Description

Summary

Micron Technologies (MU) is a significantly undervalued, misunderstood equity in the midst of a multi-year industry transformation. The transformation, which has reshaped the DRAM industry into a highly profitable oligopoly, is in its middle innings and is experiencing a moment of significant investor doubt thereby providing one of the best entry points into the stock over the past couple of years. Short term weakness in PC DRAM pricing due in part to weak PC demand (and Samsung temporarily shifting mobile bits into the PC channel) has overshadowed overall very strong DRAM industry fundamentals and positive industry and MU specific developments which we explain below. Historically, PCs were the dominant end market for DRAM (representing >70% of bits sold as of 2009) and hence were the driver of DRAM pricing. Reflexively, investors learned to extrapolate poor PC fundamentals into poor DRAM fundamentals and sell off DRAM producers on any hint of PC weakness. This dynamic has rapidly changed; PC  fundamentals are no longer the critical factor to DRAM fundamentals with PC’s now representing less than 25% of bits sold and dropping rapidly due to explosive demand growth for bits from mobile and cloud (enterprise computing/networking) end markets. Particularly, mobile bit demand growth is driven by content growth (more DRAM per handset) which is a trend that should persist for some time. In 2H 2015 as the iPhone 6 moves from 1GB to 2GB per phone handset producers are bracing for tight DRAM supply. As such, we envision a significant shift in investor sentiment in the coming months as investor focus moves away from an over emphasis on PC industry fundamentals and towards the strong overall DRAM fundamentals driven by producer supply discipline, stable pricing, and explosive demand growth. For MU we envision strong earnings growth, multiple expansion, and accelerating share buybacks in the coming 12-24 months.

We expect the weakness in DRAM pricing to subside within a quarter or two at which time we envision MU's stock to continue its multi-year upward trajectory. At 5.7x our forward (FY‘8/16E) EPS estimate the multiple is undemanding while we expect earnings growth for the next few years to exceed 25% per annum. We believe the DRAM industry transformation is about 2-3 years behind the Hard Disc Drive (HDD) industry transformation and expect that DRAM producer multiples (including that of MU) will gravitate closer to those of Seagate and Western Digital (10-13x) as the DRAM transformation becomes more apparent. Using an '8/17 FYE for MU, and assuming they can earn $6.00 of EPS and using a multiple of 10x would bring MU's price to $60 or a return of 120%.

We highly recommend reading Aviclara’s stellar VIC write-up from 04/15/2013 which can be found here: http://www.valueinvestorsclub.com/idea/MICRON_TECHNOLOGY_INC/97865.

Our write-up seeks to address developments that have occurred over the two years since his write-up and to highlight the currently attractive entry point into the stock.  (incidentally, MU is due to report their fiscal Q2 earnings tonight after the close).

Description

Headquartered in Boise, ID, Micron is a leading manufacturer of memory semiconductors. The company manufactures a full range of DRAM, NAND-flash and NOR Flash memory products, for leading-edge computing, consumer, networking, automotive, industrial, embedded and mobile products.

The business is split broadly into four main segments: 1) DRAM, 2) NAND-Flash, 3) NOR and 4) Other. We believe that the most important story here for MU is its DRAM segment, which generates ~78% of the company’s gross profits (68% of revenues), followed by NAND at 20% of its gross profits (28% of revenues).

We believe that MU holds about a quarter of the DRAM market, as well as around 14% of the NAND market:

Concerns that have weighed on DRAM and Micron

We believe that MU has been overly penalized due to the negative anecdotes that we have seen, mostly related to weaker-than-expected PC shipments as well as the strengthening US dollar:

Weaker-than-anticipated PC shipments

We think that PC weakness started precipitating soon after Microsoft reported 4Q’14 earnings on 01/27/15, when the company reported a fall in profits amidst a strong US dollar and a slump in personal computer sales, cooling demand for Windows software. On 03/12/15, IDC had announced that global PC shipments will decline by 4.9% this year, a larger decrease compared to its prior guidance of 3.3%, marking a sign of weakening demand for desktop and portable PCs. This came a week after INTC had negatively pre-announced, citing weak corporate demand and sluggish economies, particularly in Europe. This has undoubtedly created a “spillover” effect onto PC DRAM manufacturers, specifically MU, which had sent ~35% of its bits (industry average of 25%) into the PC vertical. As a result, analysts across the board started reducing estimates for more drastic ASP (average selling price) declines and hence earnings potential of MU.

