来自 Maksym Misichenko · Nasdaq ·
AI智能体对这条新闻的看法
The panel is divided on Micron's future, with concerns about cyclical memory market volatility, inventory overhang, and geopolitical risks countering optimism about AI-driven demand and long-term agreements.
风险: Inventory overhang and potential price collapse once supply normalizes
机会: Structural demand from AI and long-term agreements with customers
本分析由 StockScreener 管道生成——四个领先的 LLM（Claude、GPT、Gemini、Grok）接收相同的提示，并内置反幻觉防护。 阅读方法论 →
英睿达最近市值突破了 1 万亿美元，这得益于一位看涨分析师的报告。
管理层表示，供应动态正在收紧。
根据未来预期，该股票定价良好。
英睿达 (NASDAQ: MU) 刚刚做出了历史上任何其他股票都未曾做过的事情。
这家存储芯片公司的市值从 5000 亿美元增长到 1 万亿美元，仅用了 48 个交易日，而大多数 1 万亿美元的公司花了超过一年才达到这一里程碑。
人工智能会创造世界上第一个万亿美元富豪吗？ 我们的团队刚刚发布了一份关于一家鲜为人知但至关重要的公司（被称为“不可或缺的垄断”）的报告，该公司为英伟达 (Nvidia) 和英特尔 (Intel) 都需要提供关键技术。继续 »
存储芯片股票一直备受追捧，英睿达的两个韩国同行 **三星 (Samsung) 和海力士 (SK Hynix) ** 也最近突破了万亿美元大关，但它们的增长速度不如英睿达快。
这家位于爱达荷州、生产 DRAM、HBM 和 NAND 的公司，由于 瑞士信贷 (UBS) 发布了一份报告，将该股票的目标价提高到 1650 美元，称其为当时的两倍多，从而突破了门槛。
瑞士信贷表示，长期存储供应协议可能支持整个行业的定价和需求可见性增加，从而在 2029 年之前支撑盈利和自由现金流的激增。
英睿达刚刚宣布了其财政第三季度的盈利日期为 6 月 24 日。在届时之前，这款芯片股票是否值得购买？让我们看看最新的迹象。
瓶颈已经为英睿达带来了收入和利润的激增。在最近一个季度，收入几乎翻 tripled 到 238 亿美元，净收入几乎增长了 10 倍，达到 138 亿美元，该公司在该季度记录了 67.6% 的运营利润率。
现在，管理层认为这些有利因素正在增强。在 **摩根大通 (JPMorgan Chase) ** 投资者大会上，全球运营执行副总裁 Manish Bhatia 表示，自上次盈利报告以来，该公司的财务前景已经加强。
Bhatia 认为，供需差距是结构性的，而不是周期性的，并预测在 DRAM、HBM 和 NAND 存储市场中，供应紧张将持续到 2026 年日历年之后。他还表示，一些主要客户只能满足其 60% 的存储需求，这表明供应受到限制。
3 月份，英睿达预计第三季度收入为 327.5 亿美元至 342.5 亿美元，调整后每股收益为 18.75 美元至 19.55 美元。Bhatia 的评论表明英睿达将超过这些目标。华尔街的预期仍然符合这一预测，预计收入为 337 亿美元，即增长 262%，调整后每股收益为 19.21 美元，而去年同期为 1.91 美元。
在 **英伟达 (Nvidia) ** 最近发布财报后的不久，首席财务官 Collette Kress 在一次采访中表示，芯片领导者比其他人更早地意识到内存价格会上涨，并且“很久以前就订购了”。对于核心内存产品，她补充说：“我们实际上正在与他们合作，以确定应该建造什么。”
这是英睿达的另一个积极信号，表明目前的产品以及未来产品都存在现有需求，这可能会推动其增长超出内存周期的通常限制。
英睿达和更广泛的存储领域以周期性著称，因为存储芯片的价格往往会随着库存水平的波动而上下波动。
人工智能热潮创造了超级周期，增长可能持续数年，但这并不意味着未来不会出现不利因素。考虑到英睿达的抛物线式上涨，在某个时候出现重大抛售的可能性很高。
然而，内存热潮可能还会持续数年，这意味着该股票仍有可能大幅上涨，尤其是在其以 16 的前向市盈率交易，并且预计 2027 财年收益将激增的情况下。
随着势头仍在增强，英睿达有望继续上涨。在 6 月 24 日的盈利报告之前购买该股票似乎是一个明智之举。
在您购买英睿达科技公司的股票之前，请考虑以下事项：
Motley Fool Stock Advisor 团队刚刚确定他们认为投资者现在应该购买的 10 支最佳股票……而英睿达科技公司不是其中之一。入选的 10 支股票在未来几年可能会产生巨大的回报。
请考虑 Netflix 在 2004 年 12 月 17 日被列入此名单时的情况……如果您当时投资了 1000 美元，您将拥有 465,733 美元！ 或者当 英伟达 在 2005 年 4 月 15 日被列入此名单时……如果您当时投资了 1000 美元，您将拥有 1,313,467 美元！
值得注意的是，Stock Advisor 的总平均回报率为 985%——与标准普尔 500 指数相比，市场表现优于 211%。不要错过最新的前 10 名名单，该名单可使用 Stock Advisor，并加入由个人投资者为个人投资者建立的投资社区。
