Micron Earnings Preview: Price-Driven Beat Expected, Cycle Duration and Buyback Resumption Are Key Focus
nashnova research
UBS expects Micron's August-quarter revenue at $52.4 billion with 87.6% gross margin, both above guidance caps — yet the beat is almost entirely price-driven, not volume. The market debate has shifted from 'how much prices rise' to how long the upcycle lasts and when buybacks begin.
Is the beat driven by prices or by volume?
UBS estimates August-quarter DRAM bit shipments grew just 3% sequentially, while average selling prices jumped 23%. NAND told the same story: bits up 4%, prices up 23%. This means → revenue and profit gains came almost entirely from pricing, not volume.
Standard DDR product prices rose 24% sequentially, broadly in line with UBS's industry contract-price estimate of roughly 26%.
HBM — high-bandwidth memory paired with AI chips — brought in about $3.5 billion, with bits up 9% and prices up 10%. In plain terms = even the hottest AI memory product grew modestly on volume; the money came from higher prices.
Is HBM still a high-margin business?
HBM's margin premium has narrowed sharply. In the August quarter, HBM sold at roughly $14.06 per GB versus $13.83 for standard DRAM — nearly the same. In early 2025, HBM pricing was more than four times standard DRAM.
Because HBM costs far more to manufacture, its gross margin came in at about 68% — actually below standard DRAM's roughly 91%. This means → on a pure-margin basis, making ordinary memory is now more profitable than making HBM.
UBS expects vendors to re-widen the gap. By Q4 2027, HBM pricing should reach about $31.11 per GB — roughly 1.8× standard DRAM — lifting HBM gross margin back to around 85%.
Do long-term agreements help or hurt Micron?
Micron disclosed 16 strategic customer agreements guaranteeing minimum cumulative revenue of about $100 billion, backed by roughly $22 billion in cash deposits and financial commitments. The locked-in price is about $10.5 per GB, holding through the November 2028 quarter.
Meanwhile, open-market spot pricing has climbed to around $23 per GB. In plain terms = Micron locked in a large batch of orders at low prices; the market has since doubled, but those shipments still go out at the original rate.
UBS expects open-market prices to rise another ~10% in the November quarter, but Micron's blended ASP will rise only in the high single digits. This means → only part of the open-market upside reaches Micron's P&L — the agreements cap the upside.
How long can the upcycle last?
UBS's latest channel checks show DRAM order-fulfillment rates still around 60% — buyers can secure only six out of every ten units they want.
Server DDR bit demand in 2027 is expected to grow roughly 80% year-on-year; server and storage SSD bit demand growth may exceed 100%. UBS sees DRAM remaining undersupplied through at least Q2 2028.
Chinese capacity expansion is already in the model: CXMT's DRAM bit share is projected to rise from about 7% in 2025 to 9% in 2027. YMTC is redirecting roughly 30,000 of its 45,000 annual wafer-start additions toward DRAM instead of NAND. This reflects a longer NAND pricing cycle — UBS expects NAND prices to peak only in Q3–Q4 2027.
What exactly are UBS and the market disagreeing on?
Near-term estimates are close; long-term estimates diverge sharply. For calendar-year 2027 EPS, UBS forecasts $212.17 versus the Street's $164.71 — about 29% higher. For 2028: $277.19 versus $169.32 — about 64% higher.
The Street expects 2028 earnings roughly flat versus 2027. UBS believes the shortage extends another year, with prices rolling over only in H2 2028.
In plain terms = the Street thinks next year is the peak; UBS thinks the peak comes the year after — the bet is whether the upcycle lasts one more year.
When do buybacks start, and how large could they be?
CHIPS Act–related buyback restrictions expire on December 9, 2026. UBS expects Micron to begin repurchasing about $20 billion per quarter from the February 2027 quarter, ramping quickly to $40–50 billion per quarter.
Over fiscal years: FY2027 buybacks of roughly $100.5 billion, FY2028 roughly $192.5 billion, FY2029 roughly $210 billion. Free cash flow supports this: about $147 billion in FY2027 and $201.3 billion in FY2028.
Diluted shares outstanding would fall from roughly 1.151 billion in FY2026 to about 873 million in FY2029 — a ~24% reduction in three years. This means → even if revenue declines, share-count shrinkage cushions EPS. However, UBS does not expect many buyback details on the September 30 call — the company is unlikely to commit before restrictions lift.
市场有风险，内容仅供研究参考，不构成投资建议。