Good day, and welcome everyone to the Vicor Earnings Results for the Fourth Quarter and Year Ended December 31, 2019, hosted by Dr. Patrizio Vinciarelli, CEO of Vicor; and James Simms, CFO of Vicor.My name is Tommy, and I'm your event manager. [Operator Instructions] I would like to advise all parties this conference is being recorded for replay purposes.And now I'd like to hand over to James Simms. Please go ahead, sir.

Thank you, Tommy. Good afternoon, everyone, and welcome to Vicor Corporation's earnings call for the fourth quarter and the full year ended 12/31. I'm Jamie Simms, CFO; and with me here in Andover are Patrizio Vinciarelli, CEO; and Phil Davies, Worldwide Head of Global Sales and Marketing.After the markets closed today, we issued a press release summarizing our financial results for the 3 months and 12 months ended December 31. This press release has been posted on the Investor Relations page of our website, www.vicorpower.com. We also filed a Form 8-K today related to the issuance of this press release. I remind listeners this conference call is being recorded and is the copyrighted property of Vicor Corporation.I also remind you various remarks we make during this call may constitute forward-looking statements for purposes of the safe harbor provisions under the Private Securities Litigation Reform Act of 1995. Except for historical information contained in this call, the matters discussed on this call, including any statements regarding current and planned products, current and potential customers, potential market opportunities, expected events and announcements, planned capacity expansion as well as management's expectations for sales growth, spending and profitability are all forward-looking statements involving risks and uncertainties. In light of these risks and uncertainties, we can offer no assurance that any forward-looking statement will, in fact, prove to be correct. Actual results may differ materially from those explicitly set forth or -- in or implied by any of our remarks today. The risks and uncertainties we face are discussed in Item 1A of our 2018 Form 10-K, which we filed with the SEC on February 28, 2019. We expect to file our 2019 Form 10-K this week, ahead of the SEC's March 2 deadline, and a refresh discussion of these risks and uncertainties that we…

Well, thank you, Jamie, and good afternoon to everyone. As Jamie just discussed, 2019 was a challenging year for us particularly in our China market with the tariff situation and potentially important customers there who ended up on the commerce department's denied parties list. We also faced an over-inventory situation and pushouts with our data center customers plus reduced spending in the semiconductor test equipment market. We did however see growth in North America and Europe with both our Brick Products and Advanced Products, and our global distributors grew at a combined 22% year-on-year. So we remain enthusiastic about our 2020 prospects in AI acceleration, supercomputing and data center servers given design wins and expanding product offering, including Power-on-Package solutions and an ever-broadening customer list.Customer interest in factorized power solutions for the 48-volt variant of the Open Accelerator Module or as it's commonly called the OAM, a design put forth by the Open Compute standards group, is high. We are working with GPU and ASIC developers on OAM solutions. If listeners are attending the Open Compute Summit in California in early March, you'll see companies there demonstrating new 48-volt-based OAM accelerator mezzanine cards. There are 2 power levels for these cards, and the lower power versions use our 48-volt to 12-volt and 12- to 48-volt modules. The 12- to 48-volt modules enable these AI cards to be backwards-compatible with data center customers who are still using 12-volt power delivery networks. The higher-power cards use our LPD, or lateral power delivery, 48-volt to load factorized power solutions.Looking forward, demand for much higher AI processor performance requires much higher current, which is bringing new customers both very large and small, such as new start-ups to our VPD, or vertical power delivery, as the enabling solution to their current density needs. Some of…

I am limiting my remarks today given a bad cold and cough that may make it even harder to comprehend my voice. I want to repeat the statement from last quarter.Vicor is executing well in the face of challenging conditions and near-term uncertainties, and we're confident the company is well positioned for long-term growth in very promising markets. We're engaged with parties interested in partnering with us for either technology or market access with the goal of accelerating expansion of vertical markets with alternate sources. Phil has spoken to a substantial head start to enjoy in AI acceleration and automotive electronics, which are both expected to follow sustained secular growth trends. And Jamie mentioned our forthcoming capacity expansion, evidence of confidence in the future.As discussed, booking patterns continue to reflect U.S.-China trade and tariff dynamics, country-specific micro uncertainty, segment-specific demand visibility challenges and, more recently, the coronavirus uncertainty. Bookings growth in Q4 was expected to set the stage for an upward trend in Q1 with sequentially higher bookings and revenues. However, program uncertainties are causing Q1 revenues to be less than earlier forecasted. We have a sizable pipeline of customer programs in AI acceleration, supercomputing and data center servers. With the sudden supply chain uncertainty brought about by the coronavirus, some large, near-term Advanced Products orders may be delayed until a clear sense of the extent and length of the outbreak is developed.We'll now take your questions. Operator?