**Stock Advisor 的回报率截至 2026 年 5 月 30 日。 *
Jeremy Bowman 持有英睿达科技公司的股份。Motley Fool 持有并推荐英睿达科技公司。Motley Fool 有一份披露政策。
本文件包含的观点和意见是作者的观点和意见，不一定代表纳斯达克公司的观点和意见。
四大领先AI模型讨论这篇文章
"MU's parabolic move already embeds most of the structural supply tightness, so the risk/reward ahead of June 24 is closer to even than the article suggests."
Micron's $1T cap in 48 days and 67% operating margins reflect genuine HBM/DRAM tightness, yet the article underplays how quickly Samsung and SK Hynix can add capacity once long-term agreements lock in pricing. At 16x forward earnings against 262% revenue growth, any June 24 miss versus the $33.7B consensus or signs of inventory restocking would trigger a sharp de-rating. The Nvidia comments confirm demand visibility but do not address execution risk in HBM3E ramps or potential gross-margin compression if ASPs peak earlier than 2027.
Management's post-earnings comments at JPMorgan already signaled upside to the $32.75-34.25B guide, so a clean beat on June 24 could extend the multiple to 20x rather than compress it.
"Micron's parabolic move reflects AI euphoria and supply tightness, not a permanent shift in the memory cycle—and the market is pricing in 2027 earnings as if margins stay at 67%, which history says won't happen."
The article conflates a cyclical supply tightness with structural demand, a dangerous leap. Yes, Micron's 67.6% operating margin and 10x net income jump are real. Yes, UBS's $1,650 target and Nvidia's long-term orders matter. But the 16x forward P/E assumes earnings don't normalize—and memory cycles always normalize. The article admits this ('notoriously cyclical') then dismisses it. Customers filling only 60% of needs sounds bullish until you realize: once supply catches up, those same customers will have excess inventory, crushing prices. The 48-day $500B-to-$1T run suggests momentum, not fundamentals. June 24 earnings could easily beat guidance and still disappoint on forward guidance if management signals supply normalization.
If HBM demand from AI truly is structural (not cyclical), and supply agreements lock in multi-year pricing, then Micron's margin expansion could persist through 2027-28, justifying 16x forward P/E on a much larger earnings base—making the stock cheap, not expensive.
"The market is currently pricing Micron for a permanent margin expansion that ignores the inevitable return of cyclical supply-side overcapacity."
