Micron became one of Wall Street’s hottest AI infrastructure trades after its market value briefly moved above Meta and Tesla during Thursday trading. The stock later pulled back to roughly the same range as those two companies, but the market narrative had already shifted: analysts were openly framing Micron as the “next Nvidia.” The driver behind that move was clear, high-bandwidth memory, or HBM, the memory technology tied directly to AI compute demand.
HBM sits at the center of the current AI buildout. According to the source material, each Nvidia H200 and B200 GPU uses 6 to 8 HBM stacks, and a single AI server can consume hundreds of gigabytes of high-bandwidth memory. As AWS, Google Cloud, Microsoft Azure, and AI startups keep expanding compute capacity, global HBM supply has tightened sharply. That imbalance has pushed Micron into the middle of the AI spending cycle.
HBM demand is reshaping how memory suppliers are valued
IDC said the shortage could last through 2027. The impact is no longer limited to AI servers. It has started spilling into consumer electronics as well. The report said Apple recently raised prices for MacBook and iPad products because memory costs had climbed. On Micron’s latest earnings call, the company’s chief executive said Micron is changing its business model at the core, moving away from being a commodity supplier exposed to cycles and toward becoming a strategic partner with revenue supported by long-term supply agreements.
That argument has found support on Wall Street. Citi analyst Christopher Danely raised his price target on Micron to $200, saying AI demand is creating an unprecedented memory supercycle. Evercore ISI analyst Mark Lipacis pointed to Micron’s strategic agreement with Anthropic as a turning point, arguing that direct long-term supply ties between AI model developers and memory manufacturers are unusual in semiconductor history.
The industry’s boom-bust record still hangs over the trade
Even with the AI narrative in full force, the memory sector’s track record remains hard to ignore. Over the past 30 years, the DRAM market has gone through at least five major boom-and-bust cycles. Each time, claims that the cycle had fundamentally changed were eventually challenged by rising capacity and weaker pricing. Samsung and SK hynix are also expanding HBM output, and if supply catches up with demand, pricing pressure could quickly weigh on margins.
Taiwan operations and suppliers are part of the story
Micron’s footprint in Taiwan has also drawn renewed attention. The report identified the Houli A3 plant in Taichung and the Guishan plant in Taoyuan as important bases for advanced DRAM and HBM packaging. Last year, Micron increased purchases of equipment and materials from Taiwanese suppliers, lifting parts of the local memory packaging, substrate, and test interface supply chain.
At the same time, SK hynix has reportedly been preparing a US IPO to raise $29.4 billion for HBM capacity expansion, showing how the AI memory race has moved beyond factory plans and into capital markets. For investors, Micron’s share performance is being watched not just as a company-specific move, but as a live read on demand for AI infrastructure.