Surging demand for DRAM and high-bandwidth memory is pushing prices, margins and earnings expectations higher. Memory stocks — Micron Technology, Inc. (MU), Sandisk Corporation (SNDK) and SK hynix Inc. (SKHY) — are some of the biggest beneficiaries of the AI spending boom.

Today's sky-high profits, however, aren't a new normal, according to Allen Bond, Managing Director and Portfolio Manager at Jensen Investment Management.

"Current supply constraints and AI-related demand have improved pricing and profitability," Bond told MarketDash in an email. But he warned that memory businesses remain highly cyclical, making today's earnings potentially misleading when investors try to value them for the years ahead.

AI Is Rewriting Memory Demand

The AI infrastructure boom has made memory a critical part of the data-center buildout. High-bandwidth memory, or HBM, is used alongside AI accelerators to move data faster and with lower power consumption, while conventional DRAM and NAND also benefit as AI workloads expand.

The result has been a sharp increase in memory prices and earnings expectations. Micron, for example, reported record fiscal third-quarter revenue of $41.46 billion in June and guided for roughly $50 billion in fourth-quarter revenue. Its fiscal fourth-quarter results are due Sept. 30.

S&P Global said consensus estimates for Micron's DRAM revenue had climbed substantially, with fourth-quarter DRAM revenue projected at $38.3 billion and gross profit estimates rising sharply as pricing improved.

That momentum has helped send memory stocks sharply higher, but Bond sees a familiar risk underneath the AI boom.

The Earnings Cycle Matters

Bond's concern isn't that AI demand is fake. It is that investors could extrapolate unusually strong pricing and profitability too far into the future.

"These businesses remain exposed to significant cyclicality in both pricing and demand," he said, "making today's elevated earnings a potentially poor measure of normalized economics."

That distinction matters because memory companies can look exceptionally cheap when earnings are near a cyclical peak. A low price-to-earnings multiple may simply reflect temporarily inflated profits rather than a permanently higher earnings base.

The current cycle does have unusual characteristics. S&P Global said Micron's management expects industry conditions to remain tight beyond 2027, with AI-driven workloads continuing to increase memory requirements.

But Bond's warning focuses on what happens when supply eventually catches up with demand.

Investment Takeaway

AI has clearly changed the memory demand equation, but that doesn't eliminate the industry's historical cyclicality.

Investors watching Micron, SanDisk, SK Hynix and other memory names should look beyond the next earnings beat and ask whether today's pricing, margins and demand can persist as capacity expands. The AI boom may be real; the question is how much of today's memory earnings are durable.