Variant-perception thesis · personal research變異觀點投資論點 · 個人研究

Applied Materials (AMAT)

The picks-and-shovels enabler of chip complexity. The reasoning is the model's; every number carries its source and date. Where the first draft ran on memory, live data rewrote the conclusion. 晶片複雜度背後「賣鏟子」的設備商。推論由模型產生，但每個數字都標註來源與日期。第一版靠記憶寫成的地方，即時資料改寫了結論。

How this is built. A variant-perception framework — the discipline of ignoring consensus to find what the market may be misunderstanding, underestimating, or ignoring. It walks the same nine steps for any stock: business → bull → bear → the variant view → catalysts → management → moat → probability-weighted odds → what it takes to double. The rule that makes it trustworthy: no number is written from memory — every figure is grounded to a filing or live data with a date, and the share price is recomputed the moment this page is built.

我們是怎麼分析這家公司的。採用「變異觀點（Variant Perception）」框架 — 刻意不跟市場共識走，去找市場可能誤解、低估或忽略的地方。對任何一檔股票都走同樣九個步驟：本業 → 多方 → 空方 → 變異觀點 → 催化劑 → 經營層 → 護城河 → 機率加權情境 → 翻倍需要什麼。讓它可信的關鍵規則：沒有任何數字憑記憶寫成 — 每個數字都標註來自財報或即時資料與日期，股價在生成這頁的當下即時重算。

Isometric illustration: a layered silicon wafer being built up step by step by robotic deposition and polishing tools
Price as of股價時間 2026-09-02 close (IBKR)收盤（IBKR） Fundamentals財報數據 FQ3 FY2026 (reported 2026-08-132026-08-13 公布) + 10-Q (filed 2026-08-202026-08-20 申報) + FY2025 10-K
$438.46
Close 09-0209-02 收盤 ✓ IBKR
+71.0%
YTD · -40.7% off high $739.67距高點 $739.67 -40.7% ✓ IBKR
~34×
Fwd P/E預估本益比 close ÷ FY26E EPS收盤÷FY26E
$9.12B
FQ3 rev · recordFQ3 營收 · 新高 ✓ release財報 08-13
Forward P/E = 09-02 close $438.46 ÷ FY26E non-GAAP EPS $12.75 — 9M FY26 actual $8.73 plus the FQ4 guide midpoint $4.02, i.e. the company's own numbers, still unchanged since 08-13 ✓ release 08-13 — this leg of the compression is pure price. On FY27E $18.44 (MarketScreener/S&P Global, 09-03; basis ~3% above the stockanalysis series this page used at 08-14 — like-for-like ≈$17.9): ~24×. Trailing TTM GAAP ~38×. 09-02 is the last completed session; this page was built before the 09-03 open.預估本益比 ＝ 09-02 收盤 $438.46 ÷ FY26E 非 GAAP 每股盈餘 $12.75 — 即 FY26 前三季實績 $8.73 加上 FQ4 指引中值 $4.02，用的是公司自己的數字，自 08-13 以來未變 ✓ 財報 08-13 — 這一段收縮純粹來自股價。若以 FY27E $18.44 計（MarketScreener／S&P Global，09-03；口徑較本頁 08-14 版採用的 stockanalysis 系列高約 3%，同口徑約 $17.9）：約 24 倍。TTM GAAP 歷史本益比約 38 倍。09-02 是最後一個完整交易日；本頁生成於 09-03 開盤前。

Default path: Finding → Numbers → Odds → Playbook → Update log (~8 min). The full argument — business, bull, bear, variant perception, management — is collapsed below.預設動線：結論 → 數字 → 機率 → 操作框架 → 更新紀錄（約 8 分鐘）。完整論證（本業、多方、空方、變異觀點、經營層）已收合於下方。

headlineThe bull re-rating has largely already happened多方期待的「評價重估」其實大半已經發生

一句話：股價已跌進本頁預設的進場區、客戶也沒砍單，但 34 倍本益比以 AMAT 自己的歷史標準看仍然貴。這次進場賭的是 AI 週期比過去長，不是撿便宜。

One line: the price has fallen into this page's pre-set entry zone and customers have not cut orders, but 34× is still expensive by AMAT's own history. This entry is a bet that the AI cycle runs longer than past ones, not a bargain hunt.

