A deepening supply deficit in high bandwidth memory and NAND flash is reshaping the semiconductor landscape, with implications that extend well beyond traditional markets.
Two chipmakers have quietly become the hottest trades on Wall Street this year, and neither of them is Nvidia. Micron Technology and SanDisk have posted gains north of 200% in recent months, riding a wave of insatiable demand for the memory chips that power AI data centers.
Micron’s stock has climbed more than 550% over the past year, pushing its market capitalization toward $600B. SanDisk, meanwhile, has been even more dramatic, with shares surging over 3,000% year-on-year and its market cap climbing above $157B. Both companies now rank among the top performers in the S&P 500.
The AI memory supercycle is real, and it’s just getting started #
Every large language model, every GPU cluster, every AI training run needs memory, and lots of it. The specific bottleneck is High Bandwidth Memory (HBM) and NAND flash, two categories of chips that sit alongside processors in AI data centers and determine how fast those processors can actually work.
Morgan Stanley forecasts this AI-driven memory shortage will persist for another 2-3 years, with no quick fix to the supply deficit in sight.
Citi is equally bullish on the space, projecting that SanDisk may offer another 40% upside over the next year, fueled by a rebound in NAND pricing. After years of oversupply that cratered NAND prices, the pendulum has swung hard in the other direction.
The crypto angle: decentralized compute and storage tokens #
For crypto investors, the AI memory shortage is a signal about where capital is flowing and what infrastructure layers are becoming scarce. Projects focused on distributed compute, like Render Network and Akash, and decentralized storage protocols, like Filecoin and Arweave, stand to benefit from a narrative that positions AI infrastructure as scarce and valuable. When traditional memory and compute resources are in shortage, the economic case for decentralized alternatives strengthens. Even a modest spillover of demand from centralized to decentralized infrastructure could move token prices meaningfully, given the relatively small market caps of most AI-adjacent crypto projects compared to a $600B Micron.
The AI infrastructure narrative has historically been one of the strongest drivers of crypto market sentiment. When Nvidia rallied in 2023, AI tokens followed. When Microsoft increased its AI capex guidance, decentralized compute tokens caught a bid. Micron and SanDisk’s parabolic moves reinforce the broader story that AI demand is accelerating.
The risk is that narrative and fundamentals diverge. Most decentralized compute and storage protocols generate minimal revenue relative to their token valuations. If the AI memory shortage resolves faster than Morgan Stanley’s 2-3 year forecast, these tokens would likely give back gains faster than the underlying chipmakers.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our