Western Digital Corporation (WDC) Earnings Call Transcript & Summary

Good day, everyone. My name is Mehdi Hosseini, Senior Analyst at Susquehanna International Group covering technology hardware. It is a great pleasure to have a fireside chat with the team from Western Dig. We have Wissam Jabre, the Chief Financial Officer, who recently joined the company; and Sivaram, President, Technology and Strategy and the IR team. Wissam comes to the WD with more than 20 years of experience in semiconductor and operation and was most recently CFO of Biala Semiconductor. Wissam will make a short statement, then we will open it up to fireside chat conversation. I have prepared a number of questions and topics to discuss with the WD team. And if there is any question from the audience, feel free to e-mail me your questions, and I will ask them towards the end. My e-mail address is [email protected]. To that end, again, I want to welcome to WD team. I want to thank them for taking the time to give us an overview. And to that end, I want to welcome Wissam to the investment community under the WD platform. So Wissam you have the mic.

Thank you, Mehdi. It's great to be here. I'm very excited to be joining Western Digital. Before I start, I have the pleasure to read the safe harbor statement. We will be making forward-looking statements. I ask you to refer to our SEC filings for the risks associated with these statements. We will also be making references to non-GAAP financials and a reconciliation to our GAAP and non-GAAP results can be found on our website. For those on the call that have not seen the press release we issued last night, let me quickly summarize it. On February 9, we and our joint venture partners issued press releases noting that a contamination of certain materials used in flash manufacturing processes has affected production operations. At that time, we noted that there could be an impact to Western Digital's availability of flash bits of at least 6.5 exabytes. Last night, we are pleased to report that in late February, the production at both our Yokkaichi and Kitakami has returned to normal operations. We now expect flash availability will be reduced by approximately 7 exabytes, very close to our original estimate. This reduction of bids is expected to primarily affect fiscal Q3 and fiscal Q4 as the facilities ramp back to full production output. We now expect revenue to be in the range of $4.2 billion to $4.4 billion, non-GAAP gross margin to be in the range of 30% to 32%, and earnings per share of $1.30 to $1.60. With that, let me turn the mic back to Mehdi for the first question.

Thanks, Wissam. And the one question advocate that I want to ask you is about your decision to join WD team, pretty much like shortages of chips. There is also a shortage of CFO in the Bay Area. And in that context, I want to hear if you could share with us a couple of things that attracted to this opportunity at WD?

Thank you. Well, look, the Western Digital is at the heart of the digital economy. When you think of storage and memory, it's pretty much in every electronic [indiscernible]. The company has great leadership, top talent engineering as well as a long and deep history of innovation. Very broad-based portfolio with a great set of also very broad-based customer base. And so we have all the ingredients to be very successful and given where we are, we also have all these ingredients to continue to drive long-term shareholder value.

Great. Just a couple of questions for you. You highlighted the press release from last night about $200 million or so revenue shortfall, 26 of EPS. If I just look at the exabyte assuming that half of those exabytes were impacted in the March quarter and just work it back to revenue and EPS, it seems to me the underlying assumption for blended NAND ASPs but a flattish March versus December. Am I in the ballpark with those assumptions?

So the announcement that we made yesterday in the updated outlook is primarily driven by the flash situation, the HDD to remains, as we talked about it earlier. When we look at the updated outlook, which is, as you noted, $200 million lower at the midpoint from the $4.2 billion to $4.4 billion relative to the previous outlook of $4.55 billion -- sorry, the midpoint was $4.55 billion. It's primarily driven by the lower exabytes -- the bids shipment on the NAND side. And so it's partly offset by some improvement on the pricing side. The better ASPs are primarily driven by the transactional market. And as we start our pricing negotiations with our OEMs for the -- in the next few weeks, we'll see how things will shape up.

Okay. And then margins not changing. Is that a reflection of your relationship with the JVs buying NAND wafers on a cost-plus basis?

So the one thing I should note on the non-GAAP margins is that they assume the exclusion of the -- any charges related to the contamination. And so this is sort of important to note. And second, as I said, we basically are -- we're starting to see in the transactional market better ASPs. And so that's also reflected in our margins being more or less where you were at the beginning of the quarter in terms of outlook.

And with transactional business, how did it trend before and after Chinese New Year holidays?

