Seeking Alpha
Deep value, special situations, growth, intensely monitored positions
Profile| Send Message|
( followers)  

Summary

  • Micron is on-target to hit $3.27 EPS (non-GAAP) this year.
  • Samsung's announcement of DRAM price increases is a game changer.
  • New substantial demand potential is another game changer.
  • My price targets have been adjusted to reflect increasing prices in DRAM.

This article is a follow-up to my recent Micron (NASDAQ:MU) article "My New Micron Price Targets", which was published June 27th. As with the first article, this article will have a lot less numbers than is usual for me and a lot more conceptualization. My conclusions are mostly judgment calls, which are difficult to substantiate other than with my educated opinion based on the available information.

I recommend that readers pause here and go read or re-read that prior article, and especially the comments. There is a lot of value in the comments.

I'm following up that first article so quickly because a positive and very significant event happened on June 30th, just three days after my article was published. On that day, Taiwan's Economic Daily News reported that Samsung (OTC:SSNLF) plans to raise DRAM prices by 10% starting in July. In my view, this is a disruptive event in the memory industry. And then, a day after that, another event happened, this one negative. A few days later, more event happenings occurred quicker than I can write about. I'm sure more will happen tomorrow and the next day, but I'm not going to start over again and re-write this article every time one happens. We'll just have to deal with them in the comments.

That's the way it always is for me. My three price targets for any given stock rarely last a whole week without something happening requiring adjustment. In the case of Micron, a day or two is the usual life span. I'll type quick.

DRAM

First, the positive.

As those of you who follow Micron and read my posts know, I have been predicting that the oligopoly in the memory industry, what William Tidwell named "MIPO," will coalesce behind a leader, which will be either Samsung or Micron.

I have also been predicting that the oligopoly will raise DRAM prices, just because they can. I drew a lot of intelligent disagreement to this prediction based on arguments that:

  1. It's never happened before. Historically prices have always come down in this industry, in particular, and the semiconductor industry in general; and
  2. Costs go down as smaller nodes are reached, and lower costs means that lower prices can maintain or increase gross margins without any price increases; and
  3. Customers expect to get the benefit of those lower costs, and if they do not, they will go elsewhere.

I also said that it looks like my expectations were early because of fixed price contracts not expiring as rapidly as I had thought they would, and I said:

It's also the result of this oligopoly's failure, so far, to either realize or exercise the pricing power which it has. The realization has probably dawned on all the members but they haven't yet acted on it. Usually, it takes a clear leader to trigger an oligopoly's choice of actions and, so far, no clear leader has emerged, as far as I can tell.

Well, that was then and this is now. Samsung announced that it was raising contract DRAM prices about 10% effective this month, thereby assuming the role of leader in the oligopoly.

It's different this time (Thanks, Russ). It really is. The significance of this price increase cannot be overstated. It's the beginning of a rapid expansion of margins, both gross and net. DRAM prices are going up. And DRAM costs will continue going down.

That's all we know so far. But, based on almost universal oligopoly historical patterns, here is what I expect will happen.

First, both Micron and SK Hynix (OTC:HXSCL) will raise DRAM prices quietly, citing market conditions or some-such generic non-speak. I expect each to raise prices between 9% and 11%. This is the follow-the-leader activity that the existence of any oligopoly fosters. Together, these three manufacturers accounted for 91.5% of DRAM revenue in the 4th quarter of 2013. The next largest DRAM supplier was Nanya, Micron's partner in Inotera, which accounted for 4.3%.

DRAM MARKET SHARES

Source: DrameXchange

Nanya can be expected to play follow the leader as well, raising prices about 10%. That leaves less than 5% for the rest of the universe, which will happily follow suit.

So now we have to guess at what rate DRAM prices will continue increasing. DRAM represents about 68% of Micron's production measured by sales. The rate of future price increases will markedly affect my projections and price targets. DRAM prices could have gone up 10% this month, for a one-time increase until forever. Or, DRAM prices could be going up 10% every quarter over the next five years. Or, more likely, something in between these two extremes. My middle and upper price targets are highly dependent on where we put that future rate of price increases.

I looked at price increases going forward of between 3% annually and 5% quarterly (21.5% annually), both of which I consider within the realm of the possible.

In the very near term, DRAM prices are going up a lot more. Digitimes reports, citing industry sources, that spot prices for DDR3 4Gb DRAM stand a chance of rising to $4.80-5.00 in the third quarter of 2014, after hitting an 18-month high of US 4.35 in late June. (subscription required for Digitimes content.)

And we have to factor in the possibility that Samsung, or someone else, will build a new DRAM plant adding external capacity (as opposed to increasing capacity within the existing fabs, which I will call "internal" capacity). Citing those same industry sources, Digitimes reports that Samsung is considering building a new DRAM fab.

But Herb Greenberg, citing The Fly, says it's not so.

