Micron Post FQ4 2018 Earnings Update – September 24, 2018

Post Earnings Update – Doing the math and recognizing that MU is now a real taxpayer

“What’s the market expecting from a company that trades at 4x earnings, perfection?” – me

While Micron had stellar FQ4 results, there were four primary things that spooked the AH trading and led the stock down ~3% on Friday

1. The stock turned on a dime in the AH (from up 4%+ to down), based on FQ1 guidance of $2.95 that was 16.5% below the blowout $3.53 delivered in FQ4. Given how sensitive the market is to changing directions in Micron’s results, the marked outperformance in FQ4, ironically, in all likelihood only served to further exacerbate concerns about the magnitude of the decline in FQ1 guidance. Midpoint of guidance ($2.95* per share +/- $0.07) also missed IBES consensus of $3.06 by 11 cents. I will discuss this in more detail following this section, but over 70% of the decline in FQ1 2019 EPS (41 cents of the 58 cent decline) is driven by Micron starting to have to pay a 12% tax rate. In FQ4 Micron only paid a tax rate of 0.5%!

2. Gross margins are expected to decline in FQ1 to a range of 57-60% vs. 61.4% in Q4 (approximately 300bps at the mid-point). While gross margins in FQ1 of 2018 were only 55.4% and a mid-point of 58.5% for FQ1 2019 gross margins is a meaningful jump from that level, those concerned about how any negative reversal in the trends in profits/margins were “triggered” by the guidance.

3. Micron affirmed that some customers are reducing inventories of memory chips (not said, but implied, some customers have elected to reduce inventories into expectations of a weakening DRAM pricing environment – in NAND this isn’t an issue because this destocking effect has already taken place). This inventory level repositioning is likely to take a couple quarters according to Micron. {1) I suspect Apple is one of the likely culprits and 2) Micron, of note, is not seeing its inventory gap out and in fact inventory turnover has increased to over 8x from over 6x in FY 2018 vs. FY 2017 when using year FYE inventory figures}.

4. Micron affirmed that the problem that Intel is having satisfying CPU demand for PCs is impacting demand for memory semis from PC customers. As alluded to in my update from 9/14, this is in all likelihood only a ~1% impact on total demand. But I estimate that between that ~1% Intel impact and the 1% impact from tariffs on gross margins, one accounts for 10 of the 11 cents difference between Micron’s guidance and Q1 consensus estimates.

Is this the “New Micron” or the old? Thoughts on post earnings price action and where we go from here.

The above chart shows a progression of adjustments made to FQ4 2018 earnings and compares a “normalized” FQ4 2018 to a likely FQ1 2019. As highlighted in the initial bullets, Micron paid virtually no taxes (a mere $20mm reflecting a 0.5% tax rate in FQ4). The days of Micron paying no taxes because of all of its past accumulated NOLs and favorable tax treatment due to heavy international presence are over with the advent of this new fiscal year. If Micron had had to pay taxes at the 12% rate they have guided to for fiscal 2019, EPS would have come in at $3.12 in FQ4. Additionally, if Micron had had to contend with tariffs in FQ4 and this Intel lack of sufficient production issue (I estimate a 1.5% impact on revenues to be conservative), that’s another 10 cents per share between those two items. All those adjustments which are aimed at putting FQ4 and FQ1 on even footing suggest that the decline in FQ1 EPS based on company guidance is likely to be negligible (less than 3%). And if one assumes that Micron beats their guidance by a mere 6% (MU has beat its quarterly guidance for eight straight quarters with a median beat of 12%), then Micron is likely to exceed the earnings it would have put up in Q4 if it had had to pay the same tax rate and deal with the same tariff and Intel issues it be dealing with in Q1. No doubt one of the benefits that Micron is enjoying in FQ1 2019 is its newly initiated share repurchase program. I calculate that buyback activity will improve Q1 earnings by at least $0.04 per share assuming at least $1.5 billion of share repurchase at a stock price of $45. As I believe it likely that, based on multiple comments from the call and follow-up comments on CNBC made by Micron’s CEO on 9/21, that Micron is likely to actually implement at least $2 billion of share repurchase in FQ1 driving EPS $0.05 higher than it would have been otherwise. As EPS is derived by using average share count during the quarter and I’m assuming a smooth repurchase throughout the quarter, the benefit of FQ1 share repurchase activity on EPS will double in the following quarter (going to 8 and 11 cents depending on whether $1.5bn or $2bn is repurchased).