I don't know, if we saw the differences in terms of before and after the New Year -- the Chinese New Year. But what I can see is over the last week or so, we saw a bit of firm up in the transactional market.

Sure. I think what is interesting with the NAND industry is the fact that we are going through an indirect consolidation with SK Hynix acquiring in terms of Dalian facility. So that, combined with the 18 exabytes that taken off the table including Kioxia's contribution for attribution, that's about 10% of the NAND bits that are now consolidated or there is more direct discretion on the pricing on those exabytes. And in that context, shouldn't that imply that pricing dynamics are going to favor the suppliers and that -- isn't that implied in the transactional segment of the market you were referring to?

So look, we're -- as I said earlier, we're starting our negotiations on pricing with our OEM customers. It's too early to make a prediction of how the prices will trend. I did mention earlier that the transactional market as prices have firmed up a little bit. So depending on how the OEM negotiations happen over the next few weeks, there is potential for ASPs to improve quarter-on-quarter. The good news is that the demand for flash remains solid. We do have a broad portfolio that serves really different end markets. And so we're well positioned to continue to benefit from that. And of course, the consolidation is good for the industry. But we also -- with our joint venture, we do have quite a bit of size and scale as well.

Sure. Great. So now I'm going to switch to Siva with some of the more detailed questions, starting with NAND and then moving on to HTD? Siva, again, thank you so much for joining us. Can you please update us on the bid cost decline? And I'm assuming that we're referencing -- well, we're trending with 15% annual decline over the next few years. What I want to learn from you is, is that -- is it an accurate expectation? And then what are the key drivers, especially as you migrate to BiCS6 and QLC becomes a larger mix?

Yes -- thank you, Mehdi. Thank you for the opportunity. As you know, we have guided this mid-teens for a long time. So we continue to see a good road map for a long-term trend of the mid-teens, 15%, as you put it, in bid cost reduction over time. Having said that, the last couple of quarters have been better than that as you guys have realized throughout fiscal '22 and probably towards the end of fiscal '21, we have done better than that 15% over time. And that's primarily because BiCS5 is an extraordinarily capital-efficient node. BiCS4 was a very high-yielding high volume. We got almost 85% of our capacity to BiCS4. Then when all of that starts to ship to BiCS5, BiCS5 being as cost efficient as it is, that consistently we've been getting good cost reduction. We also reported that towards the end of fiscal Q2 that we cost over. There are more BiCS5 shipping out of the fab than BiCS4. And that conversion as it continues to go up, we'll continue to see cost reduction. So that's on the pure technology side, and that BiCS6 cost, also we have this 18 months between site, between nodes that we continue to go through, which is coming along very well. Then the superimpose on that is QLC. QLC is still a very small percentage of bits today. Even client is where we'll start to see more QLC adoption. So BiCS5 will start to get that ball rolling. BiCS6 is where we start to see a lot more adoption of QLC in our anchor business in the client business that will drive additional cost reduction. So overall, that 15% cost reduction is a very, very healthy assumption to make.

Just a quick follow-up, what would catalyze adoption of QLC for enterprise application?

As you know, QLC initially is behind this in operation, meaning most of the interaction with the host occurs to an SLC. As the IO gets better and the densities get larger, QLC will become a part of the enterprise going forward. And I expect that when the performance IO performance and the retain performance of QLC gets to be solid, you will end up seeing the bigger driver will be larger density NAND in enterprise.

With the migration to PCI 5.0 make a difference.

One of the things of normal migration will happen, so the IO getting faster else, right, because of what he just said because of the SLC buffer, it helps yes. But if that migration probably is more in the second half of BiCS6 is when enterprise QLC will become hard. So think of it as once we ramp into BiCS6 and all the consumer client mobile all get done and then get to enterprise. So that's sort of the time span for enterprise in USA.

Sure. I just want to -- I want to focus on enterprise SSD a little bit here. Over the past, let's say, 6 years, we have gone through, I would characterize it as share loss and then regaining some of that share in enterprise SSD. Can you give us an update where we are in WD's market share enterprise SSD and what are the longer-term targets?

