beyond that, I'm not going to, go into specifics. I think as a matter of general policy, we're going to refrain from making customer-specific or, application-specific, commentary, unless we have, an agreement, with the customer and, in effect, are going public with respect to, to, sensitive information is clear beforehand by the customer.
Okay, the next question, you know, some of this, you know, I think came out in the lawsuits that you had filed against Delta Electronics, that, you know, one of your big motherboard partners had, you know, started to convert from a factorized power single stage conversion to a two-stage power conversion with a 48-volt, 12-volt first stage, and then traditional 12-volt voltage regulators. Can you talk anything about the impact of that transition on Vicor? Do you think that customer will continue to use, at least for a portion of its business, factorized power, or do you see, you know, kind of a pretty broad scale shift to two-stage at that customer? Because I know it's been, you know, a large customer for you.
So again, in the spirit of answering your question to the extent I can, given the constraint that I just outlined, I'm not going to make any customer-specific or application-specific comment, but I will answer your question in broader terms. So to your point, with the advent of 48-volt, right, in the data center space, and the desire to have an ecosystem for 48-volt that would enable solutions from a multiplicity of suppliers. To your point, OEMs have migrated or still, to some degree, migrating to a form of intermediate bus architecture with non-isolated bus converters, which again, Vicor itself, you know, pioneered. And I'm referring specifically to the NBM. Now, this is a competitive alternative to our older factorized power system capability.
It enables almost the same level of current density, you know, up to about One amp per square millimeter, and is a viable approach for these things systems and systems of the recent past. So Vicor benefits from these developments because the NBM is proprietary, heavily patented, the subject of ongoing litigation with respect to enforcing that IP. But getting back to the broader question of where does this lead us, to be clear, I do not expect the two-stage NBM multi-phase, multi-source solution to remain viable before too long, certainly for high current, you know, high-end types of system. It will remain viable for applications, you know, below 1,000 amps or thereabouts. But it isn't able to support the three to four amp per square millimeter that Vicor's factorized power before too long will deliver.
So you can see, in effect, the adoption of a two-stage involving an NBM technology is, to some extent, a step in a direction that has got near-term value, but not a long-term valuable position.
Got it. So the current, power you think brings some of those solutions, and again, we're not trying to speak about any particular customer, but just current and power density probably brings some of those solutions back to Vicor.
Yes, we believe so. We, we think that, you know, every or the many companies that are vying for the AI opportunity, large as it is, are, and many of them have come to their realization already. There are some that they haven't quite come to that realization, but based on my visibility, I think it's not going to take long for the space at large to recognize that it's a dead end for, a two-stage architecture at the high end. What you need is a factorized power system solution.
Okay.
A second-gen form of vertical power delivery that actually works, is scalable, is cost-effective, not handicapped thermally, mechanically, availability-wise, as these hybrid applications have. I mean, what we're seeing with customers that have adopted a first-gen VPD from buying competitors and infringing competitors, at least to some degree, is failures, issues, that I think are going to cause agonizing reappraisals.
Okay. One other, you know, question we get a lot from, from investors, and again, I'll apologize that it is a little bit more customer-specific, but, folks have asked, you know, with NVIDIA likely going end of life with its A100 family of GPUs at some point in the future, as it ramps its new B100 family... you lose your sort of last major design where, where you had been, you know, at least initially sole sourced. Without talking about specific GPUs, can you just say whether you still see opportunities at this customer, or do you still feel like you've got good engagements, with NVIDIA?
So again, we're not going to make specific comments with respect to customers' applications, but I can generally tell you that, customers need the technology that Vicor has, particularly, you know, 5G, technology, second gen VPD, and we believe that, we have a major role to play with each and every customer in that space.
Got it. Okay. Thank you. I do, again, I realize that these are more sensitive. I wanted to step back, you know, there have been some concerns voiced by investors that, you know, your, your revenue or share positions at, at, you know, any of these particular customers may be due in part to either quality or reliability issues. And so I wanted to kind of let you address those concerns, you know, to the extent you think there have been any quality, reliability issues. Are there any programs that the company's put in place to try to avoid those issues in the future?
Well, so I think these issues are made up by companies that either infringe some of our IP or, you know, lack the capabilities that we have. Fact of the matter, take the NBM as an example. Our field reliability for NBM measured at the 10 PPM level is better than infringing competitors.
Yeah. Okay. Lastly, just on some of the product opportunities, it does look like some of the major hyperscalers are now moving more en masse to 48-volt adoption, which I think gives you opportunities for the NBM or NBM-like products. We're also seeing, you know, increasing adoption of the OAM socket, where I think, you know, historically, Vicor has been pretty well positioned. Can you talk about just general adoption of 48-volt at some of the leading hyperscalers? And, you know, are you seeing growing, you know, traction behind the OAM accelerator module socket?
Yeah. So I think that, to your point, Quinn, the hyperscalers are mostly there now with rack designs at native 48 volts. There's still a number of, well, a couple of them that have got older data centers that they're still retrofitting, you know, AI systems into 12-volt racks, and they're still doing that 12-48 volt, you know, with the power blade based on actually NBMs. So we still see some of that. But for the CPU-type servers, yeah, 48 and 48-12, as Patrizio talked about, is very common now. That's becoming a very big market.
