Hynix 业绩不及预期早已在亚洲买方圈流传，但美国 ADR 投资者措手不及 [1]Tae Kim · 29/07institutions who bought ADR def didn't expect it买入ADR的机构肯定没预料到这一点 市场对 YMTC/CXMT 扩产及消费端需求疲软的担忧加剧，但多头认为 LTA 将提供支撑 [2]fehmi.karagulla · 29/07This is what KIS wrote 2 weeks ago. He nailed it . Jukan flagged this report SK Hynix (000660) - 2Q26 miss is start of stable long term upside *KIS* Semicon analyst Minsook Chae previews SK Hynix (000660) 2Q26 OPe at W60.4tn (+61% QoQ, +556% YoY), 8% below W65tn consensus. Analyst lowered the ASP growth assumption for 2Q26 due to Co's; 1) higher HBM sales mix vs peers (slower HBM ASP hike vs commodity), 2) lower commodity DRAM px hike by Co vs original street estimates, and 3) LTA effect (more stable ASP) starting to kick in. Accordingly, 2Q26 DRAM blended ASP growth rate estimate is lowered to +28.9% QoQ (previous +50.0%) and commodity DRAM ASP growth rate lowered to 34.2% (previous 60.6%). But, the blended ASP growth will return to market average rate of c. 10% QoQ in 3Q26 as HBM4 mass production begins in earnest. HOWEVER, this is NOT a downturn signal. Rather, the near-term estimate revision is due to more realistic reflection of 3-5yrs LTA structures that will lead to more stable earnings growth over longer term (FY26/27/28 OP growth +419%/+53%/+19% YoY). In fact, analyst estimates Co-wide OPM to continue to rise from the record high 74.6% OPM expected in 2Q26. Expect valuation re-rating to reflect the sustainability of profits; LTAs are reducing earnings volatility and increasing visibility, which have been the discount factors. Yet, upon the successful listing of the ADR, the diverged flows may continue between Hynix KS and SKHY ADR shrs in the near-term (cost benefit for US instos as cash trading now possible vs swap for KS), and with possible pair trading between the two. Thus, SEC (005930) / SEC-P (005935) may outperform Hynix KS in the near term. => Recommend SEC as top-pick with the expected earnings upcycle through FY27~28 driven by superior pricing leverage in conventional DRAM/NAND negotiations and Co's leadership position in HBM4/4e over competitors. Analyst also trimmed SEC's FY26/27 OP estimates by -4%/-5% to reflect LTAs (vs -9%/-11% for Hynix), but again, for more stable longer term growth. The imminent employee compensation related shr buyback and the biggest ever shareholder return to come (special dvd at end-FY26 + start of new 3yr shareholder return cycle) should also support the shrs.这是KIS两周前写的。他说得很准。Jukan标记了这份报告。 SK海力士(000660) - 2Q26不及预期是长期稳定上行的开始 *KIS* 半导体分析师Minsook Chae对SK海力士(000660)2Q26营业利润的预测为60.4万亿韩元(环比+61%，同比+556%)，比65万亿韩元的市场共识低8%。分析师下调了2Q26的ASP增长假设，原因是公司：1) HBM销售组合高于同行(HBM ASP涨幅慢于大宗品)，2) 公司大宗DRAM价格涨幅低于原始市场预期，以及3) LTA效应(更稳定的ASP)开始显现。因此，2Q26 DRAM混合ASP增长率预估下调至环比+28.9%(此前为+50.0%)，大宗DRAM ASP增长率下调至34.2%(此前为60.6%)。但是，随着HBM4开始大规模量产，混合ASP增长率将在3Q26恢复到约环比10%的市场平均增速。 然而，这并非下行信号。相反，近期预估修正是因为更现实地反映了3-5年LTA结构，这将带来更长期稳定的盈利增长(FY26/27/28营业利润同比增速分别为+419%/+53%/+19%)。事实上，分析师预计公司整体营业利润率将从2Q26预期的创纪录高位74.6%继续上升。预计估值重估将反映盈利的可持续性；LTA正在降低盈利波动性并提高可见度，而这些此前一直是折价因素。 然而，在ADR成功上市后，短期内Hynix KS和SKHY ADR股票之间的资金流向可能继续分化(美国机构投资者现在可以进行现金交易，相比KS的掉期交易有成本优势)，并且两者之间可能存在配对交易。因此，SEC(005930)/SEC-P(005935)短期内可能跑赢Hynix KS。 => 推荐将SEC作为首选股，预计其在传统DRAM/NAND谈判中拥有更强的定价权，以及在HBM4/4e领域相对于竞争对手的领导地位，将推动FY27~28的盈利上升周期。分析师还将SEC的FY26/27营业利润预估下调了-4%/-5%以反映LTA(相比之下，海力士下调了-9%/-11%)，但同样是为了更稳定的长期增长。即将到来的与员工薪酬相关的股票回购以及有史以来最大的股东回报(2026财年末的特别股息+新的三年股东回报周期启动)也应支撑股价。。
