Micron's biggest quarter ever: HBM sold out, margins near 81%
Fiscal Q3 2026 reported June 24, record revenue on AI memory, data-centre revenue up ~150%, full-year HBM capacity already booked
Summary
Micron Technology reported fiscal Q3 2026 after the US close on 24 June, in what analysts billed as its biggest quarter on record. Consensus pointed to roughly $34.7B revenue (~283% YoY growth) and gross margin near 81%, company guidance had been ~$33.5B at ~81%. The driver is high-bandwidth memory: Micron began volume HBM4 (36GB, 12-high) shipments in calendar Q1 for Nvidia's Vera Rubin platform, with the ramp running roughly twice as fast as HBM3E, and its data-centre revenue grew about 150% on HBM3E. Micron has said its entire 2026 HBM output is already sold out. The same supply discipline tightening AI-accelerator memory is lifting commodity DRAM and NAND prices downstream, a cost that flows through to PCs and phones.
By the numbers
- ~$34.7B, consensus Q3 revenue (~283% YoY); company guidance ~$33.5B ±$750M.
- ~81%, guided gross margin (consensus reaching toward 81.6%).
- ~150%, data-centre revenue growth on HBM3E supply to AI accelerators.
- 2x, HBM4 ramp speed vs. the HBM3E ramp.
2.8 TB/s, bandwidth of Micron's 12-high HBM4 for Vera Rubin (>20% better power efficiency vs HBM3E).
- Sold out, Micron's full-year 2026 HBM capacity.
Why it matters
Micron is the US-headquartered HBM supplier in a market otherwise led by Korea. A clean print at ~81% margin validates the AI-memory supercycle as structural, not a spike, and confirms memory, not just logic, as a value-dense chokepoint. It also sharpens the political stakes of the China memory probe and CXMT's rise against the three incumbents.
What to watch
- Q4 revenue/margin guidance, whether the supercycle is still accelerating.
- HBM4 yield and any Vera Rubin allocation gains vs. SK Hynix and Samsung.
- Pass-through of memory tightness into consumer DRAM/NAND prices through H2 2026.