Vicor Corp (VICR) 0 Q0 法說會逐字稿

Good day, ladies and gentlemen, and welcome to the Vicor Earnings Result for the First Quarter Ended March 31, 2017.

My name is Christine, and I will be your coordinator for today.

(Operator Instructions) As a reminder, this conference is being recorded for replay purposes.

I will now turn the presentation over to your host for today's call, Mr. James Simms, CFO; and Dr. Patrizio Vinciarelli, CEO.

Please proceed.

Thank you, operator.

Good afternoon, everyone, and welcome to Vicor Corporation's First Quarter 2017 Conference Call.

As stated, I'm Jamie Simms, and here with me in Andover are Patrizio Vinciarelli, CEO; and Dick Nagel, Chief Accounting Officer.

Today, we issued a press release summarizing our financial results for the period ended March 31.

This press release is available on the Investor Relations page of our website, www.vicorpower.com.

And we also filed a Form 8-K earlier today with the Securities and Exchange Commission related to the issuance of this press release.

As always, I remind listeners, this conference call is being recorded and is the copyrighted property of Vicor Corporation.

I also remind you, various remarks we may 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 the statements regarding customers, opportunities, sales growth, spending and profitability, are 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 implied by any of our remarks today.

The risks and uncertainties we face are discussed in the 2016 Form 10-K we filed with the SEC on March 7, 2017.

Please note, the information provided during this conference call is accurate only as of today, Tuesday, April 25.

Vicor undertakes no obligation to update any statements, including forward-looking statements made during this call, and you should not rely upon such statements after the conclusion of the call.

A replay of the call will be available beginning at midnight tonight through May 10, 2017.

The dial-in number is (888) 286-8010, and the passcode is 77516145.

In addition, a webcast replay of today's call will be available shortly on the Investor Relations page of our website.

I'll start this afternoon's discussion with a review of our financial performance for the first quarter, and Patrizio will follow with his comments, after which, we will take your questions regarding our business.

Beginning with consolidated results.

Vicor recorded, as stated in this afternoon's press release, a net loss for the first quarter of $974,000 or $0.02 a share compared to the fourth quarter net loss of $2.7 million or $0.07 a share.

Consolidated revenues for the first quarter increased to $54.5 million from $48.1 million for the prior quarter, representing a sequential increase of 13.3%.

Gross margin dollars rose 10% sequentially.

The product mix and the impact of ramping production of new products caused a decline in gross profit margin to 43.3% from 44.7% for the prior quarter.

For the first quarter, we recorded an operating loss of $1.4 million, $1.2 million less than the prior quarter's $2.6 million operating loss.

I'll now break down our revenue for the quarter.

The BBU, Brick Business Unit, consisting of our Andover-based Brick product lines, Vicor Custom Power and the Brick-based activities of our Japanese subsidiary, VJCL, recorded a 5.7% increase in revenue, led by a rebound of shipments of Andover's legacy products.

VI Chip recorded the highest dollar increase of revenue of our segments, up $3 million sequentially, representing a 32% increase, much of which was associated with initial shipments of the ChiP VTM portion of our VR 13.0 data center solution.

Similarly, Picor recorded a 39% sequential revenue increase, also reflecting initial shipments of the SiP PRM portion of the VR 13.0 solution.

International revenue, which we identify by the ship to address, increased 22.2%, reflecting increased shipments of the VR 13.0 solution to Asian contract manufacturers, as well as an increase in shipments of Brick products to Asian customers.

For the first quarter, consolidated turns volumes, that is orders received and shipped within the quarter, rose 9.8% and was steady at 41% of Q1 revenue.

Concluding on consolidated revenue, recognized distribution revenue declined slightly quarter-to-quarter, totaling $4.6 million.

Turning to product level profitability, as stated a moment ago, consolidated gross profit margin as a percentage of sales declined to 43.4% from 44.7% for the prior quarter.

This decline was largely driven by the mix of products and the ramp of new products making up the quarter's increased volume.

Consolidated results mask an ongoing improvement of gross margins for VI Chip as volumes increase.

Consolidated operating expenses sequentially rose 4%, but the increase, all within R&D, was expected in view of the high level of design activity.

Approximately 1/3 of the increase was a shift in the timing of expense recognition and expense deferral associated with certain customer-funded engineering projects.