Putting the math aside, the reason for the sell-off in Micron shares is the company was not able to put to bed concerns that we are not entering another typical DRAM ASP bear market. While Micron management made multiple assertions about the health of the industry, proactive steps they and others have taken and the long-term secular demand drivers (I will provide multiple examples at the end of this update), they didn’t say anything sufficiently concrete to send the bears into hibernation. Customer re-orientations on inventory levels (running with lesser DRAM inventory) are affirmation that customers expect further ASP declines (which will presumably make negotiations on every pricing call with customers that much more challenging).

There are undoubtedly those on the buy and sell-side that are pressing here. One of the biggest challenges to being long Micron’s stock is the company’s long history of volatile operating performance. Not only has that history led to ingrained institutional investor skepticism about the persistency of any rally, that bias has arguably been compounded because the institutional skeptics still can’t believe the impressive performance that Micron has put up the last six quarters. Skeptics effectively believe that the blowout profits that Micron has been posting are akin to a 1,000-year flood. Because Micron did not put to bed concerns about how long or severe pressure on ASPs will last, there are those skeptics and bears that are fear mongering, communicating /assuming the worst and conveying that profits are likely to drop like a stone for the next 6+ quarters.

On the positive side, Micron is doing just about everything right (as it seems is the industry) and it could very easily be argued that the industry is taking the steps necessary to ensure this period of weakness isn’t nearly as severe or long lasting as any of downturns in the last decade. To wit:

– Micron is going to continue its policy in 2019 (as it did in 2018) of adding no incremental wafer capacity in DRAM or NAND (other than wafer capacity needed to balance out the loss of throughput from node transitions).

– Micron continues to expect to deliver its bit growth in both DRAM and NAND from node migrations. Not from incremental wafer throughput.

– Micron continues to accelerate Capex spend to close the gap in DRAM (so it can accelerate its migration to the 1Y node which should lead to another major cost per bit reduction).

– While Micron is already the low cost producer in NAND, it plans to continue on its path to start delivering the next node (96 layer 3D NAND) in calendar 2018 with significant cost benefits in fiscal 2019 as the Company transitions to that node.

– On the NAND front, it has to be recognized how impressive Micron’s recent performance has been. In each of the last two quarters, Micron had ASP declines in NAND but gross margins actually improved (in FQ3 NAND ASPs declined YoY in FQ4 ASPs declined QoQ). This was especially notable in FQ4 where ASPs dropped in the mid-teens QOQ, but due to having a better mix of high value solutions vs. trade NAND, selling 30% more bits than in FQ3 and having achieved material cost reductions per bit the Company drove higher profitability.

– In NAND, Micron is showing noteworthy discipline. The Company is cutting back on NAND capex in 2019 with the intent to spend less than it did in 2018 as Micron is already the leader in terms of having the lowest cost of production 3D NAND. It is also, in all likelihood, the case that Micron had more planar NAND capacity to convert to 3D in 2018 than peers and those conversions are largely behind the company.

– Micron now has over 1/3 of its revenues coming from enterprise customers. Cloud SSD quarterly revenues doubled from last year. Micron now has #1 market share for enterprise SATA.

– About 50% of revenues are now comprised of high growth enterprise customers and industrial customers that require “long life cycle products.” A prime example of a long life cycle products is how Micron sells its memory into the auto industry (where Micron has #1 share). When the 2019 XYZ model car comes out, it is going to use the same memory products throughout that model year. These two areas (enterprise and industrial) alone are bringing much more solidity to Micron’s business.