[Operator Instructions] The first question is coming from the line of Quinn Bolton.

Congratulations on a nice order in the fourth quarter. Obviously, some near-term uncertainty around a number of moving factors, but I was hoping you could just clarify your comments again about the orders that pushed out. Did you say that it was an HPC customer or a hyperscale server customer that pushed, I guess, from Q4 to the month of April?

We saw pushouts from a couple of our data center and HPC customers actually. It was a couple of very large companies that we've been dealing with for a couple of years now.

Got it. And then I guess the second question, Jamie, you sort of said in your script that you did not expect revenue in the first quarter to advance or some similar term. Wondering if that's effectively sort of you're looking for revenue to be roughly flat. Or do you think it actually takes a step down with coronavirus and continued weakness out of the China, Hong Kong part of the business?

I would expect it to be slightly up but not substantially up contrary to earlier expectations. And going back to the question that Phil answered regarding the scheduling. So we are on the verge of -- with large customers in the server space, it's a ramp starting in April. And at this point, that's locked in forever. We're also on the verge of what's expected to be a major ramp for a new duty or obligation. That's also due to start ramping in the next couple of months. So what they've seen -- I'm sorry.

No. No. Please continue, Patrizio. I'm sorry.

So while there's been some rescheduling with those customers at this point, we have line of sight to ramps that are imminent. Now we should all be cautious though with respect to the potential impact of coronavirus, right? Because it has a disruptive effect on supply chain. And so we need to sit tight over the next several weeks to see what actually ends up happening. It's good news over the weekend that the Chinese government instructed Chinese companies to get back into business. That will help relieve some of the supply chain constraints we think. But until the dust settles, there's still uncertainty in the near term.

Great. And just last question from me. It sounds like the Q2 ramp and beyond is driven from multiple customers, hyperscale servers, new GPU, the HPC business coming back. Do you now have firm orders on the books for delivery beginning Q2 for those programs? Or are you still waiting for some of those orders at this point?

No. We have orders on the books for a lot of the new programs for this coming year, and we're in good shape.

Yes. We did take more planning. So what's in the books covers the near term, including in particular the ramp in the server, the application starting in April.

The next question is coming from the line of Jon Tanwanteng.

This is [Brenden] on for Jon. I just wanted to ask real quick. I know a lot of customers have moved their supply lines out of China due to tariffs. That's caused you to have a little rain delay. And wondering if you've been able to pursue some more strategies in your production at all. And do you see more of that kind of impact looking ahead?

Yes. So we've been pursuing the job strategy of limiting our exposure to Chinese supply because of the ongoing tariffs and other considerations. Unfortunately, these initiatives take time. We're looking forward with some of the key suppliers having established an alternate source outside of China in the first half of this year. So the dependency has been lessened, but it's still significant.

The next question is coming from the line of Rich Shannon.

Well, let's see. Maybe a quick question on the brick business here. If I'm to read the tea leaves right here, is it fair to say that your brick business is getting reasonably good bookings outside of China, but the ones for China are the ones that are seeing some issues?

Yes, that's correct.

Okay. As we look at China going forward and if I caught the numbers right from Jamie, you had 20% of your sales for 2019. I didn't catch the number exiting the year but -- and kind of a worst-case scenario when the tariffs and the supply chain issues that you've talked about already continue to be enforced and you can't solve the supply chain problem. Do you see a big risk to a lot or most of that China-based business disappearing over time? Or do you feel like you can still sustain some of that?

So I think it's a complex question to answer. There's a lot of imponderables. But in general, bricks are proven to have a high degree of resiliency to a variety of factors, the passage of time and the recent set of issues. That said, we obviously took a hit with respect to brick business in China last year. Phil, what do you see happening going forward with the brick business in China?

So I think that from the tariff point of view, the business has sort of flattened out in terms of the -- actually, the overall brick market in China is actually growing as a total market opportunity. It's just that there are local supply now for Chinese-made brick products, but the overall market is growing and our position in that market is still very strong with a very high brand of high quality, high reliability, ruggedness. So I still see a good brick business for us going forward for a number of years in the Chinese market.

Yes. So to be clear, the copycat bricks from Chinese makers don't work all that well when going up against our 30 or 40 years old bricks. And we are cannibalizing our own bricks with advanced products that perform the brick function in a fraction of the space and with significant performance advantages. So particularly, when we take into consideration the long-term competitive advantage of the advanced product version of an old-fashioned brick, demand in that market is subject to avoidance of [indiscernible] interference should stay strong.

