Micron's Blowout Quarter: How Much Longer Can the AI Memory Supercycle Run?

一、数字有多炸裂 | The Headline Numbers

美光（Nasdaq: MU）于6月24日美股盘后公布2026财年第三季度（截至5月28日）财报，几乎是教科书级别的"营收、利润、指引全面超预期"。

Micron (Nasdaq: MU) reported fiscal Q3 2026 results (quarter ended May 28) after the bell on June 24 — and it was about as clean a "beat across the board" as you'll see.

营收同比暴增346%，环比从上季度238.6亿跳到414.6亿，是公司历史上单季环比增幅最大的一次。

Revenue surged 346% year-over-year, and jumped from $23.9B to $41.5B sequentially — the largest single-quarter dollar increase in company history.

更关键的是指引：美光给出的Q4营收指引为500亿±10亿美元，远高于财报前市场预期的约400亿——这才是真正点燃市场情绪的地方，因为"超预期"早已被股价提前消化，"指引超预期"才是稀缺品。

But the real story is guidance. Micron guided Q4 revenue to $50B ± $1B, well above the ~$40B the Street had penciled in pre-print. A beat was already priced in — a guidance raise of this size was not.

二、为什么会这么猛 | Why It's This Strong

核心逻辑没有变化，只是兑现的速度超过了所有人的模型：AI数据中心对高带宽内存（HBM）和DRAM的需求，已经把行业从"周期性大宗商品"重新定价为"结构性紧缺资产"。

The underlying logic hasn't changed — it's just playing out faster than anyone modeled. AI data-center demand for HBM and DRAM has re-rated the entire memory industry from "cyclical commodity" to "structurally scarce asset."

几个细节值得注意：

A few details worth flagging:

• 数据中心业务是真正的引擎
  ：核心数据中心部门单季营收从56.9亿环比跳到115.2亿，毛利率高达87%。 The Core Data Center unit alone nearly doubled sequentially (from $5.7B to $11.5B) with an 87% gross margin.
• 供给侧的稀缺是真实的，不是炒作
  ：管理层此前披露，目前只能满足部分关键客户50%~2/3的需求；2026年全年HBM产能已经售罄，订单已排到2027年。 Management has disclosed it can fulfill only 50%–two-thirds of key customer demand; 2026 HBM capacity is fully sold out, with orders booked into 2027.
• "战略客户协议"是新变量
  ：美光在财报中特别提到正在执行多年期的Strategic Customer Agreements，这意味着收入的可预测性和持续性可能比上一轮周期更强——这也是CEO Sanjay Mehrotra在新闻稿中刻意强调的一点。 Micron flagged multi-year Strategic Customer Agreements as a key new lever — designed to make this cycle's revenue durability look less like the boom-bust memory cycles of the past.

三、市场怎么反应的 | How the Market Reacted

财报公布后的即时反应是上涨约2.5%，相对克制——这其实是个值得思考的信号。

The immediate post-earnings reaction was a relatively modest +2.5% — and that muted reaction is itself worth thinking about.

财报前，美光股价已经年内上涨约300%、过去12个月上涨超800%，市值刚突破1万亿美元，期权市场隐含的财报后波动幅度高达13%~17%。换句话说，"利好"早就被股价提前买入了，这次财报与上季度（Q2交出31%的超预期幅度，财报后一周却反而下跌约20%）形成了有趣的对比：这次美光不仅超预期，还把指引一起拉高了，这是市场愿意"买账"而不是"逢高出货"的关键原因。

Going in, MU was up ~300% YTD and ~800% over 12 months, had just crossed a $1 trillion market cap, and options were pricing a 13%–17% move. In other words, a beat alone was already in the price. That's exactly what happened last quarter too — Micron beat Q2 by 31%, yet the stock fell ~20% over the following week on guidance concerns. This time, the beat-and-raise combination appears to be what kept sellers from showing up.

四、风险与平衡视角 | The Other Side of the Ledger

作为内容创作者，这里也想给一个相对平衡的视角，而不只是唱多：

In the interest of balance, a few counterpoints worth keeping in mind:

• 估值已经很贵
  ：目前美光的PE接近48倍，对应"超级周期"叠加"AI叙事"的双重溢价，意味着任何一点指引不及预期，都可能引发剧烈回调。 Valuation is no longer cheap — MU trades around 48x earnings, pricing in both a memory supercycle and an AI premium. Any hint of guidance disappointment from here could trigger an outsized drawdown.
• 竞争格局在变化
  ：SK海力士在HBM市场份额上仍领先（约62%），且在HBM4每针速度上目前领先美光；三星也在加速扩产。 SK Hynix still leads HBM market share (~62%) and is currently ahead on HBM4 per-pin speed; Samsung is also ramping aggressively.
• 存储芯片本质上仍是周期性行业
  ：历史上每一轮"超级周期"最终都会被新增产能追平。美光、三星、SK海力士都在大举扩产（美光未来两个财年资本开支预计超250亿美元），产能释放通常发生在2027-2028年，这意味着当前的高毛利率窗口期是有时间限制的。 Memory remains a fundamentally cyclical business. Every supercycle eventually meets new supply — and Micron, Samsung, and SK Hynix are all spending aggressively (Micron's capex is projected above $25B over the next two fiscal years), with new capacity expected to land around 2027–2028. The current margin window has a shelf life.

五、写在最后 | Closing Thought

这份财报本身没有任何争议——它是货真价实的超预期，而且是"超预期+上调指引"的组合，含金量比单纯的"营收利润双超"更高。真正的悬念从来不是这一季，而是：当2027-2028年新增产能逐步释放、AI资本开支增速也可能放缓时，今天86%的毛利率能维持多久？

This quarter itself isn't really debatable — it was a genuine beat-and-raise, which carries more weight than a simple top-and-bottom-line beat. The real question was never this quarter. It's whether an 86% gross margin can survive the new supply landing in 2027–2028, just as AI capex growth itself may start to decelerate.

这篇内容仅为财报数据梳理与公开信息分析，不构成任何投资建议。

This piece is a factual recap and analysis of public earnings data. It does not constitute investment advice.