摩根士丹利Codex · 实盘不做把握 6/10🔁 +1 同一事件
北京时间·🕒 发布 11:38:05时区推断·⟳ 抓取 11:44:06·⚡ 决策 07-28 11:45:24 +1m18s·91.2 天前
Amkor二季度毛利率和EPS超预期,但亚利桑那投产稀释框架未变
Amkor二季度收入18.98亿美元、EPS 0.70美元、毛利率16.8%,均明显好于摩根士丹利和市场预期,计算与汽车工业需求是主要正向增量。
📄 原文摩根士丹利↗ 原文
Amkor Technology Inc: GM% surprise; Arizona framework reiterated
AMKR reported a strong beat and raise on GM% and EPS, driven by Computing and Automotive/Industrial strength and broad-…
不做 Amkor Technology(AMKR.US)
财报毛利率和EPS虽超预期,但报告维持EW、亚利桑那中期稀释未变且通信业务疲弱;盘后未确认强单边利好,当前没有足够正期望追多。
展开详情(成交明细 / 研究档案 / 原文全文 / 决策记录链)
研究档案
消息传播链
- Amkor发布二季度财报,确认收入18.98亿美元、毛利率16.8%、EPS 0.70美元,并给出三季度收入19.5亿至20.5亿美元、毛利率18.5%至19.5%的指引。
- 公司召开二季度电话会,报告中提到的亚利桑那成本稀释和NVIDIA预付款细节主要来自电话会解读。
- 媒体报道财报超预期但销售展望偏平,称AMKR盘后下跌,说明市场对正负信息的即时解读并不单边。市场反应:AMKR 7月27日常规交易收跌6.54%,盘后报价来源分歧但未显示强劲上行确认。
- 摩根士丹利将目标价从69美元小幅上调到70美元但维持EW,核心结论是执行改善被亚利桑那中期稀释部分抵消。
关键理由
- Amkor二季度收入18.98亿美元、EPS 0.70美元、毛利率16.8%,均明显好于摩根士丹利和市场预期,计算与汽车工业需求是主要正向增量。
- 三季度毛利率和EPS指引高于预期,但收入指引约20亿美元仅大致持平或低于部分市场预期,通信业务还受到SiP迁移和Android疲软拖累。
- 亚利桑那项目仍会在2027年形成1-2个百分点经营利润率压力、2028年进一步压低毛利率,且公司盘后反应并未确认强单边重定价。
主要风险
- 若市场更重视AI/HPC先进封装产能被NVIDIA预付款锁定,AMKR次日可能低开后修复。
- 财报类事件披露后传播快,且AMKR近一个月已大幅波动,追多容易遇到高开低走或卖方修模分歧。
反向证据
- 做多前查反向 · 来源 web · 日期 2026-07-27 · 影响 weakens
反向证据包括三季度收入未显著超预期、通信业务环比高个位数下滑、亚利桑那2027-2028年稀释框架未变,削弱追多边际。
检索:Amkor Technology Q2 2026 earnings gross margin EPS guidance Arizona dilution communications weakness 原文 ↗
原文全文
展开原文 · 4,749 字
Amkor Technology Inc: GM% surprise; Arizona framework reiterated
AMKR reported a strong beat and raise on GM% and EPS, driven by Computing and Automotive/Industrial strength and broad-based utilization improvement.
GM% came in well ahead of expectations at 16.8% (+250bps sequentially), ⅔ volume-driven and ⅓ mix. EPS of $0.70 included a $14mn discrete tax benefit (ETR of 14% vs. ~20% full-year guide); the underlying beat was still substantial. Our estimates move higher on stronger GM% execution and an accelerated compute ramp, though the Arizona dilution math - reiterated on the call - partially offsets the magnitude of our upward revisions.
We are giving the company credit for strong execution - GM% is running well ahead of schedule (high-teens initially guided for 2H), Phase 1 Arizona is now fully committed, and the NVIDIA prepayment ($1.5bn, to be received in 2027 and booked as deferred revenue, returned to the customer as services are delivered; Amkor owns the capex separately and the prepayment does not reduce gross capex). That said, the ~$2.50 2026 EPS exit rate is not a straight-line jumping-off point: 2027 brings a 1-2% operating margin headwind as Arizona startup costs ramp in OpEx, and 2028 brings a second wave of GM% dilution as those costs flip into COGS from an underutilized new facility. Comms adds near-term complexity, with Q3 guided down high-single-digits sequentially - ~50% strategic SiP migration from Korea to Vietnam (a multi-quarter dynamic extending into Q4 and 1H27), ~50% market-driven Android softness and memory constraints. We remain EW.