The article presents a narrative of a 'supercycle' in memory, but it ignores the fundamental volatility inherent in commodity-like DRAM and NAND markets. While HBM (High Bandwidth Memory) is currently a high-margin bottleneck for AI, Micron's historical tendency to over-expand capacity during peak pricing cycles remains the primary risk. The valuation at a 16x forward P/E (price relative to expected future earnings) assumes that the current 67% operating margin is a new baseline rather than a cyclical peak. If supply-demand equilibrium returns sooner than 2026, those margins will compress rapidly, leading to a significant multiple contraction for MU.
The structural shift in memory toward custom, AI-integrated HBM may finally decouple Micron from historical commodity cycles, justifying a permanent re-rating of their valuation multiples.
"The bull thesis rests on a multi-year memory supercycle with persistent supply tightness, but the key risk is a faster-than-expected price and demand normalization that could erode margins and valuation."
Micron’s surge to a $1T market cap in a few months is exciting but not a free pass. The article leans on structural supply tightness into 2026+ and AI-driven demand, plus a plausible upside to UBS's $1,650 target. The caveat: memory cycles are notoriously volatile, and pricing can snap when inventories normalize or capex slows. Even with guidance implying near-term beat potential, any surprise weakness in NAND/DRAM prices, slower AI memory pull from hyperscalers, or a faster roll-off in capex could compress margins and drive multiple compression. The USD, customer mix, and commodity cyclicality add more near-term risk to earnings visibility. Forward P/E of ~16 looks cheap only if the cycle lasts.
Memory cycles are notoriously volatile; even with near-term tailwinds, demand or pricing could normalize faster than expected, eroding Micron's margin expansion and valuation. The AI-driven upcycle may prove less durable than implied, inviting multiple compression.
"HBM3E execution and yield risks threaten margins sooner than broad cyclical normalization."
Claude assumes a classic inventory overhang will crush prices once supply normalizes, but this overlooks how multi-year LTAs with AI customers can lock in pricing floors even if commodity DRAM rebounds. The bigger gap across all takes is Micron's HBM3E yield curve versus Nvidia's required volumes; any slippage here compresses ASPs before Samsung or SK Hynix capacity arrives, directly threatening the 67% margins the article celebrates.
"LTAs reduce but don't eliminate commodity risk if hyperscalers shift from scarcity-driven hoarding to inventory liquidation."
Grok flags HBM3E yield risk as the real margin threat, but sidesteps whether multi-year LTAs actually protect ASPs if commodity DRAM collapses. Claude's inventory overhang thesis assumes customers behave rationally post-normalization; what if hyperscalers hoard HBM inventory during peak scarcity, then dump it, forcing Micron to compete on price anyway? The LTA floor only holds if demand stays structural. Nobody's quantified what percentage of Micron's $33.7B guidance is locked-in LTA versus spot pricing.
"Micron's aggressive capex and geopolitical exposure create a rigid cost structure that leaves them vulnerable to margin compression regardless of HBM contract status."
Claude, your focus on LTA protection is missing the capital intensity trap. Even with locked-in pricing, Micron’s massive $10B+ capex plans for 2025-2026 create a rigid cost structure. If HBM3E yields improve, the 'scarcity premium' evaporates instantly. We are ignoring the geopolitical risk: if China restricts legacy DRAM imports in retaliation for US tech curbs, Micron loses the volume base needed to subsidize its R&D for the AI-specific HBM products. The cycle isn't just demand-driven; it's policy-sensitive.
"Policy/geopolitical frictions could delay Micron's HBM3E ramp and capex, eroding margins even with LTAs."
Gemini correctly flags geopolitical risk, but the bigger, underappreciated drag is policy-driven capex delays. If export controls tighten or China restrictions widen, Micron's equipment access and supplier shipments could slow HBM3E ramps, not just price floors. That dynamic can compress margins even with LTAs, and would curb the 'scarcity premium' sooner than memory cycles alone suggest. LTAs help, but they don't immunize supply chains from policy shocks.
The panel is divided on Micron's future, with concerns about cyclical memory market volatility, inventory overhang, and geopolitical risks countering optimism about AI-driven demand and long-term agreements.
Structural demand from AI and long-term agreements with customers
Inventory overhang and potential price collapse once supply normalizes