1. 股價大修正，殺的是本益比，不是獲利。 AMAT 於 07-01 前後盤中見頂 $739.67，09-02 收 $438.46，距高點 -40.7% ✓ IBKR · 09-02，較本頁上次生成時的 08-13 收盤再跌 18%。同一段期間 FY26E 每股盈餘 $12.75 沒動（前三季實績加 FQ4 指引，公司自己的算式）✓ 財報 · 08-13，所以這段下跌純粹是市場在殺本益比：預估本益比從 6 月的約 57 倍一路降到 55、46、42，現在約 34 倍；以新的 FY27 共識 $18.44 計約 24 倍。✗→ 08-14 版為 ~42×／~31×

1. The correction is killing the multiple, not the earnings. AMAT peaked intraday around 07-01 at $739.67 and closed 09-02 at $438.46, -40.7% off that high ✓ IBKR · 09-02 — a further 18% down since the 08-13 close this page was last built at. FY26E non-GAAP EPS $12.75 has not moved over that stretch (three reported quarters plus the FQ4 guide, the company's own arithmetic) ✓ release · 08-13, so this decline is pure multiple compression: forward P/E fell from ~57× in June through 55, 46, 42, to now ~34×; on the new FY27 consensus $18.44 it is ~24×. ✗→ was ~42×/~31× at 08-14

2. 本業沒壞，客戶還在砸錢。 SK 海力士(000660.KS) 核准 ₩54兆 蓋新的龍仁 DRAM 廠與清州 NAND 廠（08-07）✓ CNBC · 08-07；台積電(TSM.N) 自己說設備採購已到 12 月計畫的約 1.9 倍（09-02）✓ Bloomberg · 09-02；SEMI 把 2028 年 WFE 預測上修到 $220B（09-01）✓ SEMI · 09-01；Q4 共識 EPS $4.06 高於公司指引 $4.02 ⚠ 單一來源。08-14 留下的兩個問題（毛利率指引持平、2027 沒有數字）還開著，但沒有變糟。

2. The business is not broken — customers are still spending. SK hynix (000660.KS) approved ₩54T for new Yongin DRAM and Cheongju NAND fabs (08-07) ✓ CNBC · 08-07; TSMC (TSM.N) says its own equipment purchases now run at ~1.9× the December-2025 plan (09-02) ✓ Bloomberg · 09-02; SEMI raised its 2028 WFE forecast to $220B (09-01) ✓ SEMI · 09-01; Q4 consensus EPS $4.06 sits above the company's own $4.02 guide ⚠ single source. The two questions left open at 08-14 — the flat margin guide, no 2027 number — are still open, not worse.

3. 進場條件成立了。 本頁 07-23 寫下的規則是兩個條件都要：股價跌進區間，而且找不到客戶砍 capex 的證據。現在進場區 < 約 $510（08-14 版為 < 約 $496，FY27 共識上修後依原框架重推導），現價低於上緣 14%。

3. The entry condition is met. The rule this page wrote down on 07-23 needs both: the price falls into the zone, and there is no evidence of any customer cutting capex. The entry zone is now < ~$510 (was < ~$496 at 08-14, re-derived under the same framework after the FY27 consensus was revised up), and the current price sits 14% below the top.

4. 最重要的警訊：現在看起來便宜，可能是週期頂端的陷阱。 34 倍仍是 AMAT 自己 2019–2024 歷史本益比帶（12.7×–23.5×）的約 1.5–2 倍 ✓ companiesmarketcap · 09-03。設備股在獲利最高峰時，算出來的本益比最好看；一旦景氣反轉、獲利下修，同一個股價會回頭變貴。所以這不是價值型進場，而是賭 AI 帶動的設備週期比過去長，本頁的變異觀點就是這件事，下方逐條寫下的絆線就是為這個賭注設的。還有一件事數字解釋不了：一個月 AMAT -14%，同期 LRCX -2%／KLAC -6%／SOXX -0.7% ⚠ 單一來源。市場可能正在為某個還沒顯現的隱憂替 AMAT 單獨定價，要盯。◆ framing = judgment定位＝判斷

4. The most important warning: looking cheap now could be the top-of-cycle trap. 34× is still ~1.5–2× Applied's own 2019–2024 trailing P/E band (12.7×–23.5×) ✓ companiesmarketcap · 09-03. Semicap stocks look their cheapest at peak earnings — once the cycle turns and earnings get cut, the same price becomes expensive again. So this is not a value entry; it is a bet that the AI-driven equipment cycle runs longer than past ones — the page's variant view — and the tripwires written below exist to guard exactly that bet. One thing the numbers don't explain: AMAT is down 14% in a month while LRCX is -2%, KLAC -6%, SOXX -0.7% ⚠ single source. The market may be pricing something about AMAT specifically that has not surfaced yet — worth watching. ◆ framing = judgment定位＝判斷

Numbers — estimates, revisions, and marking our own homework數字 — 預估、調整，以及替自己的作業打分