对 PM 来说 = 警惕产能过剩叙事 尽管短期超跌，但制造商在需求放缓下仍积极扩产，反弹空间受限 [7]Zerocool · 29/07And BNK cut target twice by now per Jukan's share. SK Hynix (000660) — BNK Investment Securities Company Analysis Report | 2026/7/29 Analyst: Lee Min-hee, Semiconductor/IT — <mailto:mhlee@bnkfn.co.kr|mhlee@bnkfn.co.kr> Headline: Short-term oversold, but with demand slowdown persisting the rebound is limited Rating Hold [Maintained] Target price (6M) KRW 1,480,000 [Lowered] — -4.5% Current price (2026/7/28) KRW 1,550,000 Earnings miss expectations for a second consecutive quarter 2Q26 revenue came in at KRW 79.32tn (+51% QoQ, +257% YoY) and operating profit at KRW 60.54tn (+61% QoQ, +557% YoY), missing consensus by 5% and 6% respectively. In 1Q26, per-chip cost increases ran higher than anticipated; in 2Q26, the DRAM ASP increase came in below expectations on product mix effects. Despite demand momentum peaking out, producers are pushing competitive capacity additions With consumer conditions deteriorating, NAND spot prices have been on a downtrend since late April, and the OECD leading indicator has simultaneously turned lower. Hyperscalers are also shifting their approach to AI capex toward better cost-efficiency, and since the bulk of this year's planned spending relies on borrowing, rising rates and funding-market instability are increasing doubts about financing. Despite the slowing demand trend, manufacturers are announcing large-scale greenfield and expansion investments one after another on the back of long-term demand optimism, raising significant concern about a shift into oversupply. In addition, the strong reception for the CXMT IPO at the cycle peak is a negative from a competitive-intensity standpoint. Maintain Hold, target price cut to KRW 1.48mn Reflecting the slowdown in demand momentum, the shares have fallen sharply to the bottom of their P/EBIT band and are in short-term oversold territory. However, the competitive capacity-addition stance among producers is unlikely to change easily, and the demand slowdown is expected to persist through 2H, so the scope for a share price rebound should also be limited. We maintain our Hold rating and lower the target price from KRW 1.85mn to KRW 1.48mn (applying 26E P/B of 3.1x).根据Jukan的分享，BNK现在已经两次下调目标价。 SK海力士(000660) — BNK投资证券 公司分析报告 | 2026/7/29 分析师：李敏熙，半导体/IT — <mailto:mhlee@bnkfn.co.kr|mhlee@bnkfn.co.kr> 标题：短期超卖，但需求放缓持续，反弹空间有限 评级 持有 [维持] 目标价(6个月) 韩元 1,480,000 [下调] — -4.5% 当前价格(2026/7/28) 韩元 1,550,000 连续第二个季度盈利不及预期 2Q26营收为79.32万亿韩元(环比+51%，同比+257%)，营业利润为60.54万亿韩元(环比+61%，同比+557%)，分别低于市场共识5%和6%。