And, in response to a Morgan Stanley report stating that Samsung plans to raise its DRAM output, Sterne Agee says it doesn't believe that Samsung is bringing any new capacity on-line. I think that the pundits were confused by reports that Samsung was planning to increase capacity into thinking that meant building a new fab. I think those reports meant Samsung was planning to increase capacity in already existing fabs by adding lines, new equipment, and the like. But a new fab increasing external capacity is a possibility.

In order to factor that possibility into our projections, we have to estimate the probability of it happening, the time the new fab will take to get into commercial production, and the quantity it will produce. Then, we have to project its impact on pricing. Attempting to quantify this number of unknowns results in so broad a range of possible effects so as to be useless as a practical matter. I tried. I wound up just guessing at the effects of these kinds of maybe or maybe-not events. My guesses, though, are not pure dart throws. I have a lot of non-quantifiable information which goes into those guesses. Is there such a thing as "informed guessing?" After going through that process, I discount the possibility of Samsung building a new fab at this time, which negatively affects pricing as 1%. Essentially nil.

NAND

What about NAND? NAND is about 30% of Micron's sales volume.

In NAND, we have a slightly different oligopoly. As of Q1 2014, we have Toshiba (OTCPK:TOSBF)/SanDisk (NASDAQ:SNDK) at about 40%, Samsung at about 30%, Micron at about 15%, and SK Hynix at about 8%.

In my prior article, I said:

Micron is committed to increasing SSD attachment rates by holding NAND prices down. And given the high quality and low price of its Crucial brand of SSDs, this appears to be a winning strategy.

Although there are credible contrary views, this is still my opinion, but I'm much less convinced that I'm right, given the comments to my first article. Whether intentional or not, it seems to be working. Micron's Crucial brand is out with the MX 100 solid-state drive. This is a consumer-oriented product, which continues along the current path of lower price per bit SSDs with higher quality and capacity. The lower price is still profitable because it is made on Micron's 16nm MLC NAND, its latest node shrink. Smaller allows more bits per wafer at the same cost per wafer, and thereby, a lower cost per bit. Reviewers who tore it down say that it is indistinguishable from the M500 except for the smaller node bits.

I think that we will see higher NAND sales internally as more NAND bits are put into Crucial SSDs and sold as finished product. In today's world, the bottleneck in the desktop PC is typically the time it takes to pull data from the PC's hard disk drive. Replacing that drive with an SSD results in quicker boot-up times and shorter compute times for data-intensive projects. The experience has been described as close to getting a "brand new" computer.

But practically everybody disagrees with my view that Micron wants lower NAND pricing. And that "everybody" is a very knowledgeable group.

Digitimes reports that NAND flash chip prices are likely to be hiked soon, saying that Micron Technology and SK Hynix have informed downstream memory clients that they have run out of certain products.

According to Josephine Lien and Steve Shen of Digitimes, Contract prices for NAND Flash chips were up 2% to 5% in the second half of June, and are expected to continue increasing into the third quarter. They attribute rising prices to the increased popularity of cloud computing applications and SSD devices.

So whatever Micron may want, NAND prices are, in fact, going up in the immediate future.

So now that you are all as confused as I am, what does this all mean? I think the answer may lie in why I think Micron wants NAND prices down. I think that is so because Micron spokespeople have hinted at that. In numerous analyst meetings, the Micron spokesperson has said NAND pricing is projected to be down next quarter. That spokesperson then followed that statement with a discussion of attachment rates for SSDs, and how attachment rates go up as prices come down and how beneficial higher attachment rates are for Micron. I took the hint after the third time that Micron did that. The hint was that Micron wasn't just saying "it's not so bad." It was saying that it was actually good. Maybe what's really going on is that Micron knows that NAND prices are coming down and doesn't mind.

So I got out my trusty spreadsheet and started playing with some numbers. Two days and 11 sheets later, I quit. All I could do by playing with the numbers is satisfy myself that there is some lower price which is profitable for Micron to sell at, which may be low enough to trigger a large enough increase in the attachment rate so as to be more profitable to Micron than the higher price. "Some" and "may" and "more" are not much help. I also came to the realization that at some lower price, a "tipping point" happens. End-use PC consumers no longer want hard disk drives in their new computers. They want solid-state drives. I cannot quantify what that price is, but if we get there, I can quantify sales enough to say that Micron's NAND sales will likely be higher than its DRAM sales. Maybe not so profitable, though, since the lower price necessary to cause that tipping will have very thin margins.