– Micron continues to migrate its NAND sales away from basic, lower margin trade NAND to higher value solutions (now over 60% of sales vs. 40% a year ago).

– Micron established a 10b5-1 on September 1 and instituted a “programmatic” $1.5bn share repurchase program for the quarter. Additionally, Micron indicated that they have the ability (and it sounded like the strong inclination) to supplement that $1.5bn with additional opportunistic repurchase activity during the quarter if the stock action provides a special opportunity (it does!). At FQ1 guided levels of profitability and a $45 repurchase price, every $2bn of share repurchase should lead to about ~$0.11 per quarter of EPS accretion once fully baked into the fully diluted average share count for the company. So if Micron engaged in $8bn of share repurchase in FY 2019 and profits prove relatively consistent with FQ1 guidance, the exit run rate benefit from that amount of share repurchase would be about $1.80 in EPS (and Micron would shrink its float by about 13%).

– Micron management affirmatively indicated that they thought their stock was “very undervalued” – not something that one generally hears companies say that often anymore in a public setting. At less than 4x consensus 2019 EPS forecasts, only a momentum trader could argue otherwise.

– Micron indicated that 2019 demand bit growth for DRAM is likely to be 20% again and NAND bit demand growth is likely to be 35-40% – down from 40%+ expected demand growth in 2018. With such continued high demand and no indications of any meaningful demand decline from end markets (only impacts from different inventory positioning and this Intel CPU related issue), it should not take long before robust demand overtakes expected slowing of1H 2019 supply growth.

– Micron highlighted how NAND supply growth is likely to be dramatically less in 2019 vs. 2018 because of how the node migration is likely to be much less impactful. Micron specifically expected a significant decline in NAND supply growth in the 1st half of the calendar year. There was a material amount of planar NAND to 64L conversions that occurred in 2018 (skipping the 32 layer node altogether). With most of the planar to 3D NAND conversion behind the industry, in 2019 the conversions will principally be from 64L to 96L. While at face value that sounds like a 50% increase in capacity, because it is estimated to take as much as 1/3 longer to make a 96L 3D NAND chip vs a 64 layer 3D NAND chip, the pick-up in capacity may be as little as 13% from that node migration – dramatically less than the expected 35+% demand growth rate. This implies two things: 1) that there will have to be additional wafer capacity added to meet high expected demand (or prices skyrocket) and 2) that the currently oversupplied state of the market is not likely to persist for any appreciable period.

– Micron highlighted how capital intensity at each node migration continues to increase, which should continue to make the industry more rational and incrementally drive more pricing power over time.

– Micron conveyed that all of the above dynamics are likely to lead to less operating volatility for the industry (but Micron’s CEO fully conceded that didn’t preclude “one or two quarters, here or there, there can certainly be ebb and flow in terms of demand or supply in the industry. But the long-term trend is positive”).

Big picture, the fundamental narrative, I would submit remains this. Micron still appears well on track to deliver $10+ per share of earnings in FY 2019 (with at least $3 of that coming in FQ1). The Company is going to be plowing its free cash flow into buying back shares. Given how these node migrations are rolling out, it is very likely that supply growth is going to slow markedly in both DRAM and NAND in the 1H of 2019. With that slowing of supply growth, absent a recession or global slowdown in 2019 (which most think unlikely), ASPs are likely to start firming in the 1H of 2019. I think it very possible, if not likely, that 2019 will prove a trough earnings year where trough earnings are likely to come in FQ2 & FQ3. It is highly typical based on past Micron trading behavior that when the market becomes convinced the bottom is in, the stock starts responding very positively.

