- The memory industry is an oligopoly that is united in its intent to reduce industry capacity and thus support higher prices and much higher gross margins going forward.
- The investing community, particularly sell-side analysts, are still mostly clueless about this new reality. MU is sporting a forward PE of barely above 10, SNDK less than 10.
- MU and SNDK deserve higher forward PE’s and investors will see these rise towards 15 later year after a strong series of quarterly results showing strong revenue and profit.
- The mechanism for price and profit momentum is explored and explained.
When the British Empire lost the last battle of the American Revolution at Yorktown, tradition has it that Lord Cornwallis ordered his pipe and drum corps to play an ancient English ditty called "The World Turned Upside Down". It was perfect for the occasion. Who could have ever believed that the powerful British would be defeated by the rag-tag collection that was the Colonial Army? The last line of the song is:
"Yet let's be content, and the times lament, you see the world turn'd upside down."
Strange things had happened in this New World. The Brits, as they marched back to their ship must have been thinking, "Who could have seen that coming?" EVERYBODY knew that the British Empire couldn't be beat. Turns out, though, the rules that governed warfare on European soil didn't work in a new environment against a highly motivated opponent. Conventional wisdom dies hard, that's for sure.
Investing in the memory industry provides yet another example of the power of conventional wisdom. The solid state memory industry has been in existence since the 1970s, and in that time it has swung through massive cycles of value creation and destruction. Investors who have followed the industry and the companies that participate in it have been repeatedly burned by large increases in industry capacity, which outruns customer demand, thereby leading to ruinous drops in average selling prices. This in turn leads to heavy losses on the part of the memory suppliers and their unfortunate investors. Rinse, wash, and repeat. Invest in a memory supplier VERY carefully, the conventional investing wisdom goes, and beware of rich valuations - you are investing in a business cycle that cannot last; a future that rarely delivers the hoped-for potential. Remember, the investor can be blind-sided by supply/demand imbalances that seem to develop out of nowhere. This is the hard-earned wisdom of memory industry investors.
Well, maybe it's time for investors to learn to love this old song, because investing in the memory business going forward is investing in a new world - a world turned upside down. The article is the second in a series whose purpose is to survey this new world and map the terrain for a new generation of investors who are willing to discard conventional wisdom and take a fresh look at their surroundings. There is still plenty of risk for the unwary and the uninformed, but the really good news is that the memory industry offers tremendous opportunities to harvest value if you know the new rules. Let's move forward and do some exploring.
To briefly summarize, the (solid state) memory business has over the last few years coalesced around 4 main entities that supply 95% of the product coming to market worldwide. Two of these entities are joint ventures (JVs) - these are Micron (NASDAQ:MU)/Intel (NASDAQ:INTC) (Micron) and SanDisk (NASDAQ:SNDK)/Toshiba (OTCPK:TOSYY) (SanDisk). Two are single companies - these are Samsung (OTC:SSNLF) and SK Hynix (OTC:HXSCL). Oversimplifying, these manufacturers produce two types of product - Dynamic Random Access Memory (DRAM) and NAND, or "flash" memory. Of the four, only three produce DRAM, excluding SanDisk, which only makes NAND. These four entities, which I have termed the Memory Industry Producers Organization (MIPO), constitute a materially different industry structure that has the power to turn old school investing wisdom on its head.
Current Manufacturing Capacities in thousands of wafers per month in the IC industry as a whole are as follows:
|2013 Rank||Company||2013 Installed Capacity (Kw/m)||2013 % of WW Total|
Source: IC Insights
Note that the top three companies in terms of volume in the IC business are memory producers. MIPO's fourth member, SanDisk, runs a fabless model that relies on Toshiba's foundries for production of their designs. They account for 208 kwm of the volume listed above. Another modifier regarding the numbers above is that the bulk of Samsung's production is used in the manufacture of their own products. Samsung's merchant production is estimated to be less than Micron's. This fact has tremendous implications for MIPO's pricing strategy going forward.