60× 50× 40× 30× 57× 55× 46× 42× 34× June build 06-30 07-23 08-14 09-03 $589 $713 $560 $534 $438 forward P/E (line) versus the share price at each build (below) — the multiple fell 40% while the price round-tripped and then broke down
Engine 2 (the multiple) has compressed at every single refresh of this page, through a +21% rally and a -40% drawdown alike. The June and 06-30 readings ran on a consensus EPS later corrected; from 07-23 the denominator is clean, and the 09-03 denominator is the same company math as 08-14 — so this leg of the compression is pure price.引擎二（本益比）在本頁每一次刷新時都在收縮 — 不論中間是 +21% 的上漲還是 -40% 的回檔。6 月與 06-30 兩次的分母用的是後來被修正的共識；07-23 起分母乾淨，09-03 這次的分母與 08-14 相同、都是公司自己的算式 — 所以這段收縮純粹來自股價。

Estimates — this update vs the last預估調整 — 本次 vs 前次

Period期間 Now (09-03)調整後（09-03） Prior (08-14)調整前（08-14） Δ差異 Consensus市場共識 Δ差異
FY2026E revenueFY2026E 營收$34.29B$34.29B—$34.25B (stockanalysis 09-01)+0.1%
FY2026E non-GAAP EPSFY2026E 非 GAAP EPS$12.75$12.75—$12.79 (stockanalysis 09-01)-0.3%
FY2027E revenueFY2027E 營收$46.13B$43.36B+6.4%$46.13B (MarketScreener 09-03)—
FY2027E non-GAAP EPSFY2027E 非 GAAP EPS$18.44 ⚠$17.12+7.7%$18.44—

⚠ FY2027E basis: S&P Global series (MarketScreener), ~3% above the stockanalysis series used at 08-14; like-for-like ≈$17.9 (+4.5%).

⚠ FY2027E 口徑：S&P Global 系列（MarketScreener），較 08-14 所用的 stockanalysis 系列高約 3%；同口徑約 $17.9（+4.5%）。

Basis, stated plainly.

口徑，講清楚。

Scorecard — FQ3 actual vs what this page carried計分卡 — FQ3 實際 vs 本頁先前所載

FQ3 FY2026FQ3 FY2026 Actual實際數 Carried on this page (07-23)本頁前估（07-23） Δ差異
Revenue營收$9,115M$8,950M+1.8%
Non-GAAP EPS非 GAAP EPS$3.50$3.36+4.2%
Non-GAAP gross margin非 GAAP 毛利率50.4%——

Marking this honestly: the "prior estimate" column is just the company's own guide midpoint, which this page adopted on 07-23 rather than modelling FQ3 independently. So the beat above is Applied beating Applied — it is not evidence that this page forecasts well. The page's actual call was the decision rule attached to that number (hold, with the guide floor as the trim tripwire), and that rule resolved correctly. Gross margin was never estimated here, so there is nothing to score. ◆

誠實打分：「本頁前估」那一欄其實就是公司自己的指引中值 — 07-23 時本頁直接沿用，並沒有獨立預估 FQ3。所以上面的超標是應材超越應材，不能當作本頁預測能力的證據。本頁真正下的判斷，是掛在那個數字上的決策規則（續抱，以指引下緣為減碼絆線），而那條規則判定正確。毛利率本頁從未預估，因此無從計分。◆

Next scoring: 11-12 FQ4 (this page's prior estimate = the guide midpoint $10.25B / $4.02; consensus $4.06 ⚠ TipRanks).

下一次打分：11-12 FQ4（本頁前估＝指引中值 $10.25B／$4.02；共識 $4.06 ⚠ TipRanks）

The three-year shape三年輪廓

Revenue $B $28.37B $34.29B +20.9% $46.13B ⚠ +34.5% FY2025A FY2026E FY2027E Non-GAAP EPS $ $9.42 $12.75 +35.4% $18.44 ⚠ +44.6% FY2025A FY2026E FY2027E Actuals are 10-K; FY2026E is 9M actual + FQ4 guide; FY2027E is consensus (basis gap ~3% ⚠)
The three-year shape: revenue (left) and non-GAAP EPS (right). ✓ 10-K · release 08-13 · MarketScreener 09-03三年輪廓：營收（左）與非 GAAP EPS（右）。✓ 10-K · 財報 08-13 · MarketScreener 09-03
FY2025A FY2026E FY2027E
Revenue營收$28.37B$34.29B (+20.9%)$46.13B (+34.5%)
Non-GAAP EPS非 GAAP EPS$9.42$12.75 (+35.4%)$18.44 (+44.6%) ⚠ basis ~3% gap口徑差約 3%
P/E at the 09-02 close以 09-02 收盤計本益比—~34×~24×