Okay. And OAM success, do you feel you've got some good opportunities there?
Yeah, again, same, same type of thing. I mean, you know, anything above 700 watts is what OAM is specifying as a 48-volt, sort of, accelerator or a CPU-type system. So yeah, we, we'll have lots of opportunities on the OAM side as well with companies that participate in that, in that standard.
Perfect. I wanted to ask just quickly on the automotive side, can you give us any update on the design wins, the traction, engagements with OEMs or tier one suppliers in the auto segment?
Yeah. So I think that the customers that we've engaged with, you know, a couple of years ago, they're, they're moving with our PPAP, power module devices into production later this year, but mostly 2025 in terms of the, the revenues that come from them. The collaborations that we've talked about are all moving ahead, fairly successfully. And recently we've added additional collaborations, you know, in the, particularly the Asia Pac, region. We've got some really good collaborations going on down there with both OEMs and, tier ones. So, that continues in a positive momentum, and, I'm still excited about that.
Great. I've gotten a couple of questions from the audience. The first is, "Your NRE revenue grew substantially throughout 2023 with the third quarter NREs achieving a record. What kind of customers are driving that growth? Is it automotive? Is it exascale advanced accelerators? And specific to the NRE growth in the third quarter, was that due to a few larger milestone completions, or was the growth driven by a number of customers working with Vicor?
So, the NRE specifically is really coming from our automotive collaborations, but we do have also some NREs coming from high-performance compute area, too.
Great, okay. Another question from the web, from the audience: "Any thoughts on a stock buyback? Do you believe your stock is undervalued at current levels?"
Let's leave that for another day, right? So, I don't think we're prepared to comment on that today.
Okay. Yeah, I wanted to come back to, to one of my questions. Certainly, one of the questions I've gotten from, from investors is, you know, I think there's a lot of excitement that you, you've talked about the technical advantages of vertical power delivery and, and your fifth-generation architecture. But, but your fifth-generation architecture doesn't really ramp in volume until 2025, and so what, you know, without giving guidance, obviously, what, what bridges the gap between, you know, today here in early 2024 and, you know, the launch of that fifth generation technology?
All the opportunities we have. We have a lot of opportunities. So, I can't think of a time in our history where we had as many opportunities as we have today.
Okay. Thank you. Another question from me, just lawsuits. You had announced lawsuits, both in Texas and at the ITC against Delta Electronics. You know, are those cases still on track? Do you have a court date? I think the ITC case was supposed to be heard this summer with a ruling in October, but can you give us any updates on the lawsuits?
Yes, at the general level. So, the ITC case is progressing very well for Vicor. We are approaching completion of fact discovery in February. The case is going to go to trial in April, and we expect an exclusion order later this year.
Okay. You know, I know from, you know, the in that 40-volt power module market, you've got Delta that you have sued. There are other suppliers of that technology, Monolithic Power, Flextronics. Do you feel that, you know, perhaps without naming, do you feel that other suppliers of those power modules may also be infringing your patents?
Let's focus for the time being on Delta. They're the largest infringer of NBMs, and for that reason, the focus of our attention. The infringement by Delta is an issue that affects a lot of OEMs. You know, with some exception, where the OEM having taken a license from Vicor is, you know, protected in terms of the outcome of the ITC case. I expect that when we get an exclusion order, Delta, the car manufacturers that use Delta modules that are infringing, and most importantly, their OEM customers are going to go lie down. And this is not going, we're all familiar with the recent events involving the Apple Watch, right?
Right.
In that case, it was as simple as turning off the blood oxygen feature, and you could still be shipping the product. But it's not as simple in this case. It's the oxygen is going to be going up.
Yeah, kind of hard to turn off the power module and still have the
That's right
Rest of the system work.
That's right.
We are approaching the end of our session, but I wanted to give the last minute to you, Patrizio, Phil, Jim, just to see if there's anything else you'd like to share with the audience or concerns, questions you'd like to address before we conclude.
I think if you have another question, it's best perhaps for you to ask that question, and we'll try to answer it.
Okay. I guess the last question for me is, you know, you've seen a nice uptick in your royalty revenue. You know, and, and I think you've said in the past that that's covered your litigation expenses. You know, do you expect that that will continue to be the case going forward? Do you see that royalty income continuing to grow in future quarters?
Yeah. Royalties are increasing at a substantially faster rate than litigation expenses. So obviously, the litigation expenses have been increasing. I think our opponents have thought that somehow bringing suits against Vicor with frivolous patents would deter our resolve and determination. I think they're finding out that that's a misplaced belief and that there's no opportunity to, quote, unquote, "level the playing field," because it's not a level playing field. Vicor is the innovator, and these companies are copycat companies that infringe the IP of an American innovator, and that's what the ITC is intended to protect against.
Okay. With that, we're at the end of our session time. So Patrizio, Phil, Jim, thank you very much for joining us at the Needham Growth Conference. We really appreciate your participation and look forward to catching up after earnings.
Thank you.
Thank you, Quinn.
Thanks, everybody.
Thank you.