在1Q26，单颗芯片成本增幅高于预期；在2Q26，受产品组合效应影响，DRAM ASP增幅低于预期。 尽管需求动能见顶，但生产商仍在推动竞争性产能扩张 随着消费环境恶化，NAND现货价格自4月下旬以来一直呈下降趋势，OECD领先指标也同时走低。超大规模云服务商也在将其AI资本支出方式转向更高的成本效益，并且由于今年计划支出的大部分依赖借贷，利率上升和融资市场的不稳定增加了对融资的疑虑。尽管需求趋势放缓，但制造商基于对长期需求的乐观情绪，接连宣布大规模的新建和扩建投资，这引发了对转向供应过剩的严重担忧。此外，在周期峰值时CXMT IPO获得热烈反响，从竞争强度角度来看是一个负面因素。 维持持有，目标价下调至148万韩元 反映需求动能放缓，股价已大幅下跌至P/EBIT区间的底部，处于短期超卖区域。然而，生产商之间竞争性的产能扩张立场不太可能轻易改变，需求放缓预计将持续到下半年，因此股价反弹空间也应有限。我们维持持有评级，并将目标价从185万韩元下调至148万韩元(应用26E市净率3.1倍)。。
META 的 Capex 指引未能打消市场对 2028 年后投资降速的疑虑 [8]AGdoom22 · 30/07“We aren’t providing a specific outlook for 2027 CapEx at this time. Infrastructure planning remains highly dynamic. *We’re gearing our current infrastructure plans toward maximizing capacity in 2026 and 2027, while giving us the flexibility to continue growing in 2028 and beyond.”* Sounds like '28 capex could be down“我们目前不提供2027年资本支出的具体展望。基础设施规划仍然高度动态。*我们当前的基础设施计划旨在最大化2026年和2027年的产能，同时为我们提供在2028年及以后继续增长的灵活性。”* 听起来'28年的资本支出可能会下降 MSFT Azure 增速虽达 43%，但未能全面提振半导体板块情绪 [9]Mo W · 30/0743% on azure...Azure上增长43%…… [10]pricedin · 30/07so...does MSFT drive a semi bid tomorrow? looking kinda like..no.那么……微软明天会推动半导体板块上涨吗？看起来……不太会。。
单票基本面让位于宏观因子与量化交易，市场陷入“杀跌一切”的流动性真空 [15]Methodica Trading Research · 30/07I think my worst fear over the years came to fruition in 2026: market flows fully dominated by algos in which single names move all correlated to factor flows in dynamics that are much more exaggerated and illogical for humans to position on the up and down. Algos move all the names together based on factors creating stupid correlations like if semis are bought railroads, adbe or aapl get sold. Feels single name fundamentals are total irrelevant and what really counts is being right on the factor/sector exposures.我认为我多年来最担心的事情在2026年成为了现实：市场资金流完全由算法主导，个股走势完全与因子资金流相关，其动态对于人类来说在上涨和下跌时都更加夸张且不合逻辑，难以布局。算法基于因子一起推动所有股票，造成愚蠢的相关性，比如如果半导体被买入，铁路股、ADBE或AAPL就会被卖出。 感觉个股基本面完全无关紧要，真正重要的是对因子/行业敞口的判断正确。
美联储维持利率不变，但长端美债收益率飙升，30年期突破 5.20% [20]HL Cap · 30/07**30*-*YEAR* YIELD HITS 5.20% FOR FIRST TIME SINCE 2007**30**年期收益率触及5.20%，为2007年以来首次 高利率环境直接威胁软件及高估值科技股的定价逻辑。
对 PM 来说 = 警惕 2022 重演 长端利率失控 + 降息路径不明 → 高估值科技股逢低买入风险极高 [23]John · 30/07observation: with yields ripping on the back end, not sure how high multiple stocks i.e. software favorites can continue to work. Everything with a high multiple could get really hurt, repeat of 2022. Just an observation, not a thesis观察：随着长端收益率飙升，不确定高倍数股票(即软件类宠儿)能否继续表现良好。所有高倍数的东西都可能受到重创，重演2022年。只是一个观察，不是一个论点。。
低频独立频道 · 有内容时才显示
MSFT 资本开支预期稳健，缓解了市场对 AI 投资缩减的担忧 [1]Matty · 30/07Pulled in some demand from next quarter by early go-live, and still have enough juice for 45% in Q1FY27 - that is such a strong demand signal通过提前上线，从下个季度预支了一些需求，并且仍然有足够的余力在 27 财年第一季度实现 45% 的增长——这是一个非常强劲的需求信号