Probability-weighted outcomes機率加權情境

Bear 25% growth miss + compression Base 50% moderate growth, steady execution Bull 25% sustained gains + margin lift A subjective call, not a number calculated from data.
judgment These percentages are a subjective call, not a statistic. The June note here — that a fresh high argued for higher bear odds — played out in July's -24% pullback. The split is kept as originally framed. As of 09-03, the bear leg's content is: the market may be pricing a 2027 WFE peak — the classic semicap pattern where the multiple looks lowest right before estimates roll over ◆. The bull leg's content: the tape is 40% below its high while every customer datapoint since 08-14 points up.這三個百分比是主觀判斷，不是統計數字。6 月在此加註的「創新高應調高空方機率」，已在 7 月 -24% 回檔中兌現。機率分布維持原判。截至 09-03，空方那一腳的內容是：市場或許正在為 2027 年 WFE 見頂定價 — 半導體設備股的經典模式：本益比看起來最低的時刻，往往就在獲利預估開始下修之前 ◆。多方那一腳的內容：盤面距高點已跌 40%，而 08-14 以來所有客戶端數據都指向上。

Which engine gets the stock to double? From $438.46, doubling means $877, a market cap of ~$696B. A rising stock has only two paths: Engine 1 the company earning more (EPS growth), Engine 2 the market paying a higher P/E (re-rating).

股價要翻倍，靠哪台引擎？ 從 $438.46 翻倍是 $877，市值約 $696B。股票會漲只有兩條路：引擎 1 公司賺更多錢（EPS 成長），引擎 2 市場給更高的本益比（評價重估）。

Playbook (not investment advice)操作框架（非投資建議）

Added at the 2026-07-23 refresh, re-derived at 2026-08-14 against the post-print numbers, and re-derived again at 2026-09-03 as the price fell into the entry zone. Every row is this analysis's judgment ◆, derived from the numbers above. The page keeps its original no-target-price qualitative frame.

2026-07-23 新增，2026-08-14 依財報後數字重推導，2026-09-03 依股價跌入進場區重推導。每一列皆為本研究判斷 ◆，由上方數字推導。本頁維持原始的「無目標價」定性框架。