But given that a firming in pricing may only start happening four to seven months from now, what does that mean for now? Well, if the market thought that trough earnings for Micron was going to be $10+ per share and the multiple on Micron’s stock at the trough of the cycle is at least 15-20x earnings, well then Micron’s stock should already be trading at over $150 per share rather than in the mid $40s where it closed on Friday. Going into this earnings report, trough earnings forecasts for many of the most conservative forecasters were $6-$8 per share. Applying historic trough multiple range to those forecasts would suggest the stock is worth anywhere from $90-$160 per share. To arrive at this week’s closing price one would have to assume trough earnings of $3 and a historically low trough multiple of 15x. I submit, it just ain’t going to even come close to happening (outside, again, a US or global recession). That said there is one outlier that is forecasting something not that far off of $3 ($3.53 to be exact) and that’s Morgan Stanley’s Joe Moore who has that as his forecast for 2021 earnings. Joe Moore has been a skeptic for a while now. Even though he keeps on having to increase his near term EPS forecasts every quarter lately because he’s been too skeptical, he has been staying true to his long-term forecast that earnings will go through a major downdraft in the coming years. The funny thing is even Joe Moore doesn’t have a sell on the stock. He has an “equal-weight” on the shares and a TP of $48, as he believes the stock’s valuation already accounts for his outlier forecasts.

Looking more broadly at the rest of the sell-side analyst community, the stock is facing the conundrum of little intellectual rigor/conviction in terms of sell-side target prices. While most sell-side analysts still rate the stock a buy and do believe there will be some degree of EPS decline in FY 2019 (with that likely to be the trough earnings year),most are unwilling to ascribe a “trough earnings multiple” to their “trough earnings forecast.” The outcome of doing so would lead to TPs that are uncomfortably high for most analysts relative to the current stock price. It’s one thing for a sell-side analyst to have a huge upside TP when a stock is galloping higher and loved by the investment community. It is another when the stock hasn’t been working lately and it’s a battleground stock (that set-up can lead to the sell-side research analyst’s obtaining unwelcome attention from their equity research management if TPs are too far out of line). The other approach some analysts are undoubtedly taking is that because they do not have sufficient conviction in their forecasts, they are making up for that by putting a near peak earnings multiple (meaning a very low multiple) on their “trough” EPS forecasts. As they become more convinced that their models are accurate, TPs and earnings multiples are likely to be ratcheted higher.

The refrain from the hardened and recently emboldened war torn memory semi trader is that profitability declines are often longer and more severe than most anticipate at their offing. To this sensibility I remind everyone that a) those same hardened, war torn memory semi traders would never have predicted that Micron would be as profitable as it is today (in FQ1 Micron will make more per share than it did in the peak year of any recent cycle) and b) per my previous update there are a lot of reasons why Micron (and the industry) is positioned much better this time around than it has ever been in the past. Further, as SK Hynix’s performance during the last downturn is evidence of, it is not a foregone conclusion that all memory semis have their profits decimated during downturns. During the last downturn when Micron witnessed its quarterly EBITDA degrade from $1.77bn to $770mm over the course of five quarters (a 57% decline), SK Hynix witnessed its EBITDA decline 36%; appreciably less. One of the primary reasons old Micron underperformed the way it did was because its tech was lagging to a much greater degree than SK Hynix (and Samsung) and its balance sheet was a mess compared to peers (which compounded the impact on earnings during the downturn). Those conditions no longer exist (Micron’s tech is either leading or only slightly lagging peers and its balance sheet is dramatically improved). And bigger picture while that last decline was only four years ago, the diversity and robustness of demand drivers is considerably different now than it was back then (when PCs still played a much more prominent roll than they do now for memory semi demand and the cloud was markedly smaller scaled).

Bringing this update to a conclusion, I believe that the secular shifts in the industry and positive trends are still clearly intact and the valuation remains incredibly compelling. This past quarter’s performance and even the company’s outlook is replete with evidence that the “New Micron” is here and continuing to emerge. While that clearly hasn’t inoculated the New Micron from trading volatility and historically common trading patterns. If Micron’s forecasts play out in line with any of the sell-side’s (save for MS’s) forecasts, investors buying into Micron at these prices should be thrilled with the outcome 12-24 months from now. I say that specifically because if the new trough annual EPS levels for Micron are even just $6-$8 and trough EPS should be met with at least double digit EPS multiples, then upside returns should be very attractive to the Micron investor by the end of the next 12-24 period. I personally am much more bullish on Micron and believe the secular thesis of how the oligopoly industry structure in DRAM, the diversity/depth of demand drivers and the likelihood of reduced downside in this next downturn should lead to a meaningful re-rating of Micron’s shares. But even without that bullish thesis manifesting itself, I submit that the Micron investor with a 12 to 24 month horizon is likely to do very well.

