Micron's FY2025 revenue hit $37.4B (+48.9% YoY) with gross margins at 39.8% and net margins at 22.8%, putting it on track to be the third-most-profitable U.S. company behind only Nvidia and Google. The AI memory supercycle is compressing the gap between commodity DRAM economics and high-margin HBM pricing, creating a re-rating setup if margins continue expanding.
Micron's FY2025 revenue hit $37.4B (+48.9% YoY) with gross margins at 39.8% and net margins at 22.8%, putting it on track to be the third-most-profitable U.S. company behind only Nvidia and Google.
MU's near-50% revenue growth and 39.8% gross margins raise the question of whether the market will finally re-rate memory as a structural AI infrastructure play or whether HBM pricing will revert as Samsung and SK Hynix ramp capacity.
Samsung and SK Hynix HBM ramp compresses pricing faster than expected, or hyperscaler capex guidance steps down — either collapses the margin expansion thesis and re-exposes the cyclical multiple.
CoverageSource: MarketWatch · Published here SUN, JUN 28 · 10:00 AM ET · the only report in this recordHow this is decided →
Micron Technology is posting financials that would have looked unthinkable two years ago, when the memory cycle was at its nadir. FY2025 revenue of $37.4B represents a near-50% year-over-year jump, driven by surging demand for High Bandwidth Memory (HBM) from hyperscalers and AI accelerator builders who are paying premium prices to secure supply. At 39.8% gross margins and $7.59 diluted EPS, Micron is operating at a level of profitability that now places it in the same sentence as Nvidia (71.1% gross margins) and Alphabet (32.8% net margins).
The headline framing matters because memory has historically been a low-multiple, cyclical business — investors have long penalized Micron relative to logic-chip peers for exactly that reason. If AI demand structurally floors HBM pricing and sustains volume, the bull case is that Micron deserves a meaningfully higher earnings multiple than the market has historically assigned it. The names most directly touched are MU itself, and indirectly NVDA and GOOGL as the primary end-demand drivers.
The tension is real: Micron is still a commodity memory maker at its core, and the cycle can turn. Samsung and SK Hynix are aggressively ramping HBM capacity, which could compress the pricing premium that is currently supercharging Micron's margins. The question of whether 39.8% gross margins are a new floor or a cycle peak is the central valuation debate.
Watchers should track Micron's next quarterly print (FY end August 2025) for gross margin trajectory and HBM shipment volume guidance. Any signal that pricing is softening or that hyperscaler capex is plateauing would pressure the re-rating thesis. Conversely, continued HBM allocation sold out into 2026 would reinforce the case that this is a structural, not cyclical, shift.
Micron's 48.9% revenue growth and 22.8% net margins reflect a genuine earnings inflection powered by HBM pricing — not a one-quarter blip. If hyperscaler capex sustains into 2026 and HBM supply remains tight, the historical discount Micron trades at versus logic-chip peers should compress, supporting a multiple expansion. The FY-end print in late August is the next hard data point to confirm or break the margin trend.
The read above, as written. kept as written · closes shown from JUN 29 on
A dated catalyst on AUG 28 · 2-4 months, into Aug 2025 earnings. Follow to be told when one lands.
Price context does not establish that the story caused the move.
With $37.4B in revenue growing at ~49% YoY and HBM allocations reportedly sold out into 2026, Micron's earnings power may be structurally undervalued if the market continues pricing it as a commodity DRAM name rather than a critical AI infrastructure supplier.
Memory has historically mean-reverted sharply — Samsung and SK Hynix are aggressively scaling HBM output, and if supply catches demand by mid-2026, Micron's 39.8% gross margins could prove a cycle peak rather than a new floor, compressing both earnings and the multiple simultaneously.
Kept as written · your side, if you take one, is graded privately against licensed closes after 10 trading days · nothing here is advice · How the Wire is made →
Reaction = the first close after the story against the close before it. Prior-session closes only; not a call.
Shares a name with this story — discovery, not a connection.
This page is kept as it was written on Jun 28. Later coverage joins it only when the company and catalyst evidence match, and what the stock did is shown from licensed end-of-day closes — never re-graded, never backdated. The judgment is yours.