Because this group of companies has so few members, they operate now in an industry regime, which is in micro-economics class called an "oligopoly" (OL). While there are a number of salient characteristics of an OL industry structure that can be enumerated, for our purposes only three are paramount:
1. There are significant barriers to entry that deter other possible participants from entering.
2. The companies of an OL operate interdependently - the action of one affects all the others. Because they are interdependent, they communicate, or "signal" each other regarding significant initiatives.
3. They emphasize profit growth, and only secondarily revenue growth. Competition among OL members is more often about factors other than price.
In the first article of this series, we could identify each of these elements expressed in quotations from Mark Durcan, the CEO of Micron, and Kipp Bedard, Micron's VP of Investor Relations. Let's now turn to some recent statements from Sanjay Mehrotra, CEO of SanDisk. At a recent Morgan Stanley Investor's Conference, in response to a question regarding industry and SanDisk bit supply growth, he elaborated on the components of the supply increase for his company and for MIPO as a whole. Asked where the bit growth (40%) for the industry was coming from, he said:
"[…] the estimation of bit growth is primarily coming through the technology transitions [… ] In terms of new supply the industry new capacity is expected to be fairly moderate, approximately 10% or so, […] So Cap/EX is primarily being driven by technology transitions [which are] the best way to give you bit growth because they give you the maximum return on investment."
What really gives this answer some punch is SanDisk's guidance in their Q4 conference call that the company plans to grow their bit supply by 25% to 35%, even though industry bit growth is estimated to be in the 40% range. At this point, I highly recommend that you take a moment and actually listen to the exchange Mehrotra is having with the moderator, who, being from the old school, can't quite believe his ears. Note that "new capacity" in this exchange is a code word for "new wafer starts":
Q. Moderator: "And you guys have contributed to the slow growth environment by your decision to focus on technology [to drive growth] and actually predict bit growth that is below the industry as a whole and recoup that on the pricing side. Can you talk about that decision process and when you might have to think about adding new capacity?"
A. Mehrotra: "If you look at over the course of the last few years our bit supply share really has stayed about the same. Having said that, I want to be very clear that SanDisk is not about chasing bit share in the market. Our strategy, is […] through high value solutions to really drive our revenue share in the industry. I believe our revenue share will continue to outperform our bit share…] This strategy is working out very well] because of our prudent focus on managing supply growth. It's really working well for us."
Note the words "high value solutions," which are central to SanDisk's strategy for optimizing profits. Why? If the basis of competition is a "high value solution," one is competing on some other factor than price. But wait, isn't he talking about revenue performance? True, he does say "revenue share," but clearly, if bit supply (a measure of cost) is reduced and revenue is increased, then operating margin goes up, right? Finally, note the phrase, "our prudent focus on managing supply growth." These words are meant for a broader audience than the investors and analysts listening to the call. They are, in fact, a signal to the other MIPO members that SanDisk is going to keep their foot on the brake when it comes to industry capacity growth.
Later on in that same exchange, Mehrotra is pressed about their 3D technology plans:
[…we will be delaying the introduction of our 3D technology and staying with 2D planar for at least two more generations because] "that offers us [the lowest CAP/EX and] the best financial results and I believe we are continuing to do well in this regard."
Is it starting to look like Durcan and Mehrotra are reading from the same memo? The similarities regarding strategy between the two are apparent. One says he's not about chasing "scale"; the other says his company is not about chasing "bit share." Decoded, what they are really doing is introducing the investing community to a new industry paradigm that fundamentally changes the way industry participants approach production and pricing decisions. In this new world, MIPO is in control of industry bit supply, and bit supply is subordinate to, and in fact, is a primary lever for attaining, higher profit.