Regarding income taxes.

Our first quarter tax calculations did not reflect any unusual or non-recurring activity.

Turning to cash flow and our cash position.

Operations experienced a cash flow deficit of $1.3 million, due to the net loss for the period and a shift in working capital associated with increased revenue.

After capital expenditures of $2.6 million, our quarter-end cash and equivalents balance declined $3.3 million to $52.8 million.

We do have the liquidity to fund the anticipated growth further -- including further capital expenditures for expanded manufacturing capacity.

The quality of our receivables portfolio remains excellent, with first quarter days sales rising to 44 days from the fourth quarter's 41 days.

Similarly, our aging schedule is in excellent shape, with no meaningful past dues as of today.

Annual turnover of consolidated inventories continues at a desired level, declining slightly from 4.3x to 4.2x, again, reflecting intentional increases in raw materials by VI Chip and Picor, in anticipation of scheduled demand.

No unusual activity occurred in our inventory reserve accounts for the quarter.

Employee headcount at quarter end was 998, of which, 967 were full-time employees.

Our full-time headcount increased 8 quarter-to-quarter, with 7 of these hires filling engineering positions.

Total headcount increased by 27, but as has been the case with recent swings, changes in temporary headcount have been associated with University co-op students as well as categorization changes as we manage the staffing of and the number of shifts in the factory.

We continue to run 2 5-day shifts for Brick manufacturing and 3 5-day shifts for VI Chip manufacturing, although, we are implementing weekend shifts as required.

Turning to new orders.

First quarter bookings increased 5.1% to $57.9 million from the prior quarter's 55.1% -- excuse me, $55.1 million, and 1 year backlog rose 6.9% quarter-to-quarter to $51.7 million.

One highlight of the quarter was a 44% increase in VI Chip bookings, reflecting new orders for VR 13.0 solutions and Factorized Power solutions for supercomputing.

We also accepted higher volume orders for certain VI Chip projects that have been in early low volume phases over recent quarters.

The rise in VI Chip bookings was offset by a 5.8% decline in BBU bookings, a reflection of the trends we've seen in legacy markets in certain geographies.

In addition, Picor bookings were off, reflecting the rather idiosyncratic way contract manufacturers manage their supplies of ChiP VTMs and SiP PRMs, which have different lead times.

Nevertheless, the bookings for the quarter were encouraging and we see the diversity of activity increasing.

As Patrizio will address in his remarks, momentum with our new advanced products further improved during the first quarter with important design wins.

Turning to our outlook.

Given our backlog, current turns volumes and our near-term forecast, we anticipate further improvement in our top line for the second quarter of 2017.

As we're not forecasting meaningful changes in mix, gross margin or spending, we are expecting bottom line profitability, starting with the second quarter.

I must remind listeners, as I regularly do, of the difficulty of accurately forecasting sales cycles for disruptive innovative technologies.

Similarly, the decisions made by early adopting customers and their contract manufacturers and the timing of those decisions are subject to changes brought about by many factors out of our control, with the consequence being sudden and unanticipated changes in operating and financial forecasts.

With that, I'll now turn the call over to Patrizio.

Thank you, Jamie.

As Jamie noted, our VR 13.0 business finally started ramping in the first quarter with VI Chip and Picor reporting improved results.

Despite the complexity of ramping new products in the midst of new equipment installation for expanded capacity, our factory was able to improve VI Chip margins, while increasing top production volume by nearly 46%, and production of VTM chips for VR 13.0 data center solutions by 66%.

As outlined in today's press release, we've also been preparing from a ramp of AC to 48 and 48 to Power-on-Package or PoP solutions for advanced supercomputing installations, which is taking place in Q2.

Our progress with bookings supports our expectation of improving revenues and profitability as soon as the current quarter.

High levels of engagement with customers across market segments, notably, data centers, supercomputing, artificial intelligence, machine learning, autonomous driving and outer space are expanding.

During my Q4 remarks, I mentioned our design wins for powering GPUs and for powering CPUs within the CPU package.

Since that time, we've expanded the scope of our engagements and believe we've established a strong competitive advantage in the emerging, high-growth market segments for our current GPUs, CPUs and ASICs that enable accelerated computing applications, including deep learning and artificial intelligence.

We expect that this segment will provide significant contribution to our growth over the next several years.