So clearly, you are correct, Mehdi that several years ago, we had some missteps in our enterprise SSD execution. We are out of that completely. We are an emerging strong player in enterprises SSD where there are only 2 strong players. Now we are becoming the third stronger player in there. Our focus is entirely on NVMe. The whole the SATA SaaS, which still is a big part of it that we don't play where we need focus on the biggest growth segment, which is NVMe SSD products. As we have probably been talking over the last several quarters, we are now qualified in 3 major cloud titans. And we are also qualified in a couple of large OEMs. This gives us the base to which we were out from. Cloud, in general, is a large percentage of our focus as you have seen in our distribution of where we put our bits both on HTD in SSD side. Overall, cloud remains a big part of our portfolio. So given the fact that initial execution issues are over. We are qualified with the titans. We are continuing to ramp with all of them. Now superimpose on top of it, our current controller issues and availability and supply chain things. Those are short-term hiccups that we need to continue to work out. But this trajectory is very good.

Sure. And to that end, if you get back on track with gaining mine share and market share with the cloud and OEMs, would that actually reduce the mix of consumer retail segment?

Probably not. Consumer is going to be a strong barter portfolio always. Consumer is our unique -- where we are very strong and gives us -- just like Wissam was talking about, gives us the first premium into every market as it grows, everything and faster access and turnaround in introducing technology and seeing the ASPs change. Consumer will always continue to be a good part of our portfolio. Client is the foundation of this. Client is the big middle that service. But then you vary on top of it, how much we go into mobile and how much we go into enterprise. Enterprise is our current largest growth segment, something that is equally matched by gaming. Gaming is emerging to be again a double-digit player for us in BiCS share. So those are -- I can't anymore say 4 legs of the stool, because right now we have 5. So I have to go out of this analogy. The fifth leg is gaming that's growing very fast.

So if you get back on track executing and gaining share, would that require you to work with JVs for higher allocation?

JV is, as you know, fixed -- we have a 60-40 output of JV. However, what we do have -- our ability to grow is in what kind of diet, whether I have TLC versus QLC, whether I end up doing the larger debt that's how we gain share in terms of bits coming out of the JV.

Got it. Okay. Now switching to hard disk drive and where are we with 18 terabyte, what's -- if you could give us a mix of nearline of mass capacity broken by 18 terabyte. When would it be like a year from now?

So as you know, 18 terabyte is sort of I even say it's moving our way past the sweet spot. 18 terabits for the entire fiscal '22, it will be the majority of our volume treatment. It is qualified in all the cloud titans, all the OEMs, in retail, in channel, everywhere 18 is fully qualified in shipping and volume. Right? So 18 is no longer -- 18 of our first generation and energy-assisted drives. We have shipped a large volume out of it we are starting to ramp 20 and we are starting to ramp SMR. Those are the 2 new messages you'll get out of this, 20 is starting to ramp. SMR in nearline is starting to ramp the significant percentage.

Siva, I think it would be great if you could just give us a very brief overview of what are the key differences between SMR and CMR, HTD?

So CMR, as you know, conventional. You have a right track and you -- and on top of it you go re. Right Track and retrack are slightly different. The right track is slightly wider than the retrack. And that gave us the idea to say, if that's the case, why can't a shine just like a roofing tile. On top of the wider track, do another track that overlap, I can still read that provides the idea of Shingled Magnetic Recording, which is normal PMR with this overlap of the right track 1 over the other. That gives us an additional density uplift. We are now qualifying this SMR in near line in a couple of cloud titans. We are starting to ship volume in the second half of the year, we'll be in double-digit volumes with SMR 20 terabits and about this. This is a big part of it. You add that with our client and the near line, SMR is becoming a very, very important part of our shipping portfolio.

In other words, SMR gives you a higher density even though the right is slower?

The right has to be controlled by the host. So you streamline it. So what you do is you take the bits match it, shape it so that you're right sequentially.

Right. And just to be clear, do competitor or competitors also offer both SMR and CMR or is that just WD?

SMR is an open standard developed by us, and we have put it in a open standard right now. In many devices, everybody is shipping. But in near line, you get mass capacity enterprise drives, we hold a substantial lead in the development of the technology, implementation of the product, shipping to the customer, getting the customers qualified. There we have a substantial lead.

Got you. And -- are we going to just see increased adoption of SMR as we migrate towards totally exabyte and we're going to make a small incremental improvement in density before HAMR is available?