In addition to the backdrop of weakening PC demand, we have also heard (from industry as well as MU management itself) that Samsung has been the price aggressor in the PC DRAM market. More specifically, for a standard 4Gb module of DRAM that goes into PC, Samsung has reduced prices in that channel to as low as $26-28, down from ~$30-31 (which was that MU was offering these modules at), which further exacerbated the weakness in PC DRAM.

Why did Samsung do that? Our channel checks and other work has led us to conclude that there were a few factors to Samsung’s aggressive pricing strategy. Firstly, as most are aware, Samsung has been losing market share in its mobile handset segment to lower-cost Chinese competitors as well as AAPL. As a result, internal consumption of mobile bits has been below Samsung’s internal forecasts. To address that issue, Samsung made a choice to direct DRAM bits away from its internal mobile consumption into the PC channel. This was quite logical because PC DRAM pricing over the past 6 quarters has been better than mobile pricing enabling Samsung to transfer the bits, price aggressively, while maintaining target gross margins. Its important to highlight that the premium DRAM pricing in PCs was somewhat artificial in that it was a legacy of Hynix’s 2013 reduced production due to a fire at its Wuxi fab. Otherwise, DRAM bits should be mostly fungible and should command equivalent pricing/margins across end markets all things being equal. As such, the over emphasis on PC DRAM pricing declines also fails to consider that much of the pricing decline is due to a normal reversion to the mean post the Hynix fire. In essence, Samsung has been a rational actor in arbitraging the pricing differential between PC DRAM and mobile DRAM pricing. While this has occurred, the fall in DRAM pricing has led investors to fear that either Samsung is becoming an irrational actor and/or that fundamentals for DRAM are in fact weaker than expected. Our work leads us to believe quite the opposite; namely that Samsung was closing the pricing arbitrage (it is in fact, we believe, close to fully closed) and in 2H 2015 Samsung will be sending their excess bits to supply the iPhone 6’s content jump (from 1GB to 2GB). As such we view the deterioration in investor sentiment to be overdone and misplaced and expect a strong 2H for DRAM fundamentals and for MU.

Furthermore, in looking at overall bit demand growth by end market we foresee strong growth which should match or exceed bit supply growth in 2015 and beyond.

We stress tested our belief that PC weakness is unlikely to severely impact overall DRAM fundamentals and concluded that even if PC bit demand declines by 10% (which would imply more like -20% unit decline (400% worse than Gartner estimates) with +10% content growth), the overall DRAM industry would still be in supply-demand balance, with bits growing at 25% industry wide:

As such, we believe that the PC concerns have been overblown and that weaker-than-expected PC demand does not threaten the supply-demand balance of the industry. In fact we believe the current industry view of “as PC fundamentals go so go DRAM fundamentals” is an outdated view that has created an attractive entry point into MU.  

Strengthening of the US Dollar

A recent concern for MU and DRAM producers has been the impact of a strong US dollar on consumer electronics sales (including mobile handsets and PCs). In general MU is a strong dollar beneficiary in that MU settles most of its DRAM contracts in USD and it has significant operations in Japan via its Elpida subsidiary (all its mobile bits are manufactured there). However, there is speculation that US centric consumer electronics OEMs are raising prices in foreign markets, in order to keep either margins or profit dollars at similar, historical levels. Thus, the concern is that that the “strong dollar”will in effect, curb demand for consumer products that have faced price increase due to the stronger dollar. We believe the strong dollar has been a secondary concern suppressing investor sentiment in semiconductors in general as well as in MU specifically. While we would not dismiss the dollar concerns outright, we’d highlight that a strong dollar is a mixed bag for MU and that DRAM, via content growth in handsets, and cloud growth has a compelling secular story that should be able to withstand a strong dollar. Secondly, we believe that DRAM producers are focused on maintaining high margins and act to ensure that bit supply is managed accordingly (more on industry structure below).