For the trader most worried about the next 4-7 months, while the stock does seem to have bottomed here in the low-to-mid $40s, things could easily happen that could take the stock in either direction. The biggest wild cards from my perspective are a) we still have China/US trade tensions front and center and b) we have this Intel Micron JV that becomes actionable for Micron that could lead to a rewarding strategic outcome. In terms of China, it now sounds highly unlikely that that situation will be dealt with before the US mid-term elections. China seems unwilling to “appear weak” by agreeing to meet/deal with the US right after the imposition of additional tariffs. Further, China appears to be wanting to use this time between now and the US mid-terms to curry favor with other foreign powers (by lowering some of their existing tariffs) and stymie the US’s attempts to bring a more globally coordinated effort to confront China on unfair trade practices. China also seems to want to see if these mid-terms a) weaken the administration and b) provide the administration the opportunity to soften post the election under the assumption that the Trump administration may be especially un-pliable while its campaigning up into the election. That all said, one could see how China and the US could come to some resolution post the mid-term elections but before January 1 as: a) time will have passed since this latest ratcheting up of tariffs b) the implications of the outcome of the mid-term elections should come into better focus and c) China can’t be interested in having all the additional tariffs come into play that will be imposed on January 1 absent some progress made on trade negotiations.

In terms of the Intel Micron JV, January 1 is coming quickly and that’s when Micron has the contractual right to buy in Intel’s remaining 49% ownership position for a very attractive price (<$1bn and the repayment of ~$1.1bn of debt that Intel has fronted the JV). Will Micron try to take advantage of the fact that Intel still doesn’t have a permanent CEO? Will Micron exercise its option straight away because at this juncture, with the US and China still at odds on trade, the chances that Intel will respond by making a bid for all of Micron are more remote? Might Micron choose to wait to exercise its option until late 2019/early 2020 because a) that’s when it is anticipating having its first 3D Xpoint products in the market and b) it allows Micron to use all its free cash flow in the interim to buy-in undervalued shares? Might Micron want to try to force Intel’s hand in terms of making a big offer to rest control of the JV (3D Xpoint is more strategic, arguably, to Intel than it is to Micron), so that Micron can highlight the value of its memory assets to the market and use even more liquidity to buy in undervalued shares? There are a bevy of considerations, but it remains very possible that something happens with this JV in the next few months and if something strategic happens around the IMFT JV it is likely to be very beneficial for Micron shareholders. In sum, I’m not going to pretend to have conviction about how the stock will trade over the next 4-7 months (why that period because it takes us one to two earnings reports out for all the memory semis), but over the next 12-24 months one has to heavily discount a de facto doomsday scenario into Micron’s stock to justify its current stock price. And there are many ways a) that doomsday scenario doesn’t happen and b) a very positive scenario does come to pass. Buy MU! As promised from earlier in the report, a list of positive statements on the outlook for Micron/memory semis conveyed on the FQ4 earnings call: “Our markets are propelled by diversified, secular growth strengths. Our Industry is more rational and the Micron team is executing well against the Company’s strategy. Over the last 12 months, we have sharpened our focus on improving our cost structure and on increasing the mix of high-value solutions in our portfolio, all of which position us well for the future.” CEO Sanjay Mehrotra (“SM”) “Our technology roadmap and manufacturing execution is driving strong cost reductions for the Company.” CEO SM “Turning to mobile. The market continues to grow, driven by content increases for DRAM and NAND across all frontiers.” CEO SM “DRAM and NAND content will continue to grow in mobile devices in fiscal 2019 driven by customer preferences and the need to support new camera and digital features utilizing machine learning.” CEO SM “As discussed at our Investor Day event, we expect industry cyclicality to be more dampened than in the past as industry supply growth