Micron and SanDisk Become the Cheapest AI Stocks in Nasdaq 100—Should You Buy?

According to Benzinga, two of Wall Street’s most favored AI概念股 have quietly slipped into the 'discount zone' of the Nasdaq 100 index, despite recording triple-digit gains year-to-date.

Micron Technology (MU-US) currently has a forward P/E ratio of just 6.2x, meaning investors are paying $6.20 for every $1 of estimated earnings over the next 12 months. This makes it the second-cheapest stock in the Nasdaq 100, behind only Strategy.

Micron has declined 23.3% since July, heading toward its worst monthly performance since June 2022. However, year-to-date, its stock has still surged 208%, having previously climbed nearly 2,000% from its 2025 lows to its June highs.

Meanwhile, SanDisk (SNDK-US) has dropped 35.7% this month, with a 24.7% plunge this week—the worst weekly performance since March 2025. Its current forward P/E ratio stands at approximately 8.1x, and its stock is still up 504% year-to-date.

Memory has always been one of the most cyclical sectors in the semiconductor industry. Building a new wafer fabrication plant typically takes two to three years and requires capital investments of tens of billions of dollars, resulting in a significant lag in new capacity.

As a result, historical patterns show that a sharply declining P/E ratio is often not a 'buy signal' but rather a warning sign. When the market anticipates that corporate earnings are about to enter a downturn, P/E ratios typically contract in tandem.

Micron has experienced this exact scenario before.

In May 2018, Micron’s stock hit a then-record high of around $64, with a P/E ratio of just 4.5x—the company’s most profitable year to date. However, as DRAM prices peaked and began to fall, the stock plummeted 57% by December 2018, and revenues declined 23% the following year.

In other words, the seemingly ultra-low P/E ratio actually reflected the market’s anticipation that earnings were at a cyclical peak.

However, bullish analysts argue that this cycle may break from past patterns.

First, high-bandwidth memory (HBM) is largely locked in through multi-year contracts, with nearly all current capacity already sold out. This contrasts with past downturns, which were typically triggered by sharp declines in spot market prices.

If AI demand represents structural growth rather than a temporary cycle, memory stocks should command higher valuations, not lower P/E ratios.

In fact, Wall Street has consistently backed these stocks during the recent rally. The real question now is whether these seemingly cheap P/E ratios are enough to lure investors back in—or if they simply reflect the market’s early pricing in of future earnings declines.