对半导体板块的意义 MSFT 的稳健指引为整个 AI 基础设施产业链提供了支撑，缓解了近期半导体板块的抛售压力。[5]kkhajavi7 · 30/07the best leading indicators for the spend are the cloud numbers for the hyperscalers and thats accelerating meaningfully across the board + backlogs are enormous - so sure we dont have anyoen spoon feeding us a number for 28 - why would they - but for now the train seems very much intact and the stocks all just got absolutely destroyed支出最好的领先指标是超大规模云服务商的云数据，这些数据正在全面显著加速，加上积压订单巨大——所以，当然没有人会直接告诉我们 28 年的具体数字——他们为什么要这么做呢——但就目前而言，这列火车似乎仍然完好无损，而所有这些股票却刚刚被彻底摧毁了
META 资本开支未大幅上修，但 Q3 指引疲软及高管对 2027 年支出的模糊表态引发担忧 [6]kkhajavi7 · 30/07don't think we learned much new on the capex front here - sounds like they are going to keep spending at a healthy clip - for 2028 nobody knows and thats not new - you guys agree?不认为我们在资本支出方面学到了很多新东西——听起来他们会继续保持健康的支出节奏——至于 2028 年，没人知道，这也不是什么新鲜事——你们同意吗？
SK Hynix 业绩不及预期，NAND 资本开支前置引发供应过剩担忧 [9]AB · 30/07<@U0AQ2UUNJAF> nothing to tear apart. I don't know what happens near term. But I do know ever since the Koreans got into NAND since 2 decades ago, they have consistently over supplied the market. Their team has always over budgeted demand. And overbuilt. From 2009-2019 we have only seen BKs / TKO of NAND players. I do not want to be in this market with irresponsible players driven I am guessing by some government help or some crazy competitive behavior. They say we will be disciplined in capex yet last two quarters in a row they have talked about pulling forward pulling forward. IMO, this is why their stock which is darn cheap was at $700B mcap dropped another 8-10%. their ROI is definitely very high at 65-75% op margins. So instead of buying back stock, or paying down debt, they just keep spending on capex. This is what Samsung did to Spansion went from 2nd or 3rd largest to BK within 4-5 years. nothing to tear apart. I don't know what happens near term. But I do know ever since the Koreans got into NAND since 2 decades ago, they have consistently over supplied the market. Their team has always over budgeted demand. And overbuilt. From 2009-2019 we have only seen BKs / TKO of NAND players. I do not want to be in this market with irresponsible players driven I am guessing by some government help or some crazy competitive behavior. They say we will be disciplined in capex yet last two quarters in a row they have talked about pulling forward pulling forward. IMO, this is why their stock which is darn cheap was at $700B mcap dropped another 8-10%. their ROI is definitely very high at 65-75% op margins. So instead of buying back stock, or paying down debt, they just keep spending on capex. This is what Samsung did to Spansion went from 2nd or 3rd largest to BK within 4-5 years.<@U0AQ2UUNJAF> 没什么可分析的。我不知道短期内会发生什么。但我确实知道，自从 20 年前韩国人进入 NAND 领域以来，他们一直持续向市场供应过剩。他们的团队总是高估需求。并且过度建设。从 2009 年到 2019 年，我们只看到 NAND 厂商破产或被技术性击倒。我不想参与这个市场，因为有这些不负责任的参与者，我猜他们是受到某种政府援助或某种疯狂竞争行为的驱使。他们说要保持资本支出纪律，但过去连续两个季度，他们都在谈论提前、再提前。在我看来，这就是为什么他们那非常便宜的股票，在市值达到 7000 亿美元后，又下跌了 8-10%。他们的投资回报率确实非常高，营业利润率达到 65-75%。但他们不是回购股票或偿还债务，而是继续在资本支出上花钱。这就是三星对 Spansion 所做的，使其在 4-5 年内从第二或第三大厂商走向破产。 没什么可分析的。我不知道短期内会发生什么。但我确实知道，自从 20 年前韩国人进入 NAND 领域以来，他们一直持续向市场供应过剩。他们的团队总是高估需求。并且过度建设。从 2009 年到 2019 年，我们只看到 NAND 厂商破产或被技术性击倒。我不想参与这个市场，因为有这些不负责任的参与者，我猜他们是受到某种政府援助或某种疯狂竞争行为的驱使。他们说要保持资本支出纪律，但过去连续两个季度，他们都在谈论提前、再提前。在我看来，这就是为什么他们那非常便宜的股票，在市值达到 7000 亿美元后，又下跌了 8-10%。他们的投资回报率确实非常高，营业利润率达到 65-75%。但他们不是回购股票或偿还债务，而是继续在资本支出上花钱。这就是三星对 Spansion 所做的，使其在 4-5 年内从第二或第三大厂商走向破产。
