Sir John Templeton spent decades watching investors talk themselves into believing history no longer applied to them. His most quoted line, “the four most dangerous words in investing are: ‘this time it’s different,'” has outlived every bubble it warned against. Analysts at Franklin Templeton, the firm bearing his name, are now pointing that same phrase directly at the AI memory chip trade.
The target is hard to miss. Micron Technology and SK Hynix have each climbed to roughly $1 trillion in market capitalization as of May 2026, fueled almost entirely by surging demand for high-bandwidth memory chips used in AI training and inference. Add Samsung to that pile, and the combined valuation of the three dominant memory players sits at approximately $4.1 trillion.
The AI chip boom in plain terms
High-bandwidth memory, or HBM, is the specialized chip stack that sits alongside AI processors and feeds them data fast enough to keep up with the math. Without it, even the most advanced AI accelerators run into a bottleneck.
Demand for HBM has been so intense that SK Hynix and Micron’s combined output is essentially spoken for through the end of 2026, locked up under long-term supply contracts.
The concern from Franklin Templeton analysts is not with the current demand. It is with what comes after, and with how much of that future is already priced into valuations that have moved at a pace that memory chip stocks historically do not sustain.
The cycle nobody wants to remember
Memory semiconductors have a well-documented rhythm. Demand spikes, manufacturers race to build capacity, supply eventually outpaces demand, and prices fall hard. It happened in DRAM markets repeatedly over the past three decades. It happened in NAND flash.
Capital expenditure across the semiconductor industry has been rising sharply to meet AI-driven demand. At some point, if AI infrastructure buildout slows, or if new entrants add capacity faster than the market absorbs it, the supply-demand balance that currently justifies $1 trillion valuations could shift quickly.
SK Hynix carries a particular concentration risk that goes beyond its own balance sheet. The company represents a significant weighting in Korean equity indices, which means investors with broad exposure to Korean markets are carrying more semiconductor cyclicality than they might realize. Franklin Templeton analysts have flagged this specifically as a structural concern for portfolio construction, not just a single-stock risk.
What this means for risk assets broadly
There is no direct crypto token tied to HBM chip production. But sentiment in crypto markets correlates meaningfully with broader risk appetite. When large-cap tech and semiconductor names are doing well, institutional appetite for risk-on exposure tends to stay elevated, and Bitcoin along with other digital assets benefit from that environment.
The $4.1 trillion in combined semiconductor market cap represents a concentration of investor confidence in one thesis: that AI buildout continues at its current pace indefinitely. For now, the more immediate read for crypto investors is macro: a sector that has absorbed trillions of dollars in market cap gains on a single demand thesis is being flagged by Franklin Templeton analysts as carrying cycle risk. The specific word from Franklin Templeton is not “sell.” It is “remember how this has ended before.”