Micron Technology (MU +0.87%) is one of the world's three top suppliers of memory chips, which play a critical role in the artificial intelligence (AI) hardware stacks in data centers, computers, smartphones, and even cars. There is a worldwide shortage of memory right now, which allows the manufacturers to dictate prices. For Micron, this resulted in a staggering 1,368% year-over-year increase in earnings to $24.67 per share during its most recently reported quarter.
A company growing at such a blistering pace would normally be expected to command a sky-high valuation as investors pile into its stock to get ahead of future potential returns. And investors have bid the stock up: Micron is sitting on a 12-month gain of around 640% -- but it's actually still trading at a steep discount to the S&P 500 (^GSPC -0.32%) and Nasdaq-100 indexes by one traditional valuation metric.
Normally, I would consider a stock like Micron to be a bargain at the current price. But here's why I'm not a buyer right now.
Image source: Getty Images.
Micron is unquestionably cheap at first glance
During its fiscal 2026 third quarter (which ended on May 28), Micron generated a record $41.4 billion in revenue -- a 364% increase from the prior-year period. That result was driven by triple-digit percentage growth across all four of its business segments:
|
Segment |
Fiscal Q3 Revenue |
Revenue Growth (YOY) |
|---|---|---|
|
Cloud memory |
$13.7 billion |
307% |
|
Core data center |
$11.5 billion |
653% |
|
Mobile and client |
$11.5 billion |
254% |
|
Automotive and embedded |
$4.6 billion |
311% |
Data source: Micron Technology. YOY = year over year.
Cloud memory is the category that includes Micron's sales of its high bandwidth memory (HBM) for data centers, where it sits alongside the graphics processing units (GPUs) supplied by chipmakers like Nvidia. HBM stores data in a ready state for GPUs so that they can access it rapidly, helping to maximize processing speeds. That's particularly valuable in intense AI training and inference workloads.
Suppliers like Micron have been reducing their production of other types of memory and reallocating that capacity to boost their output of HBM because demand for it is so strong.
Micron has now generated earnings of $44.23 per share over the last four quarters, placing its stock at a price-to-earnings (P/E) ratio of just 19.8. That's cheaper than both the S&P 500 and the Nasdaq-100, which have P/E ratios of 25.2 and 32.6, respectively.
But Micron's blockbuster financial performance is widely expected to continue. The average forecast among Wall Street analysts covering the company (provided by Yahoo Finance) suggests that its earnings will surge to $155.56 in its fiscal 2027, which begins in September. That gives the stock a forward P/E of just 5.6, which would normally constitute an irresistible bargain in my book.
Valuation isn't everything in this situation
The semiconductor industry -- and particularly the memory segment -- has historically been extremely cyclical. In the past, companies would build data centers and upgrade them only once every few years, resulting in lumpy revenues for chipmakers. The AI boom has condensed the upgrade cycle to 12 months or less because companies like Nvidia and Micron keep bringing out faster chips to meet the market's insatiable demand for computing power.
NASDAQ: MU
Key Data Points
But this can't go on forever. The Financial Times reports that Amazon, Alphabet, Meta Platforms, and Microsoft have spent a combined $1.1 trillion on AI infrastructure since 2023, and they are still increasing their annual capital expenditures. That kind of spending makes economic sense only if there is a tangible return, but it appears that the end-users of AI are starting to feel the pinch financially.
Alphabet CEO Sundar Pichai recently said he is fielding complaints from Google Cloud customers about the rising cost of deploying AI. Moreover, a recent price increase by Anthropic for the use of its AI products caused Uber Technologies to blow through its entire 2026 AI budget in just four months. As a result, the company's chief operating officer said it's getting harder to justify the current rate of spending.
Uber has now imposed limits on AI usage for its employees, as have other large companies including Walmart, AT&T, and Amazon. If infrastructure costs keep rising, AI companies will have to continue hiking prices, and this will cause even more of their customers to watch their spending to prevent budget blowouts.
In my opinion, this explains why investors aren't piling into Micron stock despite its low P/E ratio. Even though the AI boom has distorted the cyclicality of the semiconductor industry, it's almost certainly a temporary phenomenon. AI infrastructure spending will eventually slow down. Plus, since Micron and its competitors are rapidly building more chip manufacturing capacity, they are also likely to surrender much of their pricing power in the future as new production comes online and helps ease the shortage -- or even creates a supply glut.
Simply put, it's possible that Micron's earnings could start shrinking in a couple of years as the supply-demand imbalance in the memory market is resolved. That would make its stock more expensive on a forward basis than it currently appears to be. As a result, I'm not buying it right now.