We are particularly well-positioned to provide PoP solutions in which our modular current multipliers, so-called MCMs, are co-packaged within the ASIC package or on a substrate next to or under the ASIC, to enable efficient and fast delivery of hundreds of amperes of currents to turbocharge ASICs.

For emerging computing applications, our Factorized Power approach to powering GPUs, CPUs and ASICs within the package is game-changing, with power density, efficiency, response speed and total performance attributes far above any alternative.

With Factorized Power MCMs capable of supporting (inaudible) computing requiring peak hours sometime reaching 1,000 amperes, we're enabling customers to achieve considerably higher GPU, CPU and ASIC performance.

I joined our team at the Open Compute Product Summit in March, and was quite pleased to see the level of attention paid to the emerging 48-volt infrastructure, which Vicor pioneered over a decade ago.

Last year, Google joined the OCP, pledged to contribute their 48-volt chassis, which they did in August, and otherwise, put its spotlight on the performance benefits of 48-volt systems.

Within last year, the servers community perspective on 48-volt has changed from curiosity to interest to early adoption, and this was obvious at this year's Open Compute Project meeting.

Numerous vendors made presentations or previewed products to enable an expanding 48-volt ecosystem.

The onset of competition, particularly from companies with roots in power management, is a welcome development as it accelerates widespread adoption, expanding the market opportunity for all, while Vicor enjoys the competitive advantages of superior performance rooted in its unique intellectual property and long-standing product development initiatives.

Vicor did some time on the stage, presenting some more mature 48-volt direct to the point of load solutions.

We also participated in high-profile [ public ] discussion with Google and Facebook.

Google and Rackspace presented the Barreleye chassis and various server board incorporating Vicor's Factorized Power architecture system utilizing cPRMs and Chip VTMs to power IBM P9 processors and memory rails.

This system is due to be in production later this year.

I'll conclude my remarks by highlighting an announcement that was made during OCP by Wiwynn Corporation of Taiwan, a leading provider of computing and search products to the cloud infrastructure market, and a significant manufacturer of equipment for Microsoft Azure's infrastructure.

Wiwynn rolled out the first OCP-certified merchant server platform utilizing a 48-volt bus for the processor and memory rails.

Wiwynn merchant product named M1 is based on Intel's Skylake CPU and is powered by Vicor's Factorized Power Architecture cPRMs and Chip VTMs.

This merchant board represents a step beyond a reference design in that it is OCP-compliant and ready for customer deployment.

As the purpose of OCP is to drive standardization, we believe the introduction of a turnkey state-of-the-art motherboard using Vicor components to power Intel's Skylake processor is a credible harbinger of broader acceptance of 48-volt systems, the architecture that Vicor has been developing and perfecting since 2002.

I'm sure listeners have many questions.

So we will open the call.

Operator?

(Operator Instructions) Your first question comes from the line of Mark Lanier, representing Pegasus Capital.

Patrizio, I was interested in the design wins during the quarter, and taking your usual steps not to disclose names and too many specifics, would be interested in where you see these design wins developing and how it informs your sense of the trajectory of some of the markets and the relative rates of adoption?

Well, without naming names, as suggested in the prepared remarks, we're seeing an accelerating rate of interest and outright adoption, in some cases.

In the general compute space, in its various forms, particularly where as suggested in prepared remarks, the demand for (inaudible) to power low-voltage, very high current logic devices is pronounced.

So the particular area of incremental excitement and opportunity has to do with variety of devices that nowadays are approaching 300, 400, 500 amperes of, in some instances, continuous power draw, with peaks ranging up to as much as 1,000 amperes.

GPU is an example of this kind of a device, but not the only one.

There are custom ASICs that also fall in that category.

Their application fields are, as suggested in the prepared remarks, related to the various new fronts in which the compute landscape is emerging to enable very advanced systems that are capable of performing functions that years ago were unimaginable.

Equally unimaginable is the challenge of powering these kinds of devices, which in particular instances, while drawing currents that while large are -- within what was envisionable in the hundreds of amperes, a burst of current demands in turbo modes of 1 form or another that take them up to double the current capability.

And it's in that space in particular that our solutions stand out as being particularly capable and very highly differentiated.

It is true on the motherboard or on the cartridge, it is particularly true on an ASIC substrate.

