I’ve had this in draft for several months. My intent in raising this topic is contrary thinking, and I was hoping to post when Micron’s stock price was over $90 per share. Since that price is more than 50% up from here, and I don’t want to wait, I’m posting this now. Those who have been reading my analysis and commentary know that I invest in Micron using a cyclical strategy. Consolidation has improved the DRAM market since Micron acquired Elpida in 2013, leading to higher peaks in the subsequent two cycles. The new CEO has managed Micron better than the previous management teams. But Micron’s financial performance and stock price are most determined by the cycles of memory pricing.
Pricing and Cash Flow
Micron and its competitors (Samsung and Hynix in DRAM, plus Toshiba/Western Digital in NAND) design memory chips, develop manufacturing processes, and making semiconductor products (from components to sub-systems). These activities are both expensive and capital intensive. Micron spent almost $2.7B on R&D in FY-21, or 10% of revenue. The company spent $10B on capital expenditures in FY-21, which was 80% of their operating cash flow during the year. They plan to spend around $12B in CapEx during FY-22. Memory was a bad business until about ten years ago because DRAM and NAND are commodities and there were too many players. DRAM is now an oligopoly of three players. NAND is still consolidating, most recently with Intel selling their memory business to Hynix. DRAM is now a good business and NAND is okay. But even though the DRAM suppliers, with three players controlling 95% of the market, have been able to keep pricing more stable than in the past, DRAM companies are all still price-takers in a technically and operationally demanding business that is highly capital-intensive. The unpredictability of memory pricing, the investment requirements of the business, and the technical degree of difficulty are why memory company stocks trade at a discount to other big semiconductor companies. Said simply, these companies are required to spend huge sums on capital investments and R&D, to make products that sell for unpredictable prices. This certainty of spending and uncertainty of profitability is the biggest reason Micron is not among the most valuable semiconductor companies.
Degree of Difficulty and Product Differentiation
Sanjay Mehrotra, the CEO of Micron, spends much of his opening statements during quarterly analyst calls talking about Micron’s new product offerings. I mostly tune out on these comments, and on related questions from analysts, because this information doesn’t matter to investors. New product announcements, fastest chip, etc.; this is all price of entry for memory makers. The part of semiconductors that makes electronic products amazing for consumers – the steady increase in performance one gets for a similar price over time – is part of why the business is so challenging to be in. Semiconductor makers need to keep designing and churning out more and more advanced products every year – more bits and bytes per phone, more flops of processing power, etc. – knowing the price per unit performance will steadily decline. The internal workings to make this happen within each company are highly complex and have evolved and been refined over decades. Memory companies need to execute extremely well day in and day out just to keep up. That’s why, when Sanjay touts that Micron has the fastest graphics chip on the market, I assign no value to that as an investor. They need to keep making those improvements just to keep up. The degree of difficulty in memory is high.
The argument against this nihilistic view is that having the best products makes some company revenue differentiated. With three DRAM companies left, there is not much room for one of them to get a share of higher margin markets such that their earnings are significantly affected. All three players need to participate in all segments. Micron can’t elect to not make PC DRAM anymore, because they don’t like the margins there. They have too many bits to do that. There is also the fact that different segments have different quality requirements. Memory makers need consumer segments (PCs, some mobile) to sell their lower-quality bits into. Also, higher margin products are also higher cost to produce, so the higher ASPs in automotive, graphics, industrial, etc., don’t all fall through into gross margin. Finally, when a segment’s margins get too high relative to the others for a given memory type, the other two players will come in to compete it away. Micron was dominant in automotive for a few years, then Samsung entered in earnest and is eroding away the margin difference in that segment.
One closing comment I have is on high-value solutions. High-value solutions are selling NAND bits in an SSD instead of selling NAND as loose components. This is an area where, depending on their strategy, memory companies can make a material difference in their margins. It is much better business to sell enterprise SSDs than to sell loose NAND components. Similarly, one would rather sell memory modules instead of DRAM components. This is an area where Micron has lagged Samsung, Hynix, Intel, and WD/Toshiba historically. The new management team, having come from a product company that didn’t do their own manufacturing, has mostly closed this gap. The run-up in Micron’s stock price in the last five years is, in part, from this structural change in how Micron sells the bits it makes.
