Micron Technology, Inc. (NASDAQ: MU) is a leading manufacturer of memory and storage chips based in the United States After going through several decades of competition and economic cycles, MU becomes a national name representing the American semiconductor industry to testify before the Congress on chip manufacturing with Intel. This stock is also heavily invested by smart investors such as Li Lu and Mohnish Pabrai. However, stock prices have historically been very volatile and subject to many external disturbances. I think MU is an excellent long-term investment opportunity given its quality and the inevitable digitization of society. Here I would like to share some of my thoughts (especially NAND fabrication).
An industry requiring large scales and investment
The memory industry has gone through a consolidation. All capabilities have narrowed down to five big players: Toshiba (OTCPK: TOSBF, OTCPK: TOSYY), Western Digital (WDC), Micron, Samsung (OTCPK: SSNLF, OTCPK: SSNNF) and Hynix. The economies of scale of these five are so good that smaller players have no way to compete with them. You have to make 100,000 wafers per month in your factory (costing billions), or you keep losing money. The market is also able to consume new technologies much faster. In NAND builds you have 64 through 96 through 128 and then the current 176 layers. New products are constantly changing the dynamics of the industry. It is extremely difficult for an entrant to disrupt the dominance of current players.
The exciting prospects of NAND
MU is one of the leaders in NAND technology. It seriously challenged Samsung and SK Hynix in the latest 176-layer high-density NAND competition. Although its market shares are still lower than those of the two Asian competitors, it is already leading in the field of technological development, as the CEO said during a recent conference call:
Our 1-alpha DRAM and 176-layer NAND ramps are several quarters ahead of the industry and are progressing well as we continue to qualify new products that use these nodes.
NAND memory (3D chip) is in high demand, with a long-term CAGR within 20 seconds. Since NAND needs no power, is cost-effective per byte with high storage capacity, and is easily replaceable, it’s the perfect option for mobile devices, data centers, automobiles, and computers. ‘Internet of Things (“IOT”). This is absolutely one of the essential components of AI adoption.
On-chip CAPEX manufacturing may not turn into profits
Moving from 2D to 3D (NAND) will lead to greater bit growth (2x for 32 layers). But it’s not free. You have intense capital requirements due to additional etching and lithography deposition processes. If a 28 nanometer 2D chip requires 50 layers of masking, transitioning to 3D definitely requires more masking layers, and each layer requires more cycle time and more lab. If the average processing time of each layer increases, the entire production cycle requires more days. Therefore, CAPEX can add capacity and complexity, but can also increase production days. Not to mention other issues such as wear reliability curves where tools drift and degrade. The cost of inspection and metrology (provided by KLA Corporation) will also increase proportionally. So the risk here is how MU ensures its technology is good enough to ensure that CAPEX returns are net and net profitable after adding manufacturing capacity. In other words, CAPEX is not wasted.
When it comes to layer stacking and manufacturing efficiency control capabilities, MU seems to be ahead of the competition so far. Here is what an engineering professor at Seoul National University said:
When talking about chip technology, higher yields are a key criterion. But when it comes to stackability, you can say Micron is ahead of Samsung and other larger rivals.
Semiconductor vendors are highly specialized
Due to the complexity and specialization of chip manufacturing, the supply chain is very thin and long. You may have 700-1000 steps in your processes and each step may only have 1-3 suppliers that qualify your requirements. For example, the 3-4 companies in Japan produce 90% of the photoresists, 90% of the fluorinated polyimides and 70% of the etching gases in the world. These are essential materials with almost no alternatives. As demand from the semiconductor industry continues to grow, a few production disruptions will lead to global shortages. More and more earthquakes, fires and storms have grabbed the headlines and become black swan events for semiconductor manufacturers like MU.
Competitive landscapes are much cleaner
Memory manufacturing is known to be a commodity-like business and highly cyclical. Analyzing the timing of the different stages of the industry cycle is often key for investors. The dominant player here is undoubtedly Samsung, with the support of the entire Korean country. But companies will deliberately source from two or more suppliers to keep Samsung’s prices fair. For most memory products, price is key because people buy when a minimum set of requirements are met. This leads to huge fluctuations in unit prices.
However, the Big Five will likely maintain their position in the market for the long term. Demand becomes more stable as DRAM spending will eventually correlate to number of servers and NAND will correlate to number of mobile devices and deployment of AI.
There are concerns about huge Chinese investments in semiconductors and partnerships with American companies. With the development of US-China conflicts in recent years, China is probably out of the global competition. I think the industry has been consolidated enough that the cyclical volatility is not as big as it used to be. It’s more like an oligopolistic market where companies focus on stabilizing profits, not cutting costs and squeezing margins. This should be true for other specialty vendors such as Lam Research, ASML and Tokyo Electron.
At the end of the line
With a market capitalization of $67 billion, MU has repurchased $4.3 billion in shares and invested $41.6 billion in CAPEX over the past five years. It also has a five-year average of $7.6 billion in operating revenue and $10 billion in cash / $22 billion in current assets on its balance sheet. As a stock working in a growing industry (10% + CAGR), there is no reason to believe this is an expensive stock. The recent capex investment slowdown plan has highlighted the flexibility MU has to stay ahead of the competition. MU already have a big enough gap to defend their position. While there will always be short-term uncertainties, the long-term return potential is very obvious to me.