Well, while it's possible to supply these currents from a motherboard through the package pins to the ASIC, with our solutions, because of its extremely high density, low noise attributes and dynamic response characteristics, it is also possible for the first time to co-package a current multiplier within the ASIC package itself.

So we have first instances of this kind of deployment.

In fact, the supercomputing application referenced in the prepared remarks is an example of this kind of a system.

This quarter, we're shipping nearly 40,000 modular current multipliers to power advanced devices within the device package as part of a large supercomputing installation.

We see more and more of these kinds of opportunities emerging coming from makers of CPUs, GPUs and/or users of these kinds of devices in the U.S., in Asia and particularly in Japan.

Your next question comes from the line of [ John Dillon ], representing [ D&B Capital ].

So Patrizio, I've got a question for you.

Right now, you're running 3 shifts for the VI Chips.

So I'm wondering, what are you doing to ramp production?

I mean, it sounds like you're expecting your bookings and revenue to continue to increase.

If you're already running 3 shifts, how are you going to get the additional production out?

Well, we have a very capable operations team that has proven its capabilities.

So this is not something that keeps me up at night.

We know the capacity.

I think, we, learning from past experiences, their impact, through the earlier days of the company, we learned how to rely on scalable equipment and scalable processes.

So we have a good deal of elasticity in our manufacturing model and we look ahead and we make contingency plans that provide the 4 large factors of expansion or capacity, particularly, for chips, which are going to be at the heart of a lot of the systems that I was referencing earlier, later this year, into next year and the year after that.

So lead times for the kind of equipment that we need to scale up capacity are typically measured in the 3 to 6 months range.

We believe, within our Andover facility, we have the capacity to at least double the business to the $400 million range and possibly the $600 million range.

And beyond that, we'd be looking at another facility, possibly in Asia to further scale up capacity.

So all of these can be done by, in effect, expanding the model through the same kinds of sales, the same processes, again, with good predictability and scalability.

The challenges typically encountered in the very early going of new products, so over the last 1.5 years in particular, there was a deal of learning with the converter house and packaged technology, which is fundamental in every technology for power components is the technology that enables modular current multipliers to be co-packaged within ASICs is also a technology that enables the latest and greatest PRMs and VTMs.

There was obviously quite a bit of challenge for our operational team and engineering teams in the early going, but we've learned a lot and we now have, again, a great deal of scalability, Witness the fact that from last quarter to this quarter, we're going from, I suggested earlier, very low, small quantities of MCMs literally 10s or few 100 samples to nearly 40,000 units in 1 quarter.

And we'll be scaling that up again in the third quarter by, I think, a factor of 4. So we can do the scale up once we have the recipe in place, and we believe we now have, for these packages, the recipe in place to do it with.

Wow.

So it sounds like you've got capacity for $400 million to $600 million in your current facility in Andover alone?

We have the space and we have the plans to support the level of capacity.

Just to be clear, we're not expecting to be at the run rate in the very near future, but we do see how we can configure our Andover facility to support the level of business.

And again, depending on how things evolve in the markets that we're addressing, and some of these markets are very dynamic, there may be scenarios where we want to accelerate the level of expansion of capacity and we'll look at some point at facility other than Andover.

Right.

So it's obviously very dynamic, but you've guided in your plan to be able to grow up to $600 million in your current facility?

$500 million to $600 million in our current facility.

Okay.

All right.

So bookings and revenue for the rest of the year, do you still expect sequential increases quarter-to-quarter in each of the quarters?

Yes.

And the evidence to date, this quarter, supports that proposition.

Okay.

So your bookings are trending pretty well this quarter already?

Yes.

Okay.

Good.

Good.

And the GMs going forward, it sounds like you had some of the VI Chip revenue was -- GMs were -- gross margins were masked because of some issues, but do you expect the gross margins to continue for the rest of the year then?

Yes.

So fundamentally, there are 2 key elements to improving the gross margin parameter.

One is absorption of overhead.

We have a large fixed cost infrastructure that we put in place quite some time ago to give us the technological capabilities, the packaging technology, the product mix that is now beginning to bring dividends.

As the level of revenues scales up, that by itself does the right thing in terms of the margins.

But also, with ChiP technology, in particular, we have a very, very cost effective product platform.