I have had mostly praise for Micron’s new executive team. The changeover began in 2017 when Sanjay Mehrotra was hired to replace retiring CEO Mark Durcan. Mark had tried to retire years earlier but had stayed on as CEO when Steve Appleton died in 2012. Micron previously survived the brutality of the DRAM business with superior engineering and technology, and by being a consolidator. Prior to 2017, the company wasn’t a good operator, and the management team weren’t good businesspeople. Mediocre management meant Micron’s cost structure was the worst in the industry. When Sanjay came in, he brought along a group of executives from SanDisk, most notably the new Chief Business Officer, Sumit Sadana, and the VP of Operations, Manish Bhatia. He also replaced the CFO with Dave Zinsner, previously the CFO of Analog Devices. The other important move he made was to retain Scott DeBoer as VP of R&D. Micron had been behind in DRAM process technology for twenty years and had not been able to close the gap with Samsung and Hynix, despite several initiatives attempting to do so. Because of this, many thought DeBoer should go. But Sanjay kept him, and DeBoer successfully led Sanjay’s most important initiative since joining Micron, to regain leadership in DRAM process technology. This has been by far the “new Micron’s” most important accomplishment because process technology determines Micron’s COGS. The new management team also cleaned up the company’s balance sheet, thanks in no small part to taking over just as a massive upturn in DRAM pricing was beginning. Sanjay and his executives claim a lot of credit for turning Micron around, and they deserve some for executing well on technology and business matters, but they also joined at the best possible time. The success Micron has had in DRAM, which is 70% of their business, has come from a favorable pricing environment and catching up on technology.
Besides the initiative to close the technology gap with competitors, the other things Mehrotra and his team are citing now are more from being in the right place at the right time. The area that SanDisk was good at was consumer products based on NAND memory. The most important market in NAND today is data center SSDs, an area Micron has continued to struggle in. Sadana, the CBO, has touted the increase in the percent of Micron’s NAND bits that go into “high value solutions.” While this is true, it has largely been driven by Micron selling more NAND into multi-chip packages (MCPs) for mobile devices. They can do this because Micron makes DRAM, which is the other “chip” in the MCP. Thus, some of this success is a product of Micron having a DRAM business rather than any system expertise from the new managers. They have recently released enterprise SSDs that seem to be competitive, so there may yet be hope that the executives, who came over from an SSD company, can finally lead Micron to success in SSDs.
Mehrotra and his new management team took Micron up two levels of performance. During their tenure, Micron has caught up on process technology and now leads in DRAM. The company has maintained the lead in NAND they had when Mehrotra joined, no mean feat, and not to be discounted. The new executive team also improved Micron’s mix of sales to get better pricing and margins by selling more bits into systems. They aren’t at the level of Intel or Samsung, who dominate in the highest-value SSDs, but their improvement has been meaningful. They have also brought more discipline to cost management and other business operations. Mehrotra and his team are professional managers in a way that Micron’s previous leadership teams never were. Micron before Mehrotra succeeded because of technical innovation and manufacturing excellence.
But I see a ceiling for Micron under Mehrotra. They are good managers and poor leaders. I make this judgement based on having direct experience with, and second-hand knowledge of, many of them since they arrived. Mehrotra is an arrogant jerk that most people don’t like working for. Evidence of this was seen recently in how Dave Zinsner, the previous CFO, departed. Once he got the job offer from Intel to be their CFO, he notified Mehrotra and the board that he would leave later that week and started at Intel the following Monday. He couldn’t get out of Micron fast enough and didn’t care about leaving Mehrotra in a difficult position. If Mehrotra and Zinsner had a good relationship, he would have stayed on for at least two or three weeks for a more orderly transition. Zinsner wanted to get away from Mehrotra and Micron so badly that he was willing to send a message to the Board and others by leaving the way he did. The two people who like working for Sanjay, Sadana and Bhatia, like him because they owe their careers to him.