Unwarranted victim of SNDK’s sell-off

We also think that MU has been a casualty of SNDK’s recent dismal performance (and negative pre-announcement). On 03/26/15, SNDK announced that Q1’15 revenues will be ~$100M less than anticipated, around a 7% reduction. That was the 2nd negative pre-earnings announcement in a row, following a mid-January cut to forecast. SNDK also withdrew its full year forecast and postponed its scheduled May investor day.  Even though NAND makes up about one-fifth of MU’s gross profits, we think that MU’s share price was a casualty of SNDK’s negative pre-announcement due to the overlap in business segments (NAND) as well as end market products.

Our research indicates that SNDK’s disappointing results were primarily idiosyncratic to SNDK’s NAND mix (heavy retail skew) and internal production issues and is thus a poor read through to MU. Firstly, SNDK’s woes stemmed from the fact that almost half of its NAND bits end up directly in the retail market, in the form of SSDs, SD cards etc which have negative repercussions in today’s global environment: increased international exposure also means heightened US-dollar conversion woes, slowing growth in international markets. In addition, SNDK also cited product qualification issues (we think it may be issues with their new node transition). We also think that the fact that SNDK isn’t a captive DRAM manufacturer had also hurt SNDK’s results: as the need for mobile memory increases, coupled with the tightened DRAM supply, SNDK is at a disadvantage as it does not own captive DRAM memory and has to purchase this from competitors, whereas someone like MU, Samsung and Hynix can package both their NAND and DRAM together for sale to their end customers.

In contrast, we view MU’s NAND business as a free option, a turnaround story that can provide future upside but one that we don’t see as a key driver of our current MU thesis. MU does not have a significant NAND retail business and is far better positioned to package its DRAM & NAND for mobile use. In terms of NAND as a source of future value for MU, our view is that NAND industry structure is trailing that of DRAM by around 2-3 years in terms of 1. Moore’s law and the increasing barriers to shrink, and 2. More curtailed/controlled production of NAND leading to less supply, albeit with explosive growth in demand due to the growth in big data computing and enterprise storage requirements (both at >50% CAGR). We view the NAND portion of MU’s business as one of a few free options from owning the stock.

Further we believe that MU has increasingly taken steps to improve its NAND business:

1. The company has since placed more emphasis on eMCP by moving more bits from PC DRAM (35% down to 25% by end of this year) to mobile DRAM (from 25% to 35%/40%), which it will package together with its captive NAND bits, which will create a lift to the company’s NAND margins in the long run (though with a short term compression in the short run as PC DRAM tends to have a higher GM profile than mobile DRAM)

2. MU announced its collaboration with Seagate to make SAS enterprise SSDs was also a net positive development, as enterprise SSDs typically have significantly higher margins than raw NAND, and so we see this deal as incrementally positive MU’s NAND margins

Positive new developments for DRAM and MU:

New Inotera agreement

On 02/10/15, MU announced that it has entered into a revised supply agreement with Inotera, effective 01/01/16. Previously, MU secured bits from Inotera on a “pricing minus” mechanism which essentially meant that MU’s margins are relatively fixed if there were any improvements in pricing and margins. As a result, Inotera had operating margins of 51% vs. MU at around 20% range.

With the revised “equal share margin” agreement, MU will be able to attain an equal share of the economics, split with Inotera.

We think that Inotera manufactures a third of MU’s DRAM bits today. According to Inotera’s press release, the company expects that it would take a ~500bps hit to its gross margins, which is around ~$240M USD. We estimate that MU’s Adj. EBITDA in 2016 to be  ~$8.3B, and so this pricing agreement will essentially contribute ~2.7% increase in MU’s EBITDA, or around 200bps to MU’s gross margin.