One is much more cost effective than the early VI Chips that were made if we go back several years in time frame, which IBM was the primary customer for these kinds of products.

Those devices were literally crafted on a case-by-case basis, no pun intended, with dedicated molding cavities and not very scalable capacity.

Everything we make today that is destined to the markets we're discussing is instead made on so-called panels, in particular.

The panels go through a process that is generally referred to as converter housing package.

Within these panels, we can in effect segment the small chips, of which we can make a very large number, let's say, 100 per panel in that particular instance.

They may be relatively low-power devices that may sell for $10, $15, or in some instances, less than that.

At the other end of the spectrum, our other panel, the same kind of a panel going through the same manufacturing process, we could make as few as capital devices, which could process 6 or 7 kilowatts, and those chips used to sell for more money.

But whether they're large or small, whether these -- many of them per panel or just a few, the manufacturing process is exactly the same.

And fundamentally, a factory, just like a semiconductor wafer foundry, is set up to process panels as we would -- as factory processes wafers.

And again, this is unique and very proprietary and every patented technologies, something we've been developing, again, over the last 10 years.

So that's the manufacturing scalability.

The manufacturing model.

It's at the heart of these initiatives.

The next question comes from the line of [ Jim Bartlett ], representing [ Bartlett Investors ].

The current quarter was heavily driven by VR 13.0, is this still VR 12.5, and is that tailing off and will be ending soon?

Yes, there's still some 12.5, it will tail off mid-year.

I think, most of the action going forward is going to be VR 13.0 for Intel sockets.

Again, there's going to be action in applications that are not related to powering Intel processors.

Reference was made in the prepared remarks to IBM P9.

There's quite a bit activity initiative relating to that for a variety of reasons.

There are proprietary ASICs.

Coming back to Intel sockets, there are Intel sockets where we power the processor and the memory rails that have not yet even started ramping, that we're still ramping in the second half of this year because Intel processors for those kinds of applications are not quite yet available.

When will they be available?

Based on what we know, second half of this year.

And the significant installation that you mentioned, the supercomputer installation where there was significant shipments, is that just 1 supercomputer installation, and does that then tail off?

Or are there more similar things coming?

Or could you just help me explain.

Yes.

So there is the first supercomputer installation.

That's a pretty large installation but it's 1 device that will be complete in a few weeks.

And so we are actually shipping at relatively high rates.

We'd be shipping at regular rates over the last couple of weeks and over the next couple of weeks to complete our part.

That is the power system of that installation.

And to be clear, this is a first installation where we're not only providing the power components that deliver high current or low voltages to the processors and memory rails of the point of load from a 48-volt bus, but we're also providing the front-end power system that takes AC, pre-phase at megawatts level of power and delivers it to a 48-volt bus within the supercomputing unit.

Obviously, we're not doing it in 1 track.

There's a larger way of what we call RFMs or rectifier modules.

In the first installation, they're 8 kilowatt devices, and there's 500 of them.

And this is an installation taking place, again, in the second quarter.

In the third quarter, there will be a larger installation, which is 4x the size of the first.

The RFM will scale up to 12 kilowatts per device and there will be 2,000 of them.

So -- and these 2,000 front-end devices are going to power a much larger multiplicity of so-called MCDs or modular current drivers and MCMs which are the current modular current multipliers that are co-packaged with the ASICs that provide the compute nodes.

So in answer to your question, a first significant step up this quarter, a follow-on next quarter at 4x the volume, and we'll see where it goes from there.

Would you say that this paradigm is a significant advantage for the supercomputer manufacturer that could be replicated or want to be replicated by others?

Yes.

I think that the level of compute density and capability that we're enabling with this first installation is state-of-the-art.

The one that we will be enabling with the follow-on installation, which is just a few months later, 3 or 4 months later, will raise the bar further.

But while this is indicative of what can be done at the supercomputing level, it is also, on a different scale, indicative of the benefits and advantages of the technology, if you will, with each ASIC, GPU or CPU that has high current demands, very high current demands.

And to be clear here, Intel processors are not in that category.

Intel invested a great deal quite some time ago in developing its own on-chip regulator technology.

It is proprietary at Intel.

It is used within Intel devices.

And it relaxes the burden in terms of current delivery to an Intel processor because this internal infrastructure provides a regulation function, which in effect converts what is typically 1.8 volts of this input to what would typically be 0.8, 0.9 volts at its output.

