Sentiment Shift. Hi everyone. Narratives can change fast.
Not long ago, investors were despondent as stocks related to artificial intelligence pulled back on bubble fears and worries about potential overcapacity. FUD (fear, uncertainty, and doubt) ruled the day.
A few weeks later, traders seem to be jumping over themselves to buy the riskiest, highest-beta AI momentum plays: commodity memory stocks. Memory chip maker Micron Technology and flash memory company Sandisk are both up more than 40% over the past month. FOMO (fear of missing out) is now the word.
Investors seem to be manufacturing new rationales to become more optimistic about memory stocks. But as sentiment shifts back and forth, the reality is the underlying fundamentals haven't changed much over the past few weeks. The AI trade has remained sound as companies continue to invest in data center capacity, but it isn't suddenly better.
It's human nature to want to chase the stocks up the most and sell after large pullbacks, but letting emotions drive decisions is a recipe for underperformance. Be careful not to convince yourself to trade stocks based on price action. There is wisdom from Hollywood. In the movie Wall Street, stockbroker Bud Fox told his client: "You once told me, don't get emotional about stock, Gordon. Don't."
Micron is a leader in the markets for dynamic random-access memory $(DRAM)$, which is used in desktop computers and servers and for the flash memory found in smartphones and solid-state hard drives. It has also become a key supplier of high-bandwidth memory (HBM) for AI servers. Sandisk's business is nearly all flash memory.
While it is true that memory prices are surging and major manufacturers haven't significantly increased production or capital expenditures, history shows commodity chip suppliers always overbuild capacity at some point in the semiconductor cycle.
Unlike the proprietary AI chips made by Nvidia, memory chips are commodities, meaning they can be easily replaced. There is a higher degree of disruption risk. It may be a year from now or later, but there is real potential for Chinese companies to expand aggressively into the memory chip and flash memory space.
HBM memory is proprietary and more difficult to make, but commodity-oriented DRAM is still likely to dominate Micron's sales for years.
The exact timing of a memory bust is difficult to predict, but it is inevitable. When that happens, the memory stocks will collapse.
There is a better way to invest in the chip sector. I've written repeatedly about the investment philosophy of TCI Fund Management founder Christopher Hohn, one of the most successful investors over the last two decades. His investment philosophy takes a long-term view by finding companies with sustainable moats that provide an irreplaceable service tied to high switching costs.
Commodity memory chips don't fit this framework, but chip makers that have defensible competitive positions do. Nvidia's moat, for example, comes from its CUDA software, which provides AI start-ups and enterprises unrivaled stability as customers have spent over a decade ironing out software issues, making it far less likely to get disrupted by future Chinese competitors.
Unless you are a nimble trader and memory supply-chain expert, it's better to avoid commodity names and follow Hohn: Stick to the stocks with the best long-term fundamentals.