So what does this mean? Have we just gone from one unknown to another? I don't think so. At least not for me. My instincts, after immersing myself in all these possible scenarios with different probabilities and different financial results, tell me that it is going to happen, and not too far into the future. I get there by looking at what I would do if I owned Micron myself, beholden to no one and nothing except my own personal bank account. I would hold NAND pricing down, if I could, to barely profitable numbers to try to tip that scale into SSDs instead of HDDs. That would not kill hard disk drives. They still have uses which SSDs cannot compete with. For example, my PC backups are to an external hard disk drive. It needs to be turned on, spin, and consume power only once a day for a short time. So the advantage SSDs have over HDDs in not using electric power continuously by spinning all the time is almost non-existent for this use. And I don't much care how fast it is. I am not waiting for the backup to finish in order to do something. So as long as it is even a little cheaper, I'll use a hard disk for external backup.

I have learned to go with my instincts when they come from data immersion. They usually turn out to be right. But for reasons which will become obvious, I have left the effect of NAND out of the projections and price targets. As of now, I'm treating it as a slight possible positive.

I took potential DRAM price increases at various rates, and projected that impact on Micron's bottom line over a fourteen-month period, through the end of the coming year. Just because the oligopoly will raise prices because they can does not mean they will raise them to an arbitrarily high point. At some point, the prices will be high enough to attract a new entrant, despite the high cost and barriers to entry. That price will be higher than the price at which a new entrant thinks that it will be able to make a "nice" profit, whatever "nice" means to that prospective entrant. It will be higher because the new entrant will be aware of the defensive strategy an oligopoly can use to defend the oligopoly's exclusive membership. The oligopoly can just lower prices to the point where it is not profitable for the new entrant. Since the new entrant has a new $2 to $5 billion depreciating investment, that will be a price at which the new entrant loses more money than it wants to but the existing oligopoly membership still make a profit. So the barrier to entry into an oligopoly-dominated business is quite high. Typically, a new entrant has to have a more efficient process or a better product and deep pockets.

Let's do some numbers. Call revenue for this year ended August 29 at $16.3 billion. At 68% from DRAM, DRAM revenue is $11.1 billion. A 10% increase in prices produces an extra $1.11 billion. Now extra revenue from price increases doesn't create extra costs except for taxes on the profit and commissions, if there are any. Let's say the extra costs are 10% of the extra revenue, or another $.111 billion. That leaves $.999 billion dropping right to the bottom line. $.999 billion on 1.190 billion shares outstanding is an extra $.84 per share in earnings. If we do price increases of 5% a quarter, using the same methodology, we get an increase in earnings per share of $1.81.

If these price increases had been in effect for all of this year, EPS would have been $4.11 instead of $3.27 on the lower assumption, or $5.08 on the higher assumption. Going ahead to the coming year ending August 27, 2015, on assumed earnings of $4.30 per share, about the mid-point of the analysts' estimates, and if we get cumulative price increases over the two years, we'd have earnings of $5.98 a share on the lower assumption, or $7.92 a share on the higher assumption. Not too shabby.

I have discounted the possibility that Samsung is building a new DRAM fab to 1%. No one has seen it or any preparation for it, nor would it make smart business sense. Samsung is not stupid. Besides, it's tough to hide a plant that size. Even if you put it in a jungle, you need roads to get the construction equipment in. And if Samsung were to start now, it would take three years to produce in commercial quantities. I think that someone has confused Samsung increasing internal capacity with Samsung building a new plant.

There's more. We have a substantial increase in DRAM demand coming almost immediately, without any corresponding increase in supply. That means more profits on more sales, and even more upward price pressure. That DRAM demand increase is coming from Oracle's (NYSE:ORCL) new In Memory Database (IMDB). The potential is very large. Electric Phred, who knows about as much as anyone I know about this industry and Micron itself, tells the story here. If you haven't read it, you should do so.

I'll summarize one point. According to Credit Suisse's John Pitzer, if Oracle achieves a 2.5% penetration rate with this in memory option, there would be an incremental demand for DRAM of about 1 billion GB. That is about as much DRAM as all the cell phones in existence. Pitzer translates this into continuing earnings of $5 a share for Micron. Since Pitzer projects Micron as earning $3.10 in 2014, I presume that he means the IMDB DRAM demand would add an additional $1.90 in earnings to get to his $5 number.

Pitzer's piece was before anyone thought Samsung would raise contract DRAM prices 10%, so it is on top of the increases we have talked about above. That additional $1.90 would be about $2.42, with the 10% increase. Remember, Pitzer thinks the $1.90 is after expenses, so the addition to revenues would have to be about triple that; triple $1.90 is $5.70, a 10% price increase adds $.57, of which $.52 drops to the bottom line in addition to the $1.90, making the increase $2.42, which comes out to $5.69 if we add it to my projection of $3.27 for this year and $6.53 if we also add in the $.84 from the 10% price increase on the lower assumption, or $7.50 on the higher assumption. Of course, these numbers are "what if", and this had been true for all of this fiscal year going back to August 30, 2013. But let's add up next year ending August 27, 2015. I'm going to use Pitzer's numbers, because I have not yet finished the necessary research in order to calculate my own. I have, however, looked at the issue and done some calculations. My admittedly premature calculations indicate that Pitzer is not overly optimistic. He may be overly conservative, however. I can't tell just yet. So let's just use his numbers for now.