And it's got capability to deliver a multiplicity of such voltage rails.

That's rather unique to Intel.

And with Intel's competitors, that kind of capability isn't generally there.

But the current needs, in some instances, at the actual point of load are even larger, potentially much, much larger.

Hence, there's a void there and there's a great opportunity to fill that void with our modular current multiplier technology, which we believe to be superior to the technology which is embedded within the Intel's package.

Superior in terms of being able to multiply the current not by a mere factor of 2 or 3, but by a factor of 48 or 64, so as to reduce the level of current delivery to the processor package by a corresponding multiplier.

So case in point, an Intel device, while embedding within the device, an Intel regulator to, in effect, expand the current capability of the device, still requires 100 amps or 150 amps or 200 amps at its input pins.

And these pins, as you can imagine, take half -- quite a bit of the real state in terms of I/Os of the package and on the motherboard next to the package.

With our modular current multiplier technology, essentially, the power delivery goes down to a few I/Os.

So fundamentally, what we do for the supercomputing application that I referenced earlier, for other kinds of applications, in different parts of the computing world, what we do is reduce the current delivery to supply an ASIC or GPU that may draw, instantaneously 500 amps, we can, through the current multiplier, cut down the current delivery by a factor of 48 or 64.

So instead of having to bust in 500 amps, we can enable the level of power delivery that is required at the current of only 10 amperes, which is much easier to deliver to the package on the motherboard, with less copper weight on the motherboard, with much less issues in terms of the power delivery on the motherboard, with much fewer pins on the package itself.

And fundamentally, by reclaiming those I/Os for computing connectivity as opposed to power delivery, we enable a fundamental advantage, competitive advantage in using our modular current multiplier technology.

And this is something that is still really not in the general radar screen, it's only in the radar screen of a few early adopters and early customers.

And how would you rate your opportunity in Intel versus the non-Intel segment?

Well, obviously, Intel applications are still the bread and butter of our VI ChiP revenues and they will continue to be a large portion of our opportunity.

But I think, it's important to note that there are competing solutions.

It's important to be aware that these competing solutions, lacking the internal regulator that Intel has developed, are even more dependent on a very efficient, dense, fast responding, low noise alternative, and that's what we're providing with MCMs.

And just to give you a little bit more of a flavor with respect to what an MCM is, MCMs and MCDs relay to PRMs and VTMs.

So there's still building blocks of factorized -- proprietary Factorized Power system, but the partitioning of the functions between the PRM and the VTM is done in a different way so as to enable a much thinner, denser solution that can be much more readily co-packaged within an ASIC package or a GPU package.

So just to give you some dimensional points of reference, our voltage transformation modules, which typically power Intel processors on a motherboard -- or from a motherboard, I should say, those devices are about 5 millimeters thick.

MCMs are 2.5 millimeters thick.

So they're half the thickness.

Also, their current density is substantially higher than VTMs.

So if you look at the x and y footprint in current density capability in terms of x and y, and also look at the z or height of the package, you get a factor of 2 advantage in height, which makes co-packaging within an ASIC package or on the bottom of ASIC package much more pragmatic, practical.

And also, you get, again, more density in terms of actual footprint.

So our MCMs can flank an ASIC on 2 sides.

They can be mounted on the opposite side.

They can be mounted under the lead of the ASIC.

They can be mounted outside of the lead of the ASICs.

There's a lot of flexibilities with MCMs to enable much higher current devices, particularly where as is often the case with some of these devices, the current draw can have instantaneous loads that far exceed the average because both our VTM technology and our MCM technology is capable of burst power delivery without penalty, unlike competitive alternatives that rely on multiphase regulators, so 1 form or another, which are all, in effect, limited by their serial inductors.

Our technology does not involve any serial inductors.

There is no energy limitation.

The power delivery is really only capped by thermal limitation.

So with our technology, on a timescale of milliseconds, if the computer node needs to, on a short timescale, call on extra resources to rapidly solve a challenge, you can do so without concern with respect to power delivery getting in the way of enabling that challenge to be solved.

You described it as game-changing, and it sounds like it is.

It is.

Your next question comes from the line of Hasnain Karim, representing Kilimanjaro Capital.

