Micron Technology Inc. (MU) CEO Sanjay Mehrotra on Bernstein's 34th Annual Strategic Decisions Conference (Transcript)

| About: Micron Technology (MU)

Micron Technology Inc. (NASDAQ:MU) Bernstein's 34th Annual Strategic Decisions Conference Call May 31, 2018 1:00 PM ET

Executives

Sanjay Mehrotra - President and Chief Executive Officer

Analysts

Mark Newman - Sanford C. Bernstein & Co.

Mark Newman

Afternoon, everybody. I hope everyone had a great lunch. We’re going to kick off the next session with Micron. Just a brief introduction. I’m Mark Newman. I cover memory globally here at Bernstein, and great pleasure to welcome Sanjay Mehrotra, who is the CEO of Micron.

Last time, he came here to present as CEO of SanDisk, so it’s great to welcome him back here in his new role heading up Micron on the back of what was an amazing Analyst Day just about a week ago in New York with some great results, some great numbers and pretty impressive buyback.

So what we’re going to do now is, Sanjay is going to make a 10-minute or so presentation to kick off things, and then we’re going to go to Q&A. Well, I’ll start off with some questions and then please do feel free to write down your question on your cards. Someone will bring them up to me and I’ll try to get your question in as well.

And I’d like to introduce you to Sanjay.

Sanjay Mehrotra

Good afternoon. Thank you for being here. It’s a pleasure to be here sharing the Micron message with you. Before I share some information, let me just point to the fact that I will be making certain forward-looking statements here. And please do refer to our 10-K and the various filings we make with SEC from time to time regarding any of the risk factors associated with the business.

I will highlight some of the aspects that we discussed last week at the Investor Day during my presentation. First, want to point out that the market for DRAM and flash are strong, are diverse and are secular in nature. And the applications all around us are demanding more and more memory.

So from memory and storage point of view, when I say storage, I mean, flash here. And when I say memory, I mean, DRAM. The applications are most hungry for more memory and more storage. I’ll talk about a little bit about the market here.

In terms of the industry, in the backdrop of strong demand drivers and the increasing complexity of the technology transitions, both on the DRAM and the NAND front, as well as increasing capital intensity involved with this transition, the memory industry on the supply side is stable. I’ll share that outlook with you. And the combination of the demand backdrop and the supply outlook makes for healthy fundamentals in our industry.

With respect to Micron specifically, and we termed it as new Micron last week at the Investor Day. We have a winning strategy. We are extremely focused on execution and we believe that Micron today is best positioned ever.

In terms of the applications and the end market, let me point out that there are, of course, these billions of – tens of billions of devices that are connected that are using data, that are creating data, that are analyzing data and providing experiences to the consumers, as well as important business information to the businesses.

And these billions of connected devices, of course, are also relying on the cloud to access information, to process information and to provide these devices valuable insight. And these devices are your smartphones. They are your PCs, they are your surveillance cameras, your smart homes, your smart vehicles, smart cities, I mean, these are billions of devices that we are talking about here.

They use the cloud to derive data, as well as valuable insights in order to provide experiences and value to the businesses and consumers. This in turn feeds more demand for cloud. Cloud with the trend, emerging trend of artificial intelligence is able to provide even more deeper insights which creates new types of devices at the edge, which provides more value to these devices, these tens of billions of devices at the edge.

So you see it’s a virtuous cycle, virtuous cycle where data-driven value is being used to really unleash new business models, save money to the businesses, as well as provide consumers worldwide great experiences. And this virtuous cycle from the cloud, more data in the cloud, more data in the devices is driving strong demand across the Board for more DRAM and more flash, because more memory and more storage is absolutely key to extract the value that is available through all of this data that is being created all the time.

In 2017, about 22,000 billion gigabytes of data created. A lot of that data gets stored, gets processed. Memory and flash gets used for that. By 2021 timeframe, it’s expected that it will almost triple to about 62 billion gigabytes, 62,000 billion gigabytes by 2021 timeframe. So massive growth in data that is ahead in memory and storage, DRAM and flash are very much at the heart of these trends.

We look at data center as one of the biggest growth drivers projecting their growth to go from $29 billion of total available market opportunity for DRAM and flash in calendar year 2017 growing to $62 billion, more than doubling by 2021 timeframe. Again, data centers are able to provide the value in terms of services to the businesses and users based on the data. They need more fast memory, that means DRAM. They need more fast storage, that means more NAND.

Similarly, mobile is increasing in the opportunity as, well adding almost $10 billion of total market opportunity for memory and fast storage between 2017 to 2021 timeframe. And automotive, it’s a relatively small market, but a fast-growing market and expected to become a larger and larger opportunity as autonomous vehicles, which we call data center on wheels, require more and more memory and storage, more DRAM and more flash in order to make for a safe navigation of autonomous vehicles in the future.