What we learned from the call/callback/release:
1.) (+) GM% well ahead in 2H - utilization and mix the twin drivers. The 16.8% GM% beat our 15.1% estimate by ~170bps and came in above the high end of guidance.
Key Takeaways
JunQ GM%/EPS beat-and-raise on Computing/Auto (mix) and utilization; 3Q26 GM% & EPS guide well above MSe/Street; revenue broadly in-line overall.
Strong execution but Arizona remains a multi-year dilution story - OpEx headwind in 2027, gross margin headwind in 2028.
PT little changed at $69→$70; ~$2.50 2026 EPS exit rate is not a straight-line jumping-off point for 2027.
Per the call, ⅔ of the 250bps sequential expansion was volume/utilization-driven and ⅓ was favorable product mix, with advanced packaging carrying meaningfully more accretive margins than the mainstream portfolio. The 3Q26 GM% guide of 18.5-19.5% is well above MSe/Street at 16.3%/17.4%, driven primarily by an accelerated compute ramp; Q4 GM% was not guided but mgmt indicated flat vs. Q3 is not unreasonable.
2.) (-) Communications - below seasonal, strategic SiP migration adds complexity. Q3 Comms guided down HSD sequentially, driven by: (i) the strategic SiP migration from Korea to Vietnam (~50% of the decline, coordinated with customers for over a year, freeing up Korea capacity for higher-value advanced packaging); and (ii) market-driven headwinds including memory constraints and build pattern changes (~50%, more pronounced in Android, which was already down 20% q/q in 2Q).
3.) (=) Arizona framework unchanged - NVIDIA prepayment clarified, 2026 exit rate creates a misleading baseline for 2027-2028. The Investor Day framework (1-2% operating income dilution starting 2027, costs moving from OpEx to COGS in 2028) was reiterated; the two customer agreement announcements since Investor Day increased visibility but did not change the dilution math. The $1.5bn NVIDIA prepayment will be received in 2027 and booked as deferred revenue, returned to the customer as services are delivered in the US over a 5-10 year term; it is an operating cash inflow but does not reduce gross capex, which remains $2.5-$3.0bn unchanged. While AMKR is tracking to exit 2026 at ~$2.50 EPS, 2027 and 2028 are not a straight-line extrapolation: 2027 takes an OpEx hit from Arizona startup costs, and 2028 takes both an OpEx hit and a gross margin hit as those costs flip into COGS once production begins.
Result Details:
• Details on the June quarter: Revenue, gross margin and EPS were above our estimates. Revenue of $1.898bn (up 12.7% q/q and 25.6% y/y) was above our estimate of $1.810bn. GM/EPS came in at 16.8%/$0.70, above our estimates of 15.1%/$0.46.
• September quarter guidance. The company guided revenue, gross margin and EPS largely above our estimates. Amkor expects revenue of $2.000bn (up 5.4% q/q and 0.7% y/y) at the midpoint, slightly below our estimate of $2.026bn. GM/EPS were guided to 19.0%/$0.77, above our estimates of 16.3%/$0.59. The company increased its 2026 capex guide to approximately $2.5bn.
• Changes to our estimates: Our FY26 estimates come up due to stronger GM% and EPS from $7.57bn/15.9%/$2.07 to $7.60bn/17.4%/$2.58, while our
AMKR reported a strong beat and raise on GM% and EPS, driven by Computing and Automotive/Industrial strength and broad-based utilization improvement.
GM% came in well ahead of expectations at 16.8% (+250bps sequentially), ⅔ volume-driven and ⅓ mix. EPS of $0.70 included a $14mn discrete tax benefit (ETR of 14% vs. ~20% full-year guide); the underlying beat was still substantial. Our estimates move higher on stronger GM% execution and an accelerated compute ramp, though the Arizona dilution math - reiterated on the call - partially offsets the magnitude of our upward revisions.