Mehrotra’s arrogance, and the arrogance of his management team, are concerning for me as an investor. I don’t think companies become great and stay great without humility, something Mehrotra and his team are greatly lacking. Those on Mehrotra’s staff often behave like him and thus don’t engender much loyalty from people. Sadana is a pessimist, a terrible characteristic for the CBO to have. Nobody ever accomplished anything great without optimism. The minimal share count reduction during the 2018-20 down cycle illustrates the poor capital allocation skills of the CEO and the CBO. They missed a major opportunity to shift Micron’s capital structure. But Sadana the pessimist was too worried about what might happen in the market and convinced the CEO to hoard too much cash and take on too little debt.
Micron’s executive team today manages with a “next man up” mentality. They don’t develop employees because they believe they can just replace them. They think it is a privilege to work for Micron and that people will line up to do so. This has been true for many roles, especially those the executives have the most interaction with. This philosophy stifles innovation and discretionary effort, making employees view their job as transactional rather than feeling part of something important. I don’t think Micron will break out of their third-place niche in memory products under these managers. They will continue to execute well and maintain their market share in DRAM and NAND. But Micron won’t accomplish another transformation. The change to “the new Micron” over the last five years was about three things: regaining technology leadership, cleaning up the balance sheet, and tightening up operations. With those three goals accomplished, what is next? The current executive team is excellent at executing what obviously needs to be done (everyone in the industry knew the most important move for Micron was to catch up on process technology). Improving the balance sheet just needed a decent CFO and an upturn in memory pricing. Operational improvements are hard and require great discipline, but they are not a breakthrough.
Adding value to the company from these three improvements was a one-time event. Mehrotra has created a culture of conformity, where executives compete to please him and conform to his ideas. Controversial, innovative products and strategies will not emerge in an environment like that. The two most powerful people in the company, the CBO and the VP of Operations, actively compete with each other to be the next CEO, rather than functioning as a team. The new CFO sees Micron as a steppingstone to a better next role and will put up with Mehrotra for career growth. Great companies have great leadership teams. Mehrotra is a competent CEO with a competent leadership team. If Mehrotra retires or otherwise moves on and a new executives comes in, Micron may be able to transform again and break out of being the third-best memory company.
The Memory Market
Micron has a market share of around 25% in DRAM and 13% in NAND, for a combined share in memory of about 20%. The DRAM side is fully consolidated with ~96% controlled by Samsung, Hynix, and Micron. NAND is down to Samsung, the Toshiba-WD dyad, the currently-merging Hynix-Intel entity, and Micron. Micron is the sub-scale player left in NAND and have a higher cost burden because of this. On the plus side, NAND growth is still earlier in its life cycle than DRAM, so Micron could organically grow their market share over the next few years to be more competitive. Also in their favor is the financial stress Toshiba is under, which makes them less able to invest, reducing their ability to protect market share. Finally, NAND is more threatened by China’s attempts to enter memory. Those who have read my China post know I think the indigenous memory companies there will ultimately fail to become significant players, but they may cause damage before then in the form of oversupply. In my view, this is much more likely to happen to NAND than DRAM. Absent nibbling away a few points of market share in NAND, Micron is stuck with around 20% of the total memory market. If they try to take share in DRAM, they risk oversupplying the market and hurting everyone. They missed their chance to buy Intel’s memory operations and are now stuck in fourth place in NAND.
Micron is a price-taker in a highly capital-intensive industry. While the barriers to entry are high and growing every day, the same factors that keep out new competitors – the process technology treadmill, long product development cycles, difficulty of manufacturing, operational complexity, stringent quality requirements, hundreds of SKUs – also make it hard to stay competitive. A small number of slip-ups in development and execution can take years to recover from. Because memory is still a commodity, with pricing dependent on supply and demand, it is highly risky for Micron to try and gain market share. Thus, they are stuck at about 20% share in a market that is growing high single digits annually. That is not bad growth, but it is low considering how hard and expensive it is to stay competitive. Micron’s management team lacks the skills and culture to lead the company through another transformation, one that will break the value of the company out of the box of DRAM and NAND memory. With the transition to “the New Micron” complete, I think the best way to invest in Micron in the future is in the cycles of memory pricing. I’ll post a separate analysis to update my views on how to value Micron through the cycle.
-Smooth Hughes (long MU)