But even today, the smart vehicles, the infotainment systems and a lot of the features that are in your smartphones are actually starting to appearing in automobiles of today as well driving growth for both DRAM and flash.

And Internet of Things, another market opportunity in its infancy, expected to grow over the years. And some of these market opportunities frankly, it’s tough to project exactly where they will be. They could be even much bigger. They could be underestimated. But the point is that the trend of artificial intelligence is going to drive greater demand for more memory and storage in all of these end markets.

So the growth trends in our end markets are secular in nature. And if we translate those dollar opportunities that highlighted earlier and reflect those in terms of bit growth what that shows there is on the DRAM side a bit CAGR of bit – on the demand side between to 2017 to 2021 timeframe of approximately 20% or slightly higher.

On the NAND side, the demand CAGR in the same timeframe is 40% to 45%. You can see that on the DRAM what used to be a market that was mostly about PCs in 2013 timeframe has evolved into PCs and mobile, but also data center becoming a big growth opportunity as you can see with the server bit demand that is shown here in 2017 and 2021 timeframe.

On the NAND side, it is about elasticity of demand, it is about client computing, client notebook computers, displacing hard disk drives with SSDs, as well as in enterprise and data centers more displacement of hard disk drive by SSD due to the shared total cost of ownership benefits and the high performance benefits that SSDs provide in the data centers. So strong secular growth trends, both on the DRAM, as well as on the flash side.

On the supply outlook, we project that DRAM supply for 2019 onward to be approximately 20%. In fact, in calendar year 2018 also for the industry, we project that DRAM supply growth to be approximately 20%. And on the NAND side for 2018, we project NAND industry supply growth to be in the range of 40% to 45%, and 2019 and beyond, we project NAND supply growth to be approximately 40%.

And relative to supply, I’m going to point out two factors. One, both on the DRAM and on the NAND side, technology complexity is increasing. CapEx intensity is also increasing. Technology complexity increase is giving you less bit gain first successive node transition. These are the factors that are leading to stable supply environment, both on the DRAM as well as on the NAND side.

Specifically, with respect to Micron, we’re focused on technology. In DRAM, as we have shared before and we went into details at our Investor Day, we have focused on narrowing the technology gap and deploying more advanced technologies into production to ultimately narrow our cost gap with our competitors and that’s producing strong results, and I’ll talk about that in a bit.

We’re also very much focused on driving a portfolio transformation, shifting our portfolio toward higher value solutions. For example, greater mix of solid state drives, as well as greater mix of manage NAND solutions for smartphone kind of applications. And, of course, we are focused on operational excellence, making sure we are taking cost in terms of product cost out of our every aspect of our supply chain, operations and ultimately, delighting our customers with excellent supply chain operations.

Technology is absolutely the foundation of our strategy and foundation of the company and very proud to share that Micron has 40 years history on the technology front. And today in the world, we have the most comprehensive portfolio of semi – memory and semiconductor storage technologies.

Our DRAM, our focus is on narrowing the cost gap with our competition and accelerating the advance toward next generation technology nodes. On CD NAND, we have the world’s smallest die in our 64 layer CD NAND production and that really bodes well for our position in the industry. CD NAND, we are focused on advancing our high-value solutions SSDs and manage NAND that I talked about earlier.

NOR is a technology, which relatively speaking is a small part of the overall industry. However, it makes us one-stop shop for our customers with complete portfolio of technologies. And, of course, we’re engaged with 3D XPoint technology and exciting technology opportunity for the future for us and emerging technologies as well.

So again, it’s – technology is the foundation of the company. And we have the most comprehensive portfolio in the world today. And that positions us very well with our customers.

So our excellent execution over the course of last couple of years has resulted in structural operating profit improvement for the company. And in fiscal year 2018, $6 billion of our total operating income is actually coming from the structural improvements that we have realized as a result of our progress on our strategic imperative since the 2016 timeframe.

These include, narrowing the cost gap with our competition, the progress that we have made in terms of getting our technology, advancement, ready for production and deploying it into production, so narrowing the cost gap plus increasing the mix of high-value solutions has resulted in fiscal year 2018 $6 billion of contribution toward operating profits.

Of course, our operating profits in total are higher for the year. What I’m pointing out here is that, $6 billion is coming from execution improvements. These are structural in nature and here to stay with us that we have done since 2016 timeframe.

But we are not done yet. Going forward 2018 to 2021 timeframe, there is another $3 billion that we are targeting. And we believe, it’s a relatively conservative number there. And these are also coming from further narrowing of the cost gap and further enriching our portfolio of revenue mix toward high value solutions and also improving our operations to ultimately translate into another $3 billion of structural operating profitability improvements between 2018 to 2021.