We are giving the company credit for strong execution - GM% is running well ahead of schedule (high-teens initially guided for 2H), Phase 1 Arizona is now fully committed, and the NVIDIA prepayment ($1.5bn, to be received in 2027 and booked as deferred revenue, returned to the customer as services are delivered; Amkor owns the capex separately and the prepayment does not reduce gross capex). That said, the ~$2.50 2026 EPS exit rate is not a straight-line jumping-off point: 2027 brings a 1-2% operating margin headwind as Arizona startup costs ramp in OpEx, and 2028 brings a second wave of GM% dilution as those costs flip into COGS from an underutilized new facility. Comms adds near-term complexity, with Q3 guided down high-single-digits sequentially - ~50% strategic SiP migration from Korea to Vietnam (a multi-quarter dynamic extending into Q4 and 1H27), ~50% market-driven Android softness and memory constraints. We remain EW.
What we learned from the call/callback/release:
1.) (+) GM% well ahead in 2H - utilization and mix the twin drivers. The 16.8% GM% beat our 15.1% estimate by ~170bps and came in above the high end of guidance.
Key Takeaways
JunQ GM%/EPS beat-and-raise on Computing/Auto (mix) and utilization; 3Q26 GM% & EPS guide well above MSe/Street; revenue broadly in-line overall.
Strong execution but Arizona remains a multi-year dilution story - OpEx headwind in 2027, gross margin headwind in 2028.
PT little changed at $69→$70; ~$2.50 2026 EPS exit rate is not a straight-line jumping-off point for 2027.
Per the call, ⅔ of the 250bps sequential expansion was volume/utilization-driven and ⅓ was favorable product mix, with advanced packaging carrying meaningfully more accretive margins than the mainstream portfolio. The 3Q26 GM% guide of 18.5-19.5% is well above MSe/Street at 16.3%/17.4%, driven primarily by an accelerated compute ramp; Q4 GM% was not guided but mgmt indicated flat vs. Q3 is not unreasonable.
2.) (-) Communications - below seasonal, strategic SiP migration adds complexity. Q3 Comms guided down HSD sequentially, driven by: (i) the strategic SiP migration from Korea to Vietnam (~50% of the decline, coordinated with customers for over a year, freeing up Korea capacity for higher-value advanced packaging); and (ii) market-driven headwinds including memory constraints and build pattern changes (~50%, more pronounced in Android, which was already down 20% q/q in 2Q).
3.) (=) Arizona framework unchanged - NVIDIA prepayment clarified, 2026 exit rate creates a misleading baseline for 2027-2028. The Investor Day framework (1-2% operating income dilution starting 2027, costs moving from OpEx to COGS in 2028) was reiterated; the two customer agreement announcements since Investor Day increased visibility but did not change the dilution math. The $1.5bn NVIDIA prepayment will be received in 2027 and booked as deferred revenue, returned to the customer as services are delivered in the US over a 5-10 year term; it is an operating cash inflow but does not reduce gross capex, which remains $2.5-$3.0bn unchanged. While AMKR is tracking to exit 2026 at ~$2.50 EPS, 2027 and 2028 are not a straight-line extrapolation: 2027 takes an OpEx hit from Arizona startup costs, and 2028 takes both an OpEx hit and a gross margin hit as those costs flip into COGS once production begins.
Result Details:
• Details on the June quarter: Revenue, gross margin and EPS were above our estimates. Revenue of $1.898bn (up 12.7% q/q and 25.6% y/y) was above our estimate of $1.810bn. GM/EPS came in at 16.8%/$0.70, above our estimates of 15.1%/$0.46.
• September quarter guidance. The company guided revenue, gross margin and EPS largely above our estimates. Amkor expects revenue of $2.000bn (up 5.4% q/q and 0.7% y/y) at the midpoint, slightly below our estimate of $2.026bn. GM/EPS were guided to 19.0%/$0.77, above our estimates of 16.3%/$0.59. The company increased its 2026 capex guide to approximately $2.5bn.
• Changes to our estimates: Our FY26 estimates come up due to stronger GM% and EPS from $7.57bn/15.9%/$2.07 to $7.60bn/17.4%/$2.58, while our
决策记录链 · 只追加
每次领取产生一条记录,重新决策追加一条并以 supersedes 指向被替代的记录(ADR 0014)。
这条决策早于决策账本(2026-10-08 起逐条追加),没有追加记录。