BE 营收超预期，AI 数据中心电力需求推动强劲增长，但空头仍质疑其长期市场份额 [11]TMT Breakout · 29/07Quick $BE recap since couldn't fit it in the Morning Wrap *BE +8%: BE looked solid with a nice beat-and-raise, AI time-to-power demand driving its first $1B-plus revenue quarter, sharp margin expansion and another material FY26 guidance increase, while mgmt directly addressed project-delay, capacity and scandium concerns*. *The #s:* Revenue was $1.065B, +165.5% y/y (last q +130.4% y/y) and +42% q/q vs Street $826M, +~106% y/y; non-GAAP EPS was $0.78 vs Street $0.41. Non-GAAP gross margin was 34.3% vs Street 30.8%, operating margin was 22.5% vs 16.8%, and adjusted EBITDA was $253M vs $148M. FY26 revenue guidance increased to $3.9B-$4.2B, +~100% y/y at the midpoint vs Street $3.729B, +~84%, while operating income and EPS guidance were also raised well above Street. *Bull vs Bear Debate inside thread* Quick $BE recap since couldn't fit it in the Morning Wrap BE +8%: BE looked solid with a nice beat-and-raise, AI time-to-power demand driving its first $1B-plus revenue quarter, sharp margin expansion and another material FY26 guidance increase, while mgmt directly addressed project-delay, capacity and scandium concerns . The #s: Revenue was $1.065B, +165.5% y/y (last q +130.4% y/y) and +42% q/q vs Street $826M, +~106% y/y; non-GAAP EPS was $0.78 vs Street $0.41. Non-GAAP gross margin was 34.3% vs Street 30.8%, operating margin was 22.5% vs 16.8%, and adjusted EBITDA was $253M vs $148M. FY26 revenue guidance increased to $3.9B-$4.2B, +~100% y/y at the midpoint vs Street $3.729B, +~84%, while operating income and EPS guidance were also raised well above Street. Bull vs Bear Debate inside thread快速回顾一下 $BE，因为早间简报里放不下 *BE 上涨 8%：BE 表现稳健，实现了漂亮的业绩超预期并上调指引，AI 的上市时间需求推动了其首个超过 10 亿美元收入的季度，利润率大幅扩张，并再次大幅上调了 26 财年指引，同时管理层直接回应了项目延迟、产能和钪的担忧*。 *数据：* 收入为 10.65 亿美元，同比增长 165.5%（上一季度同比增长 130.4%），环比增长 42%，而华尔街预期为 8.26 亿美元，同比增长约 106%；非 GAAP 每股收益为 0.78 美元，华尔街预期为 0.41 美元。 非 GAAP 毛利率为 34.3%，华尔街预期为 30.8%；营业利润率为 22.5%，华尔街预期为 16.8%；调整后 EBITDA 为 2.53 亿美元，华尔街预期为 1.48 亿美元。 26 财年收入指引上调至 39 亿至 42 亿美元，中点同比增长约 100%，而华尔街预期为 37.29 亿美元，同比增长约 84%；同时，营业利润和每股收益指引也大幅上调，远高于华尔街预期。 *多头与空头辩论见帖内* 快速回顾一下 $BE，因为早间简报里放不下 BE 上涨 8%：BE 表现稳健，实现了漂亮的业绩超预期并上调指引，AI 的上市时间需求推动了其首个超过 10 亿美元收入的季度，利润率大幅扩张，并再次大幅上调了 26 财年指引，同时管理层直接回应了项目延迟、产能和钪的担忧 。 数据： 收入为 10.65 亿美元，同比增长 165.5%（上一季度同比增长 130.4%），环比增长 42%，而华尔街预期为 8.26 亿美元，同比增长约 106%；非 GAAP 每股收益为 0.78 美元，华尔街预期为 0.41 美元。 非 GAAP 毛利率为 34.3%，华尔街预期为 30.8%；营业利润率为 22.5%，华尔街预期为 16.8%；调整后 EBITDA 为 2.53 亿美元，华尔街预期为 1.48 亿美元。 26 财年收入指引上调至 39 亿至 42 亿美元，中点同比增长约 100%，而华尔街预期为 37.29 亿美元，同比增长约 84%；同时，营业利润和每股收益指引也大幅上调，远高于华尔街预期。 多头与空头辩论见帖内