So if you look at in aggregate from 2016 to 2021, both on fiscal year basis, we’re talking about making $9 billion of structural profitability improvement of the company. And these are really things we are focused on in terms of our execution day-in, day-out, and these are the kind of improvements that are here to stay with us. All of this is resulting also in very strong financial performance for the company.

Our fiscal second quarter was a record in terms of revenue performance, as well as in terms of I’m very proud to show here free cash flow $2.2 billion of free cash flow generated just within fiscal Q2. And for fiscal Q3 last week, we upped our – we have provided a revised guidance fiscal Q3, which ends today. Last week, we had guided that the revenue for fiscal Q3 will be in the range of $7.7 million to $7.8 billion. And we had also provided a revised EPS guidance made it higher in the range of $3.12 to $3.16.

So Micron really is doing very well on all fronts. We had, of course, also focused on making all our investments in OpEx, as well as CapEx all on return on investment basis, making sure, we apply a prudent discipline in making these investments. We’re focused on continuing strong execution on some of the key strategic objectives that I discussed earlier.

And also last week, we announced $10 billion buyback starting fiscal year 2019 and at the rate of 50% of our free cash flow on an annual basis to go towards our stock buyback program. Again, starting fiscal year 2019 and at least 50% of free cash flow going toward stock buyback showing confidence that we and the Board have Micron’s ability to execute well, as well as the strong industry environment for our business going forward. And of course, also reflecting our focus on providing shareholder returns.

So with that, I would like to now open it up for questions. Thank you.

Question-and-Answer Session

Q - Mark Newman

Thanks, Sanjay. So as a reminder if you have any questions write it down on a piece of paper. We’re – we’ll get to those late – later on. I’m just going to kick off with a few questions, Sanjay. So it’s been a remarkable few years really in the memory industry. Profitability for Micron really all-time high, for the industry all-time high. And lot of people talk about consolidation, it is much, much more than consolidation, right?

Clearly, this is driven by – you talked about in your slides demand diversification, new demand growth drivers in cloud data centers, et cetera. And I think most importantly, just the shrink is becoming increasingly more difficult, which is meaning that supply growth is increasingly more and more constrained. And this is really kind of like the new memory paradigm, thesis we’ve had from several years ago is playing out even better than expected.

I guess, my question for you though just to start off is, it’s been amazing improvement in your numbers. How much of that do you think that is related to this overall industry improvement, rising tide lifts all boats versus Micron specific execution?

Sanjay Mehrotra

So clearly Micron specific execution has contributed immensely to our strong performance, financial performance this year. As I showed in my presentation, $6 billion of our operating profits in fiscal year 2018 are coming solely from Micron execution strength. They are coming from Micron’s ability to narrow the technology and the cost gap over last couple of years to increase its mix of SSDs in our product portfolio, that as an example.

But overall it’s a mix of high value solutions, this has resulted in $6 billion of operating profit and that $6 billion is structural improvements in Micron’s ability to execute, these are solely coming from Micron’s ability to execute and of course as we provide these benefits, we are taking these forward as well. And as I pointed out, between 2018 to 2021 timeframe we are looking at another $3 billion of operating profit improvement, again coming from cost, as well as high-value solutions and of course operations there.

So just to give you an example, in 2016 Micron’s sales to cloud customers were less than $500 million. And in the first half of fiscal year 2018, our sales that we have shared with you before, to the cloud customers are $1.6 billion. So if you use that run rate for fiscal year 2018 you get to about $3 billion of sales to cloud customers. So going from less than $500 million in 2016 to $3 billion plus in fiscal year 2018, that shows how Micron successfully has deepened its engagements with the cloud customers, how the demand is being driven certainly by the cloud customers, but Micron execution in terms of bringing quality products and right levels of performance and right level of delivery, this is important, otherwise this kind of growth does not happen.

And then just thinking about specifically on that topic of execution, considering your background, you came – you were one of the co-founders of SanDisk, you were a CEO at SanDisk. How much of that improved execution in your opinion is from yourself and a few other people you have brought with you Sumit Sadana and the few others from SanDisk, which is helping – is that helping, is that a part of the improvement on the NAND side perhaps?

Sanjay Mehrotra

So certainly it is part of the improvement, however I will point out that improvement numbers that I have given here are over the course of last couple of years. So certainly Micron, a strong company with a strong 40 year legacy, strong set of capabilities, certainly execution improvements were already underway even when I joined the company. We now have expanded management team, including some of the new members of the management team at Micron have doubled down on our focus on execution and of course over the course of last year, the team, entire team of Micron, the newcomers, as well as the old-timers have absolutely been extremely focused on accelerating our execution and our focus here is on acceleration, which has produced results in fiscal year 2018, but more opportunity for us to go.