For 2015, we have $4.30 in analysts' estimates for EPS. Let's add in a cumulative 10% price increase on the low end and a 5% per quarter price increase on the high end, both cumulatively for the two years. We get $5.98 on the low end and $7.92 on the higher end. Now let's add in the $2.42 for profit on the IMDB revenues. We get $8.40 on the low end and $10.34 on the high end. These are numbers for the next eight quarters, so they don't track the fiscal years exactly.

OK, OK, I get it. I must have made a mistake in math somewhere. These numbers just can't be right. Would one of you please find my mistake and point it out to me so that I can redo the numbers correctly? I can't buy any more Micron stock. I have way too much already.

So in this land of unbelievable numbers, what are my price targets? Remember, I use three price targets in my investing:

  1. A Risk Target. This target represents what I expect any downward price movement to be limited to, in the event that things go wrong in a weak market climate.
  2. A Reward Target. This represents the price I'm looking to get to. It's the point at which I feel that the stock has reached its then-current potential, and represents the point at which I will be moving any remaining invested funds into another investment which I feel has not yet reached its potential. I often do not reach this price target. The reason that I don't is because by the time the price has gotten to that level, higher earnings are in the books and the growth story has been going on for a longer time and is more accepted, so this target has to change to reflect those facts.
  3. A Balanced Risk/Reward Target. This is the price at which I think the risk is about equal to the reward. It's basically my "hold and watch-it-like-a-hawk" price. Over this, I will begin to lighten up a little.

So what are they now on the various possibilities outlined above?

Well, the Risk Target is the same under all scenarios, since it assumes that nothing goes right and therefore none of those scenarios come to pass. My Risk target hasn't changed since my prior article. It's still $32. That's as low as I think Micron can go and remain for other than a very brief period.

How about the Reward target? It's different for the lower price increase scenario, the higher price increase scenario, the IMDB demand increase scenario, and the combinations of IMDB/low price increase and IMDB/high price increase. That's five possible scenarios. Let's do them all.

1. Low DRAM Price Increase/No IMDB demand:

We have annualized EPS running at $4.11. Flat-to-low growth supports a PE of 10, making this price target $41.10. Higher with annualized growth of 10% or more.

2. High DRAM Price Increase/No IMDB demand:

We have annualized EPS running at $5.08. Flat-to-low growth supports a PE of 10, making this price target $50.80. Higher with annualized growth of 10% or more.

3. IMDB DRAM demand increase/no DRAM price increases beyond 10% already announced:

We have annualized EPS running at $5.69. Low growth of maybe 12% annually supports a PE of 12, making this price target $68.28.

4. IMDB DRAM demand increase/Low DRAM price increases:

We have annualized EPS running at $6.53. Growth of maybe 14% annually supports a PE of 14, making this price target $91.42.

5. IMDB DRAM demand increase/Higher DRAM price increases:

We have annualized EPS running at $7.50. Growth of maybe 14% annually supports a PE of 14, making this price target $105.

OK, if we swallow (GULP!) those numbers as possibilities, what's my Balanced Risk/Reward Target. At what price, before I know which scenario is going to play out, would I start to lighten up? This is difficult, because there are so many permutations of the probabilities. There is no way I can do this one mathematically. I could assign probabilities to the various scenarios and lay them out statistically, but I deal with this kind of situation differently, because I find it's more accurate for some unknown reason. What I actually do is immerse myself in the research (e.g. write this article), and then sleep on the immersion and dream. When I wake up, I have two cups of espresso, wait a half hour, and pick a number. This number is influenced by my personal financial circumstances. Yours will be different. How we feel about risk depends on what, qualitatively, we are risking. I'm not risking much qualitatively, since I have a lot more than I need. So this number is personal to me.

I come up with about $60, up a lot from my $48 balanced target just a month ago. But a lot has happened since then. My Balanced Risk/Reward Target is the price at which I would sell a little Micron stock. If nothing changed (which is virtually impossible), I would sell some at $60, a little more at $63, a little more at $66, etc. But by then, circumstances will have changed. If for the better, this target goes up. If for the worse, I sell more.

Your mileage will, no doubt, vary. Mine will, after I read your comments. I change my mind a lot when I read articles and comments on Seeking Alpha. That's why I love the site. So please don't be reluctant to tell me how stupid you think I am. If I think you're wrong, I won't take offense. If I think you're right, I'll thank you for it. Either way, it works to my benefit.

Editor's Note: This article discusses one or more securities that do not trade on a major exchange. Please be aware of the risks associated with these stocks.

Source: Micron: Some Recent Developments And New Price Targets