Just for the first question, in the opening script, you mentioned diversification in terms of bookings, and I was wondering if you're speaking to diversification within the large cloud providers as well with regards to Skylake introductions?

Or were you talking more to Intel versus non-Intel processors in terms of bookings growth?

I think, it's all of the above.

And the other element in the mix, which is beginning to payroll this quarter more so next quarter and going further into the future is the AC to 48.

So again, Vicor is, because of our technology, both the power conversion engines, the packaging technology, uniquely equipped to address power system needs from the source to point of load.

And part of the diversification strategy is to combine point of load sales with power system sales in the front end because, typically, customers require both.

Historically, those needs have been addressed through different vendors with disparate technologies and without the opportunistic comp from a grander vision and a more homogenous capability, which is what we're beginning to deliver.

So in the short term, the diversification is going to come primarily from expanding our footprint in the data center space, cloud computing, supercomputing of 1 form or another.

But longer term, it is also going to come from power system solutions that are not the point of load devices that may come from the short-term revenue outlook.

Okay.

If you look out a few quarters, looks like by the end of the year, when Skylake is in general availability, do you expect to be shipping to at least 2 or 3 of the top 5 clouds vendors in the world?

So I'm not going to be specific with respect to that.

Beyond the prepared remarks, I can say that we see traction across a broad end spectrum.

As suggested in the prepared remarks, as recently as a year ago, in March, I think, when Google made this announcement at Open Compute 2016, they were the only customer for 48-volt solutions.

As you've heard, a lot has changed in the last year.

But to be clear, as pointed out in the prepared remarks, we've gone from being a curiosity to being a lot more than curiosity, more early adopters getting involved.

But we're still scratching the surface of the opportunity.

There's no doubt in my mind that given all of the benefits of a 48-volt infrastructure, which many people don't fully yet appreciate, given the many benefits, it's only a matter of time before this conversion is going to be very pervasive.

But it is not going to be an overnight change.

These are infrastructure changes that, on the part of major cloud providers, require a fundamental change in their power system architecture, something that obviously requires quite a bit of preparation and time to deployment.

And so it will take time.

It will take years.

This is not a timescale of months.

It's a timescale of years.

But what I can predict is that the ones that are going to make a judicious change sooner are going to reap those benefits earlier, as Google itself has advertised.

Okay.

So is it fair to say, I mean, a couple quarters of ago, last call, I think, your prognostication was small changes in bookings but still double-digit at times to both the fourth quarter and the first quarter.

And then, as Intel starts production of Skylake, you expected a step up in bookings and backlog in the June or October quarter.

Is that still how you see slope of growth?

Yes.

We see a monotonic growth.

I think, the rates of growth from quarter-to-quarter may be different.

I think, we may be able to do better this quarter than the last quarter, but we'll have to wait and see until the end of the quarter.

And what the third quarter brings about and fourth quarter brings about, let's wait and see.

But our forecast shows continuing improvement in preparation for what should be also a very exciting 2018.

Okay.

Just last question for me.

With regards to the gross margin, you mentioned that some of the headwinds this quarter was start-up costs.

As you see revenues getting to about $60 million or so, I guess, in the second quarter maybe, if you could talk about it, are most start-up costs done with at this point or do you still expect some sort of increased cost with regards to gross margin as you ramp in Q2 and Q3?

So this quarter, as mentioned earlier, we are in fact executing very well, I would add, or as well as could be expected.

A step function from very little to tens of thousands of MCMs, MCDs and commensurate quantities of RFMs.

This is taking place this month into early next month before we start waiting for the next wave of these kinds of devices.

I would expect that the efficiency of making these particular products will come to the fore as we get past this first wave and begin to execute on the next wave.

Yields are getting to the right place, but inevitably, in this kind of a ramp, there's a price to be paid for some level of novelty in the products themselves.

So I wouldn't predict, at least for those products, the kinds of efficiencies we're going to have starting in the third quarter until the third quarter.

With respect to other products, I think that higher volumes will bring about, again, greater efficiency.

I would say that the demands on the operations team to really step up to the bar are in terms of addressing these novelties are largely behind us at this point.

And that while not coasting, don't get me wrong, I think, it should become more routine and more efficient to deal with volumes coming around for these kinds of device and other devices in the second half of this year.

Okay.

Maybe I can just add one more.