Mark Newman

Yes, so actually just drilling down a little bit more on the DRAM side. I think what we are seeing in DRAM in addition to the entire industry becoming somewhat structurally better because of supply growth being much more constrained. It also seems to be that very clear less benefit from being one node ahead.

So, for example Samsung being one year ahead, one node ahead would be a 30% to 40% cost advantage is the past. Now that advantage seems to be much smaller, because one node is a much smaller advantage. I’m just curious what your opinion is on, do you think that trend is going to continue? Do you think that because each node is giving less and less benefit that as a result you will continue to kind of narrow that gap, not in calendar month, but in terms of economic impacts versus the leaders like Samsung, or do you think it’s possible that as things get really, really difficult for example when you go to EUV, that suddenly Samsung will like jump ahead again, what is your opinion on what happens?

Sanjay Mehrotra

So you’re absolutely right to note that each successive generation of technology in DRAM as well as NAND is getting more complex and is giving you less, more slow of scaling becoming more and more challenging. Each successive node is giving you less and less bit gain per wafer and that’s of course having the moderating effect on the supply growth. Successive generations of technology are also becoming more complex, needing more capital investments. So the combination of increased capital intensity and increased – reduced ability of technology transitions to give bit gain needs to moderation of the supply-side on the industry.

And yes, it gives us an opportunity that when we have to catch up on the DRAM technology nodes, certainly gives us an opportunity to catch up, bring our cost reduction faster than the rest of the industry as we showed at the Investor Day, therefore our COGS, if you look over a two-year time period from end of 2016 to end of – end of 2015 to end of 2017, we outpaced the industry by 30% in terms of cost reduction capability. So certainly when we are behind and we are trying to move faster, it gives us some benefit in terms of closing the gap, but you are absolutely correct that cost declines going forward with each successive node on a year-over-year basis will be moderate as well, absolutely right.

And regarding future advanced technology nodes, in terms of EUV, we shared at the Investor Day our roadmap on the DRAM side which actually shows that we do not need EUV. We think that’s a cost-effective way of scaling DRAM further. Today we are in production with one X technology node bringing out one Y technology node of DRAM in the second half of this calendar year and we shared our roadmap that actually takes us beyond one Y to one Z to one Alpha and to one Beta nodes and those do not need EUV, which is the cost effective way of scaling DRAMs.

Mark Newman

I think we’re going to have all relearn the Greek alphabet as we keep shrinking. Any thoughts though on that – on when the limit is to shrink in DRAM?

Sanjay Mehrotra

One thing I have learned over my 38 years in this industry is that engineers never cease to amaze you and engineers always find solutions and do scale so I, at this point, we’re not sharing plans beyond one Beta technology on the DRAM side.

Mark Newman

Okay, fair enough.

Sanjay Mehrotra

But I can tell you that engineers, the innovators at Micron absolutely focus on looking at future opportunities.

Mark Newman

Fair enough. Looking at the DRAM supply demand side, what seems pretty clear to me from a – although we have just talked about is, that the supply growth from technology is not enough to keep up with demand. The industry is getting barely 15% per year growth from technology migrations. And as you pointed out in your slide, I mean demand is projected to grow, it has been recently growing 28% or low 20s.

So it seems to me that there is a need for some new wafer capacity and we are seeing wafer capacity being added by the Korean players, Samsung in particular. I’m just curious what, how you think about that? And how Micron thinks about the need to add more wafer capacity given that technology migration is not really enough to supply demand.

Sanjay Mehrotra

So let me be clear in my estimation that we shared, of approximately 20% supply growth in the industry in DRAM and approximately 40% in NAND, 2019 and onward. That actually includes in it an assumption of new wafer capacity getting added in the industry as well. There will certainly be some amount of wafer capacity that would have to be added by other players in the industry in order to achieve that level of bit growth for the DRAM side I’m speaking of, the new wafer capacity to be added on the DRAM side and a small percentage of new wafer capacity on the NAND side as well.

Without that new wafer capacity the industry won’t even get to the approximately 20% range. For Micron, we project our – in calendar year 2018, our supply bit growth to be approximately 20% in range with the industry as well. And our focus is on technology transitions, because technology transition gives us an opportunity to lower our cost, as well as provide us the bit growth that we are looking for in terms of managing our business, so approximately 20%. So we are not adding new wafer capacity, we are transitioning the existing production base to the new technology node to achieve our bit growth objective, as well as our cost reduction objective. And technology transition as I always say is the best ROI method of meeting your customers’ demand requirement.