Okay. That's helpful. A couple more questions from me and I'll jump out of line. First of all, last quarter, you got a sense of increasing breadth of your orders particularly -- and I'm particularly interested in the Advanced Products. Maybe if you could provide some context to how that finished the fourth quarter. How are you seeing the first quarter? Maybe give us a sense of what you expect in the breadth of this order and sales book exiting this year.

So exiting the year, we had some very nice orders from some new applications actually. We mentioned the video, the very large video wall display that we're involved with, the 20-megawatt level. That was for a new front-end, 3-phase, AC-to-DC product to AC to 48-volt. And it's our first entry, if you like, into a new AC-to-DC market that is actually 4x bigger than the DC-to-DC converter market. And it's a market that we have big plans for as we introduce new products through this year. So that order came in, in the fourth quarter. We've been working on it for a while. And then we started to see preliminary orders coming in for the launch of these new data center, HPC and GPU customers, early ramps, if you like, in preproduction phases, getting systems in place for their early customers. So that came on in the fourth quarter as well.

Okay. Perfect. One last question from me. I think in your prepared remarks here, Phil, I think it was -- you were discussing the OAM modules. A, just want to confirm that that's really a market at least today that are really serving by your NBMs. And b, how many of the announced partners for OAMs both on the inference and training side are you working with?

Well, to answer the last question, we're working with a lot of them. When it comes to 48 volts, I mean the NBM is the densest, highest efficiency product on the market. It's very easy to use, drop-down solution, and pretty much all of the OAM guys going 48 to 12 are using it. And then for the 12-volt infrastructure market where they have to go 12 to 48, they're using NBMs there as well. So we've got really great penetration with that product. And also, there's a regulated version of that product that's getting great traction out there as well right now. So we have high hopes for the 48 to 12 market actually.

So the regulated version is called the DCM, and customers can choose between fixed ratio and regulated alternatives with somewhat different trade-offs. And in both, we have, by far, a superior solution.

The next question is coming from the line of [Jon Dylan].

I just wanted to say it was really a nice thing gross margins go up this quarter even though your revenue went down. So that was nice to see. But my questions are more on -- first of all, Patrizio, I just want to make sure I understood. The bookings for next quarter, you expect to be up sequentially. Is that what I heard?

No, I didn't say that. So we see this quarter bookings being above the revenue level with greater than 1 book-to-bill, but we don't expect it to be this quarter at the level of Q4.

Okay. Okay. So I misunderstood. All right. And let's talk about the backlog for a second. It seems like you have a great backlog, over $100 million in backlog. And I'm wondering is that all scheduled to ship within the year.

Yes.

And is there any orders in-house that are scheduled for more than a year out that are not included in the backlog numbers?

There are some, but they are the minimum. So they're not all that significant because generally speaking, customers don't book that far out.

Okay. I wasn't sure if some customers -- kind of the customers who keep coming back, you would put in some long, multiyear orders to get better pricing or something. I didn't know if that was possible or not.So for Phil, you talked about some of the ramps that you're seeing. We've seen one hyper data center, Google, who's been taking product from you for quite a while now. And I imagine your strategy is to use some of the AI chips to get your foot into the door of the other data centers, the big data centers. But I'm just wondering can you give us a little bit more color on that. And also, is there any visibility as to when another hyper scale data center will adopt 48 volts on a grand scale across the whole data center?

Okay. So a lot of questions. Okay. So...

So we'll leave the specific names of customers out of this, right?

Yes.

So talking generally.

So it's safe to say we're working with pretty much all of the data center companies here in the United States across a variety of different applications. So some have talked about -- which is the 48-volt to 12-volt and 12- to 48-volt sort of applications. So we're working with pretty much all of them on that. With regards to factorized power applications, we're working with again quite a few of them on their own internal ASIC development. And those are new engagements that we -- came on board in Q4 and further ones that we are in the initial stages on, which will come on in Q1, Q2 of this year.

So one of them, they expect over a year in terms of development.

Yes.

But there have been more recent additions.

That's right. Yes. Yes. So in terms of deployment of large-scale, 48-volt racks, that is starting. And the Open Compute forum, you'll see a lot more 48-volt at that show this year than you did last year, although last year was very encouraging. So you'll start to see, I think, 48 volts in the rack in terms of the power delivery networks at the big other hyperscalers like Facebook, Amazon, Microsoft and companies like that. I would think at the 2021, 2022 sort of area. Up until then, they have to use the 12 to 48 and then our 48 to load factorized power solutions for the AI accelerators that they're going to use. So we're sort of...