So SanDisk and Micron are signed up. What about Hynix and Samsung? Before moving on, let's take a look at what SK Hynix thinks about the supply/demand balance. In response to a question from an analyst regarding DRAM supplies:
Q: [ed. - Given the industry consolidation to three companies - Hynix, Micron, and Samsung…] "So given this situation, what do you believe will be the impact on the second quarter of 2013 on 2014 and 2015, so, going forward? And do you have any strategic thoughts regarding this situation?" [Young Park Woori Investment and Securities]
A: "The key would be on how much investment we'll make to expand the capacity. But as you might well know, we are facing physical limits on the technology side. So we believe that in two to three years, the next-generation memory technology will have to arrive. And when that time comes, we'll have to make heavy investments. So, until that time, we'll have to be relatively conservative in terms of making any further investments. So, summary would be this, that for the next two, three years we'll be investing very conservatively. So the DRAM supply and demand situation will be maintained very tightly."
[unidentified company officer - SK Hynix]
In another portion of the call, NAND supply was addressed:
"[We think that] memory suppliers' investment decision will be made more prudently than ever as the shrink technology is approaching the physical limit. Temporary supply/demand imbalance may occur in adjusting the mix of memory products within the limited capacity, but overall will be a fairly stable supply/demand situation."
[Jun Ho Kim - President and Head of Corporate Center]
So that's three of the four members of MIPO signed on to keeping capacity under tight control. All reference the challenges they face technically and financially to continue increasing node density. What about Samsung? They are the largest producer of memory, but Micron claims they are smaller when accounting for the product Samsung consumes internally. This makes perfect sense when you consider that Samsung has almost a third of the smart phone and SSD market share. How likely is it that Samsung would flood the merchant market and drive down the memory costs of its fellow competitors? This seems highly unlikely. That makes four for four of MIPO who have moved from the old world to the new. The landscape of the memory business is changed, at least for the foreseeable future.
What does all this mean for the memory investor? I hope by now you're convinced that the reality of the business, and thus, the motivations of the MIPO companies have changed. They are now in the driver's seat and looking to increase ROI and profitability over time. But how, concretely can that happen? Aren't they going to have to build new FABs and thus risk igniting the capacity glut all over again? The short answer is no, for three reasons:
1. The industry is already anticipating very high cap/ex costs from the move to sub-20nm DRAM and sub 16nm NAND. Adding new FAB construction costs on top of the already high tooling and process costs caused by the 2D to 3D transition is clearly out of the question.
2. Even given the pessimistic forecasts of tech-node driven bit increases over the next five years, the industry has the FABs it needs to drive margins much higher (60-80%) while supplying the high value added solutions that will be demanded of NAND.
3. In the new operating paradigm, the question of supply is axiomatic. In the world of oligopoly, supply is delivered at the point where ROI of the business is optimized. In other words, supply just doesn't happen, supply is in the company's (and industry's) control. It is managed as an output of the oligopoly strategy.
Let's take a look at how the numbers lay out. I'm going to focus on NAND because it both bigger and growing faster than DRAM, and thus, is more relevant to long-term profitability of industry participants. Now, remember that node density increase is also a financial driver of each business and of the technology industry as a whole. Smaller and lengthier node decreases (or "shrinks") mean higher costs for every consumer of memory, or, in this case, a much slower reduction in costs over time. This is a fundamental reality no matter what the supplier's profit goals are. Please bear with me while I get granular on this. Depending on who you believe, the NAND customers consumed roughly 380 exabytes of storage in 2013. That's bytes, so let's convert to bits and we've got 3.04 E+20. Using Micron's numbers (which are very similar to SandDisk's), here's the bit forecast for the next four years:
|Process Node||16||1Y - 3D||3D||3D -3D g2|
Since they have announced they will be supplying very few new wafer starts over this period, we can infer that the "supply" increase is a direct function of the productivity of the shrink they think they can achieve from each process node. Using this data as our baseline, assuming that the industry advances in sync with Micron, the impact on NAND costs and industry production are as follows:
This represents a 60% increase in the overall supply of NAND bits through the period. If MIPO suppliers were to pass on this increase without gross margin increases, customers could expect to see roughly 60% lower price per bit. This is highly unlikely to happen, for reasons we will discuss below.
Here is the Micron forecast for NAND byte consumption over that same period starting from the 2013 Base to 2017.