Mark Newman

And your plan therefore is to expand this space to keep roughly, your wafer capacity roughly flat that’s…?

Sanjay Mehrotra

That is absolutely correct. As we transition to new technology nodes, they need more advanced equipment in order to process those wafers, that means in order to place that equipment you need new clean room space. So yes, we are adding clean room space for DRAM in our Hiroshima fab. We have said adding about 10% of – expanding that clean room space in the Hiroshima fab by about 10% to implement our one Y technology transition.

And in the future we’ll continue to assess our clean room space requirements. And we are doing the same on NAND, our fab in Singapore, we are expanding the clean room space in its Phase 1 by approximately 35% to implement the technology transitions, not adding wafer capacity but adding clean room space for technology transition purposes.

Mark Newman

Correct, correct. Moving over to technology, again on the DRAM side, there has been, over the last few years there’s been, from my perspective, more and more. There has always been technology issues and delays, but it feels to me that more delays on not Micron specific issue, I’m talking about the industry. So for example Samsung 1Y nanometer has been pushed out, they are having some problems with it.

I mean, do you think that we are going to see more and more of these technology nodes delay, because every time we see a technology node pushed out then supply growth surprises to the downside and therefore pricing is better. Do you think we are going to continue to see those? Or do you see any specific, like particular node that’s particularly harder where you expect more delays between over the next few nodes?

Sanjay Mehrotra

These technology transitions are really getting more and more challenging, more and more difficult. They take incredible amount of engineering expertise to be able to develop a leading edge node today. And lots of physics are the same everywhere. The challenges on the chemistry of processing these structures or the physics of it is all the same and yes, I mean those challenges are common and they are only increasing and you are right to point out that all of those, along with the facts that I had mentioned before are contributing to declining supply growth rates in the industry.

Mark Newman

Moving over to NAND Flash, you made an announcement at the Analyst Day that Micron is moving away from a floating gate to charge trap flash on your – and you call it Gen4 post 96 layers. Can you talk a little bit about why you are making that move now, not earlier, not later and I think there has been quite a bit debate in the industry about what’s better. There has been a lot of us have expected Micron eventually to go to charge trap flash, but is it just you at the limits for what you can do on floating gate, can you talk about that transition?

Sanjay Mehrotra

So floating gate has served Micron very well through three generations of technology, 32 layer, 64 layer, and later this year for us to be introducing the 96-layer technology. From the fourth generation onwards, that means higher than 96-layer technology is where we are switching to the, what we call replacement gate and charge trap technology. And our engineers have been working on that for quite some time and we made the determination that this is the right technology node to transition over to replacement gate, keeping in mind the future scalability options that replacement gate provides us, as well as we at Micron want to make sure that we are adjusting diversified end-market applications for mobile for data center SSDs, as well as for automotive applications and millions of other applications.

So, at this point while the replacement gate – floating gate technology have served us very well over the course of three generations, going forward the best class forward from the point of view from scaling, as well as meeting the diverse need of end-market applications, the right technology for us is the replacement gate technology.

Mark Newman

And I don’t think you have said, maybe you are not ready to say how many layers in Gen 4?

Sanjay Mehrotra

Yes, we have not disclosed that.

Mark Newman

Have you disclosed or can you say or can I assume correctly that it’s two-stack approach?

Sanjay Mehrotra

It will be stack to technology we have not discussed anything beyond that.

Mark Newman

Okay. I see.

Sanjay Mehrotra

Again, that’s something Micron has really mastered well in terms of its ability to stack layers, for example 64-layer technology is made with those stacks of 32-layers.

Mark Newman

Right. I would like to talk about the cadence of technology migration, recently NAND and D-RAM has been on a four to six quarter cadence in terms of how frequently the industry upgrades, do you think that continues to get longer? But do you think it stays roughly [indiscernible]?

Sanjay Mehrotra

As we move from floating gates to the replacement gate technology we expect our cadence to be fairly familiar, and to be able to introduce the technology in the marketplace in a competitive fashion.

Mark Newman

Because it does seem that since say 5 years 10 years ago that it is getting a little bit longer than the past and so 3D NAND I think what we’re seeing with Samsung now is they were, they thought it was going to be four to five quarters and it is looking like it is more like six quarters per migration in 3D NAND. So yes, it’s choice, I thought it. I does seem to me that it is getting a bit longer, maybe not micro specific, but …

Sanjay Mehrotra

I mean technology complexity is absolutely increasing for DAND, as well as for NAND and you know those kind of technology complexities do have an effect in terms of how much time it takes to move from one node to the next node. And technology complexity is increasing not only in terms of getting a technology developed, but as you deploy it into production ramping up yields on that technology bringing it to mature high yields all of those are involving greater and greater complexity, but again those challenges are same everywhere and…

Mark Newman

Okay so, separate topic, one I think a threat a lot of people worry about when they’re investing in the memory space is China, a threat from China, you know there is three major players in China, YMTC for NAND, Fujian Jinhua and Hefei Innotron on the DRAM, I am just curious how do you think about those, are they a realistic threat, do you worry about, do they keep you at night?