The AI accelerators have transitioned to 48-volt, the ones that really matter.

Yes. So we're getting into those data centers sort of like a Trojan horse strategy, if you like.

So you're kind of getting your foot in the door with the AI stuff, but then it sounds like the data center customers are starting to realize that 48 volts can go across the whole data center and that would be beneficial to them. Is that what I'm hearing? Is that kind of the lay of the land?

There's no turning back from 48-volt either in the data center space, in AI or, for that matter, in automotive. It's coming.

We see the same thing happening in China as well. So the China -- Alibaba, Baidu, Tencent, they're all actively developing 48-volt racks based on the Scorpio standard, which is their equivalent of the Open Compute.

The next question is coming from the line of Gus Richard.

Just quickly on the reschedules that you saw. Can you give a little bit more color as to what's driving that? I know one was a com satellite and the other was a hyperscaler. Is there an issue with their facilities or products? Any color there would be helpful.

No. It was the bring-up of new products, new platforms, and they were over-inventoried quite a bit on the -- some of the older programs and platforms. And so they build those out and then they start working on the next-generation microprocessors and GPUs, and that was the reason for the pushout. The programs are there. They're healthy. They'll come on in Q2.

Got it. And then on the supply bottlenecks, any additional color there? Is that issues getting magnetics or something else?

So thus far, we've not been impacted on the supply side from coronavirus. We may be impacted if shops in China stay closed. Again, as suggested earlier, it was encouraging to see over the weekend that the mandate from the Chinese leadership is to get back to work. So we'll have to see how it plays out. Obviously, if coronavirus turns out to be -- they cease and the recent reversal in China were to turn yet around, that could have serious effects with respect to the supply chain. But thus far, we've been able to get, for the most part, the components we need from China. Obviously, the chain -- supply chain was effectively closed over the Chinese New Year. And in this year, because of coronavirus, what's usually a 1-week shutdown turned into a 3-week shutdown. And things are still not really running smoothly. But in anticipation of the usual shutdown, we obviously had buffer materials. So we're covered through the next several weeks, and we're taking a wait-and-see attitude and obviously taking steps where possible to make sure that our needs are taken care of.

Got it. And then the last one from me. You're working with a number of ASIC vendors, GPU vendors, et cetera. And I'm just trying to understand is most of those customers -- is your products required when they move to 7-nanometer or 5-nanometer? Or is it independent of what process now they are currently running?

It's really more to do with the level of current that they need. Certainly, as you go to 7-nanometer, the performance is going up. Therefore, the current is going up, the power is going up, and the same at 5-nanometer. The challenge as they move down those process nodes is that the -- sometimes the operating voltage that they're working at is dropping. And our factorized power solution with current multipliers is a fantastic way of going from 48 volts down to -- we've got customers as low as 0.35 volt. And so you can still do that with a factorized power solution, much easier than you can an IBA sort of multiphase solution. So it's across the board really, but it really depends on the amount of current.

But there's a correlation between the voltage and the current because generally speaking, the technology trend of going down to 7-nanometer and 5-nanometer is a technology trend that leverages a reduction in voltage, an increase in current capability for -- comparable to a power dissipation to achieve greater processor performance. And to Phil's point, that's also strongly correlated to our solution in that with our solution -- and that's unique to our solution, lower voltages and our currents, the more the merrier with the so-called competitive alternatives. It's very difficult to average them to a fractional volt from bus voltage that is acceptably efficient at distributing power. There's a fundamental conflict between efficient power distribution and the ability to support very low voltages because the alternative technologies lack the current multiplication function that is a core of factorized power.

The next question is coming from the line of [Jon Dylan].

So Patrizio, I wonder if you can give us an update on the RFM.