We see that the application view is that the demand in the market is 172,000,000,000, or 1.7 E+20 in bytes and 1.38 E+21 in bits at the end of this four year period. This represents a 60% increase through the five years. What's our bit supply plan? 1.04E+21 compared to a bit demand of 1.38E+21. In other words, Micron is projecting that the industry in 2017 will be under supplied, with bit demand exceeding supply by 25%.
Unlikely? This can't happen? In the world of oligopolistic market management, this kind of thing happens all the time. Who's going to stop them? Think back to Prince Bandar for a moment. Think he was writing any checks to the big SUV owners who could no longer afford to drive their big SUVs? The genius of capitalistic markets is that consumers adapt to the price/value equation in any good or service. NAND is a perfect example of this because of its almost limitless price elasticity of demand ("PED"). After all, who wouldn't prefer (all else being equal) a flash drive to an HDD drive?
In this case, the memory market will adapt to less rapidly decreasing prices by changing the product mix to offer different solutions. After all, even without the CAFÉ standards, the auto industry would have still been emphasizing smaller, more fuel-efficient vehicles. In the flash business, for example, we may see much slower growth in the client SSD market than might otherwise occur. We might, in fact, predict a really hot reception for the new generation of hybrid drives that our HDD vendors are just now bringing to market. Why? Because the value equation for NAND in that application is less than many others that will be competing for the scarce, and higher priced, bits.
Talked to anybody in the oil industry lately? Ask them the question "how much oil can be produced?" Two things will happen: one, they will look at you like you're an idiot, and two, they will ask you, "at what price per barrel?"
So herein is RULE #1 of the three rules of this "New World":
· TELL ME THE PRICE, I'LL TELL YOU THE DEMAND.
Corollaries to Rule 1:
1. Price is the throttle that MIPO will use to manage the market. Put your foot on the throttle (i.e., decrease prices) and demand will increase. Take your foot off the throttle and demand will decrease.
2. Competition among MIPO members is based on profit and ROI, not on market share, revenue, or any other metric. Highest market cap wins.
3. Long-term contracts are the rule - the throttle, in other words, is governed by the backlog of the supplier. Too much backlog? Raise prices. Too little? Reduce bit supplies.
4. Customer service and "collaborative partnerships" are vital. MIPO will learn that with great power comes great responsibility: to communicate with accuracy and integrity, to offer great value, to provide great service. Customers will learn that with less power comes the need to collaborate to help mold MIPO strategies from a technical and business perspective. Nobody, not the farmer or the goose, wants the golden goose to stop producing, right?
This is a new world indeed, but it's not going to happen overnight. 2014 is the transition year - a year in which customers and MIPO each come to grip with the new realities. Gross margins, especially for Micron, will be steadily increasing during 2014 (starting with a big beat in the Q2 report on April 3). SanDisk will surprise on the upside, beating its GM guidance of 45 to 48 per cent. By 2015, we should really start to see gross margins increasing toward the 60% range by the end of the year and overall industry profitability climbing rapidly. By 2016, the new order is in place and cemented in and, as the bit supply really begins to decelerate with the transition to the 3D and 3D gen 2 nodes coming on line, the industry will more than be able to cover the increased CAP/EX with its greatly improved gross and operating margins.
So what could possibly go wrong with this rosy picture? Plenty, but almost all of the risks are well within MIPO's control. (More on that topic in the next article in the series.) One thing that's not is the world economy. MIPO ultimately can only manage supply and demand within the context of a relatively benign business environment. World War III (Putin invades Canada) - all bets are off. China disintegrates into chaos and rebellion, thereby crashing the world economy? MIPO is in the tank with everybody else, obviously. Outside of real-world economic pain, however, MIPO is set for a long run of improving margins (and oh, by the way, revenues and cash flow), and investors are set for higher multiples, higher prices, and happy, happy days. Gotta love it, this strange new world, this "World Turned Upside Down"!
Oh, almost forgot, what are Rules 2 and 3?
· See Rule 1
· See Rule 2
Disclosure: I am long MU. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.
Additional disclosure: Also long SNDK and am about to initiate call options on MU.