Sanjay Mehrotra

So, I will tell you, we never discount any competition, we absolutely keep an eye on the developments in China. In terms of staying up at night, I am totally focused on our execution at Micron, our opportunities that we described earlier are just immense and we want to make sure that we execute well to our priorities that’s what I stay focused on. And I would just like to point out some of the things that I mentioned earlier, I mean today when you look at industry players, they have decades, multiple decades of experience of developing technology and taking it into production with quality products.

So, this really requires lot of IP, which our customers value too. They always want to make sure that any new entrants do have bona fide IP, license to bona fide IP in order to bring products, so that they themselves don’t drive – gone any risk to those products, but aside from that the knowhow that is required to bring up these technologies and products into production and then all the know-how of designing it, making sure it has the right spares, right performance that meets the customer’s requirements in diverse set of end markets, as well as bringing it up in yields, bringing it up in volume production, achieving scale, these are all pretty high barriers to entry for new entrants. So, we keep an eye on it. However, it remains to be seen that’s how this works.

Mark Newman

Sure. I mean what are you doing, what is Micron doing to try to limit that kind of IP’s theft worries that we all have for many tech industries. For me particularly in memory, the Chinese memory makers are poaching. They have poached many people from Micron, also from Samsung and Hynix. And there are some lawsuits out there, and I’m not sure if you can talk about any of that just curious how do you try to stop that?

Sanjay Mehrotra

I think you are right. There is some litigation that is involved related to IP theft and obviously because it is pending litigation I’m not going to comment on that. We of course remain very vigilant about protecting our intellectual property. Protecting our talent. This is extremely important to us. It is a priority to us and we are extremely focused on it and we will take aggressive actions as needed to make sure that our intellectual property is protected and respected.

Mark Newman

Final question from me before I go to this list here. Data center demand, you know there is a huge amount of demand. We’ve talked about it, you had some slides on this hyperscale data center big data, I mean there is a lot of big trends, buzzwords here that’s driving, essentially at the end of the day, lots of demand for the DRAM and NAND flash. How predictable is that? I know you get a PO from a Google or Amazon or whoever, but how predictable is it that the rate of growth of data centers, because to me, it doesn’t seem as easy to predict that versus a smart phone sales.

Sanjay Mehrotra

You know data center growth is in very, very early innings. I think even if you look at today versus what people were thinking a year ago that growth is much faster. If you look at CapEx investments that are going into the data centers, a year ago versus what they were announced at the end of last year the CapEx is already higher versus previous projections.

We have shown that CapEx going up by about a factor of 3 to going to about $100 billion in data center infrastructure by 2021 timeframe and a lot of that, majority of that is actually going towards memory and storage. Because again, the emergence of these trends like artificial intelligence, which is really very, very early, I mean we’re barely seeing the tip of the ice berg there. And again, I feel that what our transformation that artificial intelligence can drive in the world today is probably underappreciated and what it means in terms of more memory and storage also is under appreciated.

So, these are the trends that are just evolving. Long term, these trends are secular, they are strong, there are several players across the globe, you know that are involved with hyperscale data centers, you know if one has a little change in their plans there are in aggregate if you look at all the others, I think the aggregate trend continues to point northward in terms of the demand projection. And we are very focused on making sure that we have the right set of products both for flash and for DRAM to address those opportunities, and I talked earlier about some of the things we’re doing there.

Mark Newman

Yes, great. So, lightning around a few questions from the audience here. So, first question is, you mentioned $6 billion structural improvement at the Analyst Day, also in your slides, does that mean you would expect $6 billion trough or $6 trough EPS?

Sanjay Mehrotra

So, if you look at our fiscal 2016 that was breakeven year in this regard, and if you look at 6 billion achieved to date and you look at into the three that we are projecting by 2021 timeframe, we have nearly for that total of $9 billion if you assumed that tax – certain tax rate there as well as assume the share count as similar to what we have today. It amounts to $6-plus kind of number. And I’m talking about there the total $9 billion that we are targeting to achieve by 2021 timeframe in terms of structural operating profit improvement since 2016 timeframe. So this is…

Mark Newman

That’s using the $9 billion.

Sanjay Mehrotra

That’s using the $9 billion exactly.

Mark Newman

Okay. Yes.