Yes, it's a very, very good question. HAMR is a technology we are all eagerly awaiting. We will make sure that we are on par and when we need to, we will introduce it at the right time. But ahead of that, SMR, OptiNAND, these kinds of technologies allow us to give us a road map from 20 to 22 to 24 to 30. I think we will need these technologies to be fully working in the short term till HAMR becomes available, maybe by the 30 terabyte generation.

Yes. Okay. Now I'm going to switch gear and read some of the questions that I've received. And a couple of questions as it relates to contamination. You talked about a onetime cost in the March quarter. How should we think about the cost beyond the March quarter as you ramp the fab, does the cost per bit go up in the short term and then it comes down or how are the dynamics?

So I mean, look, the -- what we're looking at is the -- what I talked about for the March quarter is the charges related to the contamination event itself. So there'll be some wafer scrap, some other types of costs associated with having under utilization. Those were -- are exported from the non-GAAP figures that we're saying the outlook is 30% to 32%. With respect to the ongoing costs, we don't expect clearly big fluctuations relative to what we've seen -- but by the time the Fab gets up to the normal production output.

Sure. And this question for the entire team. Can you give us an update how Kioxia and JVs Fabs or broken? Like you talked about a 4 exabyte impact from contamination. I divide that by 0.4%. I get to almost 18 terabyte as a combined aggregate impact. Is that going to remain that distribution? Is it going to remain the same looking forward?

Siva, how do you want to put this.

Yes. So on the first starter, maybe that's true. But there's always differences as which node they are running in volume, how many bits per wafer they get, which [indiscernible] they are running, those kinds of things are going to be different. So to the first starter, we have 40%, they are 60% and all of us got damaged the same proportion. It's the same Fab that is there. So when the wafer is in the Fab, nobody knows who's wafer it is. Even it comes out, do we decide that it's my wafer, right.

Okay. And then one other question that I received has more to do with the drama associated with Toshiba. Now that they went through another CEO and the company is going to break into 2 parts. Is there any update on the ownership of Kioxia?

Maybe Kioxia has been a trusted partner with us. We have worked very well together. Our relationship between them -- between us and is very, very good. We are working very well together in the JV. So their parent ownership is not -- most of it is not public, not just exactly where, what their ownerships are. And I know Toshiba has a big stake and Toshiba has made some public statements on they want -- what they want to do Kioxia. All we can do is wish Kioxia success in all of this, and we just want to be cooperative with them.

Okay. And then just to -- we have a few more minutes one -- let me just read the question. One more question, and then I'll turn the floor to you. As we look forward, it's a 2-year anniversary of the new team being on board and the first thing that happened, breaking up the company into 2 organization like making Flash and HTD separate. And now, hopefully, the final touches on the management team is done. Now should we expect -- the team is in place and now is like, okay, we're going to get back on track with execution. And I know I'm trying to ask you a very sensitive question. I was hoping that you read my mind and give me some bullet points?

I will start then Wissam who is the new guy can always -- look, Dave Goeckeler came in March of 2020. And the week after he came, the pandemic hit. You can see his content everywhere. He has made the focus on the 2 businesses of paramount importance. So we have a well-focused HTD business and a well-focused Flash business. He has rearranged the executive team. You can see lessons here. We just got [indiscernible], the VP of operations last week. So he has got a team of extremely well-seasoned professionals in the mix. So it's up to us to come back and make sure we realize the synergies of this -- it's 2 organizations that will come together. The 2 organizations have to independently worked very well. Only then we can have synergies that then become better in those. So we are in that place -- the JV is working very well despite these kind of one-off events, they are doing very well. HTD, you can see our continuous cost reduction curves, 18 is fully qualified and running and 20 is coming up and ramping, SMR is ramping. BiCS5 is ramping very well. BiCS6 is coming up [indiscernible]. You can see this train that is -- that's now running and is a well [indiscernible].

I think the next step is to take Wissam on the road and actually have in-person meetings in New York and Boston.

Yes. Hope for that.

Okay. Well, thank you so much for your time. Thanks for entertaining all my questions. Is there any closing remarks you guys want to make before we wrap it up?

Well, I mean, I'd say, well, thank you so much for the interest in Western Digital. And we're very focused on execution and continuing to deliver and drive much more value going forward.

Okay. Great. Thank you, and thanks to the IR team. I wish you a great rest of the day and stay healthy.

Thanks, Mehdi.

Thank you, Mehdi.

Thank you, Mehdi.