Further, we view MU’s restructuring of the contract towards a more variable structure (‘i.e.’ more upside if pricing is strong, more downside if its weak) as strong evidence of management’s confidence in the DRAM industry’s bright prospects and improved industry structure.

Buybacks / Return of Capital

Over the past year management has made a few important adjustments and enhancements to its capital allocation efforts all of which signal long term confidence in their business. First, the company has ceased issuing dilutive convertible bonds and has instead moved toward issuing low cost, long dated bonds. We view MU's ability to raise debt at 5% (when FCF is 10% going to 15%+) to retire converts and buy back stock as a strong indication of management's confidence in their future prospects. Secondly, there is now a clear path towards a future where share count will be decreasing. As such, under the current capital allocation regime equity investors are poised to gain greater upside leverage if a base case or upside case plays out.

On 10/27/2014, MU initiated a $1B buyback program. So far, the company has spent ~$200M on buybacks. We expect MU to continue to opportunistically buy back shares, especially at today’s level. In addition, MU has also spent ~$2.8B in the past 15 months to deal with its dilution management, and we believe that the company will continue to prudently return capital through both avenues. Going forward to expect the company will increase the size of its buyback program as well as issue a dividend. Both represent additional catalysts and positive signals to the market place.

Industry structure

Underpinning the MU thesis is the premise that the DRAM industry has become a well functioning oligopoly. Of greatest concern is Samsung and their willingness to chase margins versus market share. Recently there have been credible reports that both Samsung and Hynix are slowing their capacity additions to ensure that DRAM supply stays tight. Incidentally, both additions are simply to replace supply which is going away due to shrink. We view these anecdotes as evidence that behavior of DRAM competitors has changed quite dramatically, consistent with what we’ve seen in many industries that have experienced similar consolidations. Further, we believe that Samsung, as a vertically integrated handset producer, is not at all incentivized to oversupply the DRAM market. This would simply allow their low cost Chinese handset competitors to buy DRAM cheaper and further undercut Samsung on handset prices. Currently most or all of Samsung’s operating profits in their handset business comes from memory products and we anticipate that Samsung will not act irrationally to jeopardize these strong operating profits.

Turning it into numbers

Reconciliation to Free Cash Flow:

Valuation

Assumptions

Note:  DRAM growth is greater in 2016 as more of the effects of MU’s 20nm migration will be witnessed in 2H 2015 and 2016.

Conclusion

We find that at its current valuation, MU is attractively priced and reflects poor investor sentiment that should reverse in the coming months. Based on our projections, we believe that MU trades at a 7x forward P/E (non GAAP earnings), and <5x EV/EBITDA, and at a 10% FCF yield, going to 15% in 2016. We think that MU’s DRAM business is undergoing a major transformation analogous to the HDD business 5 years ago, and expect to see both earnings growth and multiple re-rating/expansion once we see stabilization in PC DRAM.

Risks

1) Samsung breaking the oligopoly by chasing market share or over supplying the market

1. Recently, Samsung has apparently told its larger shareholders that they are paring back its DRAM capacity addition (which was for the loss of bits due to technology migration)—this push-out signals Samsung’s willingness to act like a rational actor in this space

2) Global slowdown / recession which may hurt end markets and the industry’s supply-demand dynamics ( a strong decline in mobile and other end markets would constrain strong demand growth albeit DRAM producers have signaled willingness to reduce supply accordingly)

3) Temporary glitches in MU’s 20nm migration (so far this process seems to be going well)

4) New entrant threatening the oligopoly (a big DRAM customer like Apple or a state sponsored entity). Unlikely in the near future due to IP issues, new fabs require billion of dollars and technology know hows, engineers are in very short supply

Catalyst

1. PC DRAM pricing stabilization over next couple of months

2. Improving 2H'15 DRAM fundamentals driven by Iphone upgrade from 1GB to 2GB per handset

3. Analysts' rolling forward valuations to FY'8/16E (next 6 months)

4. Continued buyback (stock and converts)

5. Newly issued dividend (2016)

6. Inotera deal impacts earnings (starting 01/01/2016)

7. NAND business turnaround

8. Validation of industry supply discipline