Sanjay Mehrotra

That’s using the $6 billion plus $3 billion, apply it, that’s operating profit, apply it – structural improvement in the operating profit and then apply it actually to it and apply today’s share count. And of course, what we’re talking about is share buy back program that I just discussed earlier as well. This will only, to the extent, we buy back in the future starting fiscal year 2019. It will only give further upside potential.

Mark Newman

Right. Second question here is Applied Materials. The CEO was here yesterday, and he said apparently, 1% autonomous vehicle penetration requires 25 times more memory than we have in total capacity today. What are your thoughts?

Sanjay Mehrotra

I mean, autonomous vehicles are certainly when we look at L3 plus vehicles, in 2025 timeframe, we’re projecting something like 26 million L3 plus cars. And L3 plus car is, including L5 cars. By that time, L5 is a fully autonomous car. L3 plus is increasingly features toward driver – driverless cars. And these are the cars that absolutely – I mean, autonomous vehicles will be like data center on wheels as we said.

If you look at some of the projections for autonomous vehicles, L5 vehicles, they show DRAM content in one car of 40 gigabytes plus. In fact, some estimations are even higher for L5 car. And for NAND, a terabyte of NAND in one car. So I mean, these are the kind of numbers that you typically see attached to a server that actually will be in your cars, so it will be like a data center on wheels.

So there is no question that autonomous vehicles will be a driver of growth. In my projection that I showed you for 2021 timeframe, both for DRAM and NAND, those are accounted for in those projections.

Mark Newman

Okay, that’s great. Two questions here. What is the right through cycle operating margin for Micron’s business? And second question, somewhat related to that is why do you think your competitors have becomes so rational?

Sanjay Mehrotra

So in terms of operating profits and margins, we chose to discuss it the way we did, with the tangible numbers, I guess, $6 billion of operating profit, structural improvement, over 20 – 2016 through 20198 timeframe and what we’re projecting in the future as well. And of course, we are extremely focused on continuing to drive our cost competitiveness, and as I said before, our high-value solutions mix. And that gives us an opportunity to be able to basically do the best in terms of margin profile in any kind of industry environment.

And in terms of overall industry environment, what I can tell you is what we are focused on at Micron. We are absolutely focused on driving profitable at Micron. As I said, totally focused on making sure any of the investments on the OpEx side or on the CapEx side have ROI considerations, total ROI considerations taken into account. This is what we are focused on and this is now as well as continuing to understand what are customer needs in these growing market environment.

Mark Newman

Next question here is on 3D XPoint and other types of new memories. So specifically, where do you see 3D XPoint fitting into your existing portfolios? Which applications do you think it best addresses? And then, I guess, my follow-on is also, how do you think about new types of technologies in memory? And how do you think about how that may impact your current business, DRAM and NAND?

Sanjay Mehrotra

So on 3D XPoint, we certainly see opportunities in the data center space there, in memory semantic applications, also excluding a portion of these under storage semantic front. Our new technology, we have a been very much focused on addressing the cost,. as well as the performance aspect of this technology.

We have shared with you before that at the end of last is when we staffed our teams for product development, we will have a product launching on 3D XPoint in 2019, latter part of 2019. And at that time, we will begin to engage in market enablement, as well as customer enablement opportunities based on those products and their acceptance of those products.

So we look at this as an exciting opportunity for the future. And we’re very proud that Micron has this capability of 3D XPoint technology. It will absolutely, I believe will be going into a variety of applications. Certainly, when you look at the trends of machine learning in memory computing, big data, these are the kind of applications, which will be well suited for 3D XPoint because of the cost of ownership, the cost per I/O that 3D XPoint enables and having the benefit of persistent memory as well.

Mark Newman

Why has it been delayed so much? So it seems like it’s two or three years since you started talking – since Micron and Intel has started talking about it, and it doesn’t seem to have had any significant impact in commercial yet. What is the reason for that?

Sanjay Mehrotra

Anytime you have a new breakthrough technology that is as significant and has meaningful opportunity as this have, it does take it a while for that technology, breakthrough technology to shape into all the right set of specifications that ultimately can meet the end market requirement, which end market requirements also keep moving forward, right? I mean, keep on becoming more and more demanding.

So this is not unusual for breakthrough new technologies to take long time to come to successful solution in terms of commercial success. And as I said earlier, new technology like this, we were at Micron very much also focused on making sure that it becomes cost objective, as well as achieve the right level of performance. And that – I mean, that’s just the difficulty of new technologies.

Mark Newman

But are you seeing – is Micron seeing commercial traction or like you’re seeing customers engaging – do you see potential revenue in the near-term?

