Micron's Memory Supercycle: Why $1,000 Is the Line Between Boom and Bubble
Memory is the market's favourite argument right now, and Tuesday night gave both sides fresh ammunition. After the close, AMD reported a 97% jump in data-center revenue to $8.5 billion, and Super Micro Computer said quarterly sales nearly quadrupled to $19.95 billion while lifting its annual revenue outlook to as much as $82 billion. Memory stocks climbed in response: Micron rose 2.6% in Tuesday's session to about $855, SanDisk gained 4.8%, and SK hynix added 2.1%. The question hanging over all three is whether this is a genuine supercycle or a peak wearing a costume.
The bull case is unusually concrete. Micron has expanded its portfolio of long-term customer agreements to 26, up from 16 just three months earlier, with contracts now covering more than 35% of projected revenue through 2030, many with minimum pricing commitments. More than 75% of its output for next year is already committed, customers are discussing purchases extending into 2031, and the company expects tighter DRAM and NAND supply conditions through 2027 and 2028. Its next-generation HBM4 memory is largely sold out for 2027 at significantly higher prices, and it is working with Nvidia on customised HBM4E for future AI processors. Micron's own guidance for its fiscal first quarter of 2027 — revenue of $61.5 billion, plus or minus $1.5 billion, at an adjusted gross margin of about 86.25% — reads less like a cyclical peak than a repricing of the whole industry.
The results from its rivals tell the same story. SanDisk's fiscal first-quarter revenue reached $11.03 billion, up 378% from a year earlier, with adjusted earnings of $49.51 per share; data-center revenue jumped 61% sequentially to $4.8 billion, and the company guided second-quarter sales to $12.2–12.6 billion while announcing an eight-for-one stock split. SK hynix reported record third-quarter revenue of 104.62 trillion won and operating profit of 83.33 trillion won, and says it expects shortages to last through 2030, with long-term agreements covering more than half its planned production for 2027 and 2028.
Wall Street has, unusually, almost no dissenters: 28 of 29 analysts covering Micron recommend buying, all 10 covering SK hynix do, and 15 of 18 covering SanDisk do. D.A. Davidson even raised its Micron target to $3,000, arguing supply constraints could persist through 2028.
But the bear case is not theoretical either, and it has already bitten once. Just days ago Micron failed to reclaim $1,000, then reversed sharply toward $840. Intel blamed high memory prices for a PC slowdown it expects to last into 2027 and guided fourth-quarter revenue to $15–16 billion, below expectations. Kioxia, Japan's largest memory maker, forecast December-quarter operating profit of only about 2 trillion yen — barely above the 1.99 trillion yen of the September quarter and below analyst expectations. And Stifel expects DRAM shipment growth to moderate to roughly 15–20% in 2027, which would cap further earnings expansion. The fear is not that memory makers cannot sell their output; it is that their customers eventually cannot afford it.
That is why $1,000 matters more than any analyst target. It is the level Micron just failed to hold, the round number between a market that believes the shortage story and one that has started pricing demand destruction. Reclaiming it before year-end would mean the AMD and Super Micro results did what Intel's warning could not undo: convince traders that AI infrastructure demand overwhelms PC weakness. Failing again would suggest the supercycle's first crack is showing. My read is that the contracted-supply evidence — output committed years out, at higher prices, with minimums — is stronger than the demand-destruction fear, which so far rests on one PC warning and one cautious quarter from Kioxia. But eight weeks is a short window, and a stock that just round-tripped from $1,255 to $840 does not hand back $1,000 easily. I put the odds just above even.