from node transitions slows structurally and supply growth requires higher levels of Capex. In addition, we continue to see robust diversified demand drivers and are confident in the long-term outlook for our business.” CEO SM “Looking ahead, we expect a moderation in supply growth (in NAND), beginning in the first half of calendar 2019 as the industry transitions to more challenging 96 layer designs which provide less benefits node-over-node. We also expect higher demand due to elasticity, resulting in higher SSD adoption and increasing average capacities across multiple end markets.” CEO SM “There is some limited inventory adjustments underway at a few customers, but the end customer demand remains solid.” CEO SM “As I reflect on the tremendous progress we have made improving our competitive position this year and amazing opportunities that drive our industry, I’m very excited about Micron’s future. Against this backdrop, we certainly view our stock as being undervalue at current prices and are aggressively implementing our stock buyback program. We will continue to maintain a healthy balance sheet and use strong free cash flow to support our $10 billion buyback and assess opportunities to accelerate the timeline for its completion.” CEO SM “Micron has historically shown strong cash flows from operations even in our most difficult years. And our current cash flow levels are now structurally higher as we have made significant improvements on our underlying cost competitiveness and enhanced our product mix.” CFO Dave Zinsner (“DZ”) “This transformation to the “New Micron” continues to build strong momentum. I am really excited about the future and the best is yet to come.” CEO SM In the answer to Romit Shah’s question “how confident are you that the adjustment you’re seeing is really just a one or two quarter issue?” CEO SM responded, “I think what’s important to understand is that end-market demand trends for DRAM as well as NAND continue to be strong. I mean, when you look at data center and cloud applications – AI is in the very, very early innings. And the whole cloud growth is in the very early innings as well. We see the DRAM requirements in enterprise and cloud data center applications to be growing much faster than the total average DRAM industry demand growth – a CAGR over the course of few years of about 30%. Similarly in mobile applications DRAM is growing nicely as well as machine learning kind of features as facial recognition et cetera get implemented in these phones. We are already starting to see 6-8 gigabyte even 10 GB coming in high-end smartphones now. And days are not too far off that you’ll even see 12 GB of DRAM in smartphones. {authors note Apple’s Xs and Xs Max are up to 4GB this year from 3GB last year and have a long way to go to catch up to the leaders in this area}. “We are extremely focused on technology conversions and technology transitions because they are the best way for us to achieve cost competitiveness as well as ROI on our investments. We are not adding wafer starts.” CEO SM “In NAND, we are in a very good position with the lowest cost technology in the industry.” CEO SM In answering a question on the outlook for demand from data centers posed by Mehdi Hosseini from Susquehanna, CEO SM responded “This is going to be a multi-year growth opportunity for us. We are extremely focused on working closely with the customers in the space. Through those discussions, we get good visibility from them in terms of our opportunity for the future. So we feel very good about our growth opportunities in this segment. . . at our Investor Day we had taken you through the details of how AI driven applications are requiring 6 times more DRAM per server and 2x more SSDs per server . . . The message is, the industry is structurally different and Micron – the New Micron is also structurally different. And we are well positioned to drive the business going forward.” “In both Sanjay and my prepared remarks, we did talk about acceleration {of the buyback program}. If you look at the $1.5 billion and kind of make some estimation around free cash flow for the first quarter, you’ll probably get a figure that is above the 50% free cash number we said we are committed to. So, I think it is in the cards for us to be a little bit more aggressive over time. And we obviously think that this is a fantastic price to be buying the stock at.” CFO DS Disclaimers: All information gathered from sources believed to be accurate, but the accuracy of the information cannot be guaranteed. A number of points made in this update are a matter of opinion and opinion is subject to change without notice.