Sanjay Mehrotra

Certainly, customers are very excited about the potential of this technology, given some of the attributes that we talked about in terms of persistent memory nature, as well as aspects of cost and performance and endurance. And as we said, for us, 2019 is when we start having first products and it will take us customer qualifications before we can start making it into a commercial revenue opportunity. And in the meantime, we will have opportunity to sell some of our production output to our partner depending upon their demand requirements.

Mark Newman

Question here, how much of the future big growth is in – is linked to assumptions about AI and particularly AI training workloads? Do you see any reason you could see the timeline pushed out or pulled in for AI?. Let’s say, the other question is, how much of an impact to all of your supply demand assumption could – is AI and how could that change?

Sanjay Mehrotra

So certainly, AI is impacting some of the markets already today. For example, when you look at facial recognition in your smartphone that’s an example of AI. Facial recognition requires more DRAM memory in the phone. And as more of these features go even into the mid-range phone, they will be driving more demand on the phone side. Similarly, some of AI workloads today are being tackled in cloud applications as well.

I did share last week at the Investor Day that when we look at AI workload servers, particularly AI training workload servers, it’s barely a sliver in 2017 timeframe. Even by 2021 timeframe, only about 10% of the servers will have AI workload capability, particularly AI training kind of workloads. And it’s more like 2025 by which time it’s like half the servers, they’ll be having those capabilities.

So there are certain assumptions. And keep in mind, those servers that are tackling AI workload tend to require more DRAM. I showed at the Investor Day about 6x the DRAM and twice the SSDs. So we – in the bit growth, the demand bit growth that I showed in the slide, the server portion of that demand bit growth in 2021 does include some element. But as I mentioned, we expect about 10% of the servers to be AI training kind of workloads in the timeframe.

So NOR, this is again, as I keep pointing out, it’s very early innings. I mean, there’s really a massive opportunity ahead and it will evolve over the course of next several years. I think, that’s why we say these trends are secular in nature.

Mark Newman

Yes, I think, we estimated that AI last year was about 1% of DRAM demand that’s AI data center, this year about 2% for what it’s worth. That’s how…

Sanjay Mehrotra

All right. So these are again the trends that are barely taking off, but these are big trends overall over for the longer-term.

Mark Newman

Very, very last question, which is a question that we’re asking to every CEO at this conference, which is, as you look out five years into the future, what is the single biggest disruptive force that potentially threatens your business and why?

Sanjay Mehrotra

So I will not answer it with what threatens our business in terms of that. I mean, of course, any macroeconomic major change or like what had happened in the 2008 timeframe, no industry will be able to escape that. So we’re not going to get into those things. I do not know. We – I can’t project those. But I will definitely talk about the disruptive trends that will absolutely change our industry again is artificial intelligence.

I do believe that in this disruptive trend, which will drive tremendous growth for memory and storage in many ways is being underappreciated in terms of the force it will become for our industry, but also for certainly, all industries and for all humans on this planet.

Mark Newman

Impact positively for these memory companies?

Sanjay Mehrotra

Absolutely.

Q - Mark Newman

Okay, great. Thank you very much, Sanjay. Thank you.

A - Sanjay Mehrotra

All right. Thank you.

Copyright policy: All transcripts on this site are the copyright of Seeking Alpha. However, we view them as an important resource for bloggers and journalists, and are excited to contribute to the democratization of financial information on the Internet. (Until now investors have had to pay thousands of dollars in subscription fees for transcripts.) So our reproduction policy is as follows: You may quote up to 400 words of any transcript on the condition that you attribute the transcript to Seeking Alpha and either link to the original transcript or to www.SeekingAlpha.com. All other use is prohibited.

THE INFORMATION CONTAINED HERE IS A TEXTUAL REPRESENTATION OF THE APPLICABLE COMPANY'S CONFERENCE CALL, CONFERENCE PRESENTATION OR OTHER AUDIO PRESENTATION, AND WHILE EFFORTS ARE MADE TO PROVIDE AN ACCURATE TRANSCRIPTION, THERE MAY BE MATERIAL ERRORS, OMISSIONS, OR INACCURACIES IN THE REPORTING OF THE SUBSTANCE OF THE AUDIO PRESENTATIONS. IN NO WAY DOES SEEKING ALPHA ASSUME ANY RESPONSIBILITY FOR ANY INVESTMENT OR OTHER DECISIONS MADE BASED UPON THE INFORMATION PROVIDED ON THIS WEB SITE OR IN ANY TRANSCRIPT. USERS ARE ADVISED TO REVIEW THE APPLICABLE COMPANY'S AUDIO PRESENTATION ITSELF AND THE APPLICABLE COMPANY'S SEC FILINGS BEFORE MAKING ANY INVESTMENT OR OTHER DECISIONS.

If you have any additional questions about our online transcripts, please contact us at: transcripts@seekingalpha.com. Thank you!