The numbers were unambiguous. Micron posted adjusted earnings of $33.42 a share, well above the $31.16 forecast, on revenue of $54.23 billion against an expected $50.45 billion. It was the sixth consecutive quarter of record revenue, with DRAM alone generating $39.8 billion, up 343% year over year. Then the stock slipped.
The reaction was not irrational. What unsettled investors is straightforward: net capital expenditures were $10.77 billion in the quarter and $27.37 billion for the full fiscal year. And that is the floor, not the ceiling. Micron guided for first-half fiscal 2027 capex of approximately $25 billion, with higher spending expected in the second half. The investment committee question is whether that spending locks in a structural shift or plants the seed of the next glut.
The Bull Case: AI Has Changed the Demand Equation
The strongest argument for the bulls is that the old memory cycle comparison simply does not fit the current supply picture. CEO Sanjay Mehrotra told analysts that industry demand has strengthened, and Micron expects supply-demand conditions in fiscal 2027 and 2028 to be much tighter than in 2026. Management backed that claim with contractual evidence: Reuters reported that customer financial commitments under Micron’s long-term supply agreements rose to $32 billion from $22 billion in June. Micron also said more than 75% of its 2027 output is already committed by customers, a figure it later clarified includes both SCA and non-SCA customers.
The data center business reinforces the point. Data center SSD revenue reached nearly $10 billion in fiscal Q4, more than ten times the year-ago quarter. The Core Data Center business unit posted $18.0 billion in quarterly revenue with gross margins of 90% and operating margins of 85%. Those are not commodity-cycle margins. Micron has also said that HBM agreements are now in place for the vast majority of calendar 2027 HBM bit supply at significantly higher year-over-year prices.
Management has said demand for memory and storage remains strong, with supply conditions expected to stay tight in fiscal 2027 and 2028. SK Hynix’s chief executive told Reuters in July 2026 that 2027 could be the worst year in the industry’s history from a supply perspective.
The Bear Case: Familiar Pattern, Bigger Numbers
The concern is not that AI demand evaporates. It is timing and coordination. Deloitte has argued that memory makers’ higher spending does not translate into immediate supply, because building and ramping leading-edge fabs can take years. Deloitte also projected that Samsung, SK Hynix, and Micron’s combined capital expenditure could reach $146 billion in 2027, about 3.4 times the roughly $43 billion the three spent combined in 2024.
Past episodes where Micron, Samsung, and SK Hynix all expanded simultaneously have tended to sow the next downturn roughly two to three years out, as new wafer capacity converts today’s tightness into oversupply. The most recent example came in 2022 and 2023, when the industry swung into a downcycle and Micron and SK Hynix reported heavy losses as demand cooled and pricing fell. The question is whether HBM’s structural complexity and AI’s appetite make this time genuinely different, or just feel that way in the middle of the boom.
What the Investment Committee Is Actually Debating
The overlooked implication here is the free cash flow trajectory. For fiscal 2026, Micron’s operating cash flow reached $89.68 billion, and net capex totaled $27.37 billion, resulting in adjusted free cash flow of $62.31 billion. That conversion looks extraordinary. But with fiscal 2027 capex expected to rise from fiscal 2026 levels, and revenue growth inevitably decelerating from fiscal 2026’s surge, the free cash flow margin compresses before it expands again. The question is not whether Micron is profitable. It is whether the spending intensity required to stay competitive in HBM changes the return profile that made the stock worth owning in the first place.
Deloitte has argued that meaningful new supply is not expected to come online quickly, constrained by construction timelines and the time required to ramp leading-edge manufacturing. That argues the near-term shortage is real. What happens after the next wave of capacity arrives is the question nobody has a confident answer to yet.
Stocks to Watch
- Micron (MU): The primary beneficiary of the current HBM shortage, but the one absorbing the full capex burden. Free cash flow trajectory in fiscal 2027 is the key variable to monitor.
- SK Hynix: Arguably the cleanest way to own the AI memory shortage, with management expecting severe supply tightness into 2027.
- Samsung: A direct beneficiary across HBM, server DRAM, and NAND as Micron’s results confirm the memory market is tight.
- Nvidia (NVDA): The accelerator platform driving HBM demand. If hyperscaler AI spending slows, it is the first signal that Micron’s forward commitments face pressure.
- Western Digital (WDC): Exposed to NAND pricing dynamics, where Micron said fiscal Q4 NAND revenue rose 526% year over year and prices increased about 30% on tight industry conditions. Any supply normalization hits NAND first.