Sure.

What -- where would do you think -- or I should ask, how are you guys thinking about licensing your technology as the TAM grows broader and customers, beyond the initial early adopters, ask for second source vendors?

So we are open-minded with respect to that.

We've done the kind of thing in the past.

We're prepared to do it.

Remarkably, there hasn't been that much pressure for the proverbial alternate outsource.

And I rationalize this reality by virtue of the compelling competitive advantage that we offer.

It's not unusual in our industry for customers to want to capture the benefits of a technology even though it's proprietary and single source.

So contrary to what I may have feared, I see more and more substantial customers willing to go single source with us because they visited our factory, they become very confident with our capabilities.

We've been around now for quite some time.

We have a good reputation.

We don't let customers down.

And we live up to our obligations.

So all of those considerations coupled with the benefit that our products offer in terms of enabling our customers to get a leg up in their competitive challenges with their competitors, it's been working wonders.

And we haven't really been muscled into any alternate source arrangement.

But we're open-minded.

We recognize that our objective is to make the most of the opportunity.

As you might have heard from past conference calls, we've invested, I think, at this count is probably close to $300 million in developing this technology.

And the money part is, in some respects, not doing justice to the sweat, the ingenuity, the effort over the last dozen years.

So we don't want to limit the opportunity of what we've invested by being too possessive of it.

But as good businessmen, we do want to make the most of the opportunities.

So should the opportunity arise, when the opportunity arises, we'll be very open to other kinds of arrangements to further accelerate the market uptake of our technology.

The next question comes from a follow-up from the line of [ John Dillon ] representing [ D&B Capital ].

Patrizio, you mentioned before, you have a $75 million target or plan for the fourth quarter this year.

Is that still on track?

Pretty much.

I'm not going to spell out a number, but as suggested earlier, we look at improvements in the 5% to 10% range taking place this quarter, and as a ballpark, next quarter and the quarter after that.

Great.

And in your remarks, you mentioned Wiwynn, and I think, you mentioned Microsoft.

I wasn't sure if you meant that you had a design win with Microsoft through Wiwynn, or was that some other reference?

No.

I think, the remark is the way it was framed.

Wiwynn happens to be a substantial supplier to Microsoft, and Wiwynn took the initiative of developing a solution, a 48-volt based with our building blocks.

What Wiwynn does with that solution, I can't comment on.

And you shouldn't infer from that.

The fact that they happened to be a substantial supplier to Microsoft, that there's any connection with Microsoft, that would not be the right inference.

But I think, it's in of itself noteworthy that they took the initiative.

And they're not the only ones.

There are many such initiatives with various Asian contract manufacturers doing similar kinds of things with a variety of end customers in mind.

So fundamentally, what's behind it is a broadening realization that 48-volt infrastructure is the way of the future, not just in data centers but in a variety of application environments.

And not surprisingly, companies that play the key role in the food chain supporting infrastructures that historically have been based on 12 volts will now be on the forefront of the evolution to 48.

And with that, we have an opportunity, they have an opportunity, and their customers, frankly, have the biggest opportunity.

Okay.

So I think, the point is, is that Wiwynn is a very substantial cloud provider, Microsoft being one of its customers.

But on their own dime, they went out and developed this 48-volt technology and they're offering it to their customers.

That's correct.

That's right.

If there's one more question, we'll take it, it's getting a little late.

Your final question comes as a follow-up from the line of Mark Lanier, representing Pegasus Capital.

Patrizio, a number of years ago, as Picor was conceived and developed, you articulated a potential path of Picor becoming an independent company as I recall.

That discussion hasn't been as crucial in the last few years.

But I'm wondering, as you take a look at the growth paths going forward, the requirements for various markets leveraging manufacturing, I'd be curious to know how your thinking about that has evolved?

Well, it hasn't really evolved in that there's no change with respect to the general plan.

We've been able to attract a key talent.

We've incentivized this key talent through financial goals that are tied to equity ownership in Vicor entities that would see the light of day in terms of liquidity opportunity.

To achieve that, it takes, as you know, critical mass.

We haven't had it.

We will have it.

As soon as we have it, we're going to have the liquidity opportunity.

Ladies and gentlemen, thank you for your participation in today's conference.

This concludes the presentation.

You may now disconnect.

Good day.