So we have a 2G RFM and we're getting very close to the 4G RFM. We got limited scope of the engagement intentionally so with the 2G, which is -- in effect, is counting product. It's not -- it's intended for really specialty application as opposed to high-volume mainstream applications, and that's what the 4G RFM is intended for. We actually recently saw a step-up in interest in at least one 2G RFM application. We're seeing other interest for 4G RFMs as with the front-end products for which we got a very substantial order late last year. I think the future of RFMs is very bright for us, very complementary to our point-of-load solutions. I would say with other internationally because customers need to power the systems typically from AC sources, and the 48-volt bus is the intermediate step which, with our technology, brings it all together. So the RFM, in effect, takes the power from the source and delivers it at the 48-volt level, which is safe. It's sufficient to distribute, and it's capable with our current multiplier technology to then be converted here actually to 7-nanometer nodes or 5-nanometer nodes with very high levels of efficiency. So the RFM products with upcoming 4G versions are going to be much high-performance and much more cost-effective than 2G, plays an important role into the overall strategy.I'll make one comment with respect to the strategy. We're the only company that has the breadth of capabilities to address general power system needs with a power component methodology that spans the gamut from 48-volt AC sources all the way to, as Phil was saying earlier, 0.35-volt subthreshold implementations of AI chips. And the RFM is a key element of the strategy. It's -- if you will, the jumbo jet that takes the power to an efficient hub, which is 48-volt, on the way to the point of load.

Yes. It's really -- it sounds really exciting. When do you expect we would see initial revenue for that? And then when do you expect we would see production revenue for the fourth-generation version?

So we are actually engaging on -- for the RFM development. We have a controller chip that's due in about 6, 7 weeks, and we're going to start powering things up in 4G RFM land in the second quarter.

Excellent.

Platform straight to go with the controller due to arrive in, again, 7 or 8 weeks.

And then production revenue, when do you expect that possibly?

Well, I think that's probably still at least a year away but we're going to have design activity. We already have some design activity taking place as we speak.

Right. And the results today kind of begs the question. We've talked about diversification in previous conference calls. I'm just wondering when do you think we'll have sufficient diversification that we can kind of smooth out the quarterly bumps that we're seeing.

Phil?

Yes. I think this year is the year to do that. I think if you look at the customer list that we had going back a couple of years, it was one big data center guy, maybe a couple of small ones. But now if you look at our customer list for this year, it's expanded significantly. So I think 2020 is the year to do that. And then 2021, with all the new programs coming on with AI -- powering AI ASICs, the OAM cards, the 48 to 12, I'm very confident about the -- our position in the data center market and the growth that's going to come from that and, to your point, smoothing out the bumps that we've gone through the last year and the year before.

Yes. At the moment, we're still 2 to 3 years away, but that's going to have a significant component.

Yes. And I'm encouraged also. I mentioned the network -- cloud networking companies that have come to us and have asked us for solutions for 48 volts now and higher current networking processors. That's a new market that parallels very closely the AI and cloud computing space. It's the same challenge, the same problem. We've got exactly the right technology and products for it. And that market is very large.

So you're talking about network processors like the Cavium chip?

No. I'm talking much higher bandwidth than that. I mean the stuff that is in the backhaul -- back plan -- backhaul of data centers that's really needed to move the data around.

We've got one more question from the line of Quinn Bolton.

I just wanted to follow up on John's question about the broader adoption in data centers. It sounds like the next 12 months or so is going to be driven by the AI accelerators, the OAM modules or the NBM where it's 12 to 48 or 48 to 12. But when you talk about broader adoption of 48-volt, 48-volt racks, are you sort of implying that everything on a card or everything in the server would be 48-volt so that would be including the Intel and AMD CPUs and so you'd have opportunity for point of load for pretty much every component on the motherboard?

Yes. In terms of the high-performance compute, right, the exascale-type computing, that's already 48 volts and we do power AMD processors and high-performance Intel processors on those cards directly from 48 with factorized power solutions. And then in terms of the general, if you like, cloud computing server market, that infrastructure is starting to move over to 48 volts at the big data center companies because they're moving to add AI capabilities to the cloud. The power and the racks are going up over 20 kilowatts to 40 kilowatts. You could not use 12-volt power delivery networks in the rack for that. So that's going to take a year or 2 to happen, but that infrastructure change is happening and is being supported by OPC with the rack usage.

Got it. Yes. And then last quick one for Jamie. The OpEx ticked up in the fourth quarter. You said some of that was project-specific development charges. How should we be thinking about OpEx in Q1 and beyond -- I mean for the rest of the year in 2020?

Well, the head count expansion was essentially a replacement and truing up of -- over the year. So as you saw, the percentage increase was very small. So I don't think there's real reason to assume that there's going to be a sustained ramp. The project materials, the prototyping expense was somewhat one-off in -- it's not to say that we won't be spending a lot on prototyping but it surged a lot just from timing.

So maybe flattish in Q1 from Q4?

Yes. Not materially more. I mean again, I emphasize the personnel nature of our OpEx. It's head count.

And with that, thank you very much. We'll be talking to you in a few months. Have a good day.

Thank you, everyone. That concludes your conference call for today. You may now disconnect. Thank you for joining, and enjoy the rest of your day.