$560 $520 $480 $440 Entry zone top (09-03) $510 08-14 zone $496 $534.54 $492.32 first close under $496 $438.46 08-13 08-21 08-28 09-02 Daily closes vs. the entry zone — 08-21 closed under the 08-14 zone ($496), 09-02 is 14% below $510 today
Daily closes, 08-13 print through 09-02, against the pre-committed entry zone. ✓ IBKR · 09-0208-13 財報後至 09-02 的每日收盤價，對照預先寫下的進場區。✓ IBKR · 09-02
If not holding未持有◆ Entry condition is met, staged in three tranches.
  • Both checks pass: fundamentals (FQ3 beat, FQ4 guided up, no evidence of any customer cutting equipment budgets, re-checked 09-03 and still holding); price ($438.46 is 14% below the zone's upper edge $510).
  • Tranche 1: at the current price. Tranche 2: after Micron's 09-30 report or TSMC's October call confirms customers haven't cut budgets. Tranche 3: after management clarifies 2027 at FQ4 (11-12).
  • If any invalidation condition fires before the next tranche, the remaining tranches are cancelled.
進場條件成立，分三批。
  • 雙重驗證都過：基本面（FQ3 超標、FQ4 指引上修、找不到任何客戶砍設備預算的證據，09-03 重查仍成立）；價格（$438.46 低於進場區上緣 $510 有 14%）。
  • 第 1 批：現價。第 2 批：09-30 美光財報或 10 月台積電法說確認客戶沒砍預算之後。第 3 批：11-12 FQ4 經營層把 2027 講清楚之後。
  • 任一失效條件在下一批之前觸發，後面的批次取消。
If holding已持有◆ Hold — the first rung of the add ladder is satisfied.
  • The add rule written on 07-23: price < ~$490, and FQ3 confirms guidance with no evidence of customer capex cuts. Both conditions now hold.
  • Default action: execute the first rung at the pre-committed proportion; the second rung waits on the same 09-30 / October checkpoints.
  • Trim tripwires unchanged: FQ4 (11-12) revenue below the $9.75B guide floor, or FY27 framing implying deceleration; any customer capex cut ≥15%.
續抱，加碼階梯第一階成立。
  • 07-23 寫下的加碼規則：股價 < 約 $490，而且 FQ3 確認指引、無客戶砍 capex 證據。兩個條件現在都成立。
  • 預設動作：依預先承諾的比例執行第一階；第二階同樣等 09-30／10 月的定點檢查。
  • 減碼絆線不變：11-12 FQ4 營收低於指引下緣 $9.75B，或 FY27 說法隱含減速；任一客戶砍 capex ≥15%。
Entry zone進場區◆ < ~$510, framework unchanged.
  • 40× FY26E $12.75 = $510; 29× FY27E $18.44 = $535 (like-for-like $17.9 = $519); the tighter leg binds, and it is now the FY26 leg. ✗→ was < ~$496 (29× × $17.12)
  • The zone rising was the pre-committed reaction written at 08-14 (FY27E ≥ ~$17.6 → re-derive), not a moved goalpost.
  • Current price $438.46 is 14.0% below the top, and 11.6% below the old $496 top.
< 約 $510，框架不變。
  • 40× FY26E $12.75 ＝ $510；29× FY27E $18.44 ＝ $535（同口徑 $17.9 ＝ $519）；取較嚴的一腳，現在是 FY26 那一腳。✗→ 原為 < 約 $496（29× × $17.12）
  • 區間上移是 08-14 就預先寫好的反應（FY27E ≥ 約 $17.6 → 重推導），不是事後移球門。
  • 現價 $438.46 低於上緣 14.0%，低於舊上緣 $496 有 11.6%。
Why not more aggressive為何不更激進◆ Not shorting, and not backing up the truck either.
  • Not shorting: nothing in the fundamentals supports it.
  • Not going all-in, three reasons: (1) 34× is still 1.5-2x AMAT's own 2019-2024 trailing P/E band (12.7x-23.5x) — this is a bet on a longer cycle, not a bargain; (2) -14% in a month against peers LRCX -2% / KLAC -6% / SOXX -0.7% ⚠ single source — the market may be pricing an undisclosed concern (flat margin, China at 26% and falling, a possible 2027 peak); (3) three key events land within ten weeks (09-30, October, 11-12) — staging trades a lower cost for more information.
不放空，也不一次重壓。
  • 不放空：基本面找不到任何一點支持。
  • 不一次重壓，三個理由：(1) 34 倍仍是 AMAT 2019–2024 歷史本益比帶（12.7×–23.5×）的 1.5–2 倍，這是賭週期拉長，不是撿便宜；(2) 一個月 -14%，同業 LRCX -2%／KLAC -6%／SOXX -0.7% ⚠ 單一來源，市場可能在反映還沒公開的隱憂（毛利率持平、中國占比 26% 續降、2027 見頂）；(3) 十週內三個關鍵事件（09-30、10 月、11-12），分批用更低的代價換更多資訊。
Invalidation失效條件◆ Four red lines — any one firing means stop and reassess.
  • (1) FQ4 (11-12) revenue below the $9.75B guide floor, or FY27 framing implying a QoQ decline → trim, review the view.
  • (2) Any of Samsung / SK hynix / Micron / TSMC cuts capex ≥15% → the FY27 premise breaks → bear case.
  • (3) Gross margin fails to resume expansion for two straight quarters → the pricing-power leg is damaged.
  • (4) China hit exceeds $600M, or a new BIS / VEU action → currently at the line, not yet triggered (yellow).
  • Upside (5): management quantifies 2027 at FQ4 → re-derive the entry zone again.
四條紅線，任一觸發就先停手。
  • (1) 11-12 FQ4 營收低於指引下緣 $9.75B，或 FY27 說法隱含季減 → 減碼、觀點重審。
  • (2) 三星／SK海力士／美光／台積電任一砍 capex ≥15% → FY27 前提破功 → 空方情境。
  • (3) 毛利率連續兩季未重回擴張 → 定價力那一腳受損。
  • (4) 中國衝擊超過 $600M，或新的 BIS／VEU 行動 → 目前在線上、尚未觸發（黃）。
  • 上行 (5)：FQ4 經營層量化 2027 → 再次重推導進場區。

One line: the entry light is green, but the market still has an unexplained worry, so this stages across three tranches and three events; if any of the four big customers cuts spending more than 15%, it trims and reviews per the tripwires. ◆

一句話：進場燈亮了，但市場還有沒說清楚的隱憂，所以分三批、看三個事件；四大客戶任一砍預算超過 15%，就依絆線減碼、重審。 ◆

The -40% checklist股價 -40% 檢查清單

Check檢查項目 Where to look去哪裡查 09-03 verdict09-03 判定 Evidence依據
1. Did any customer cut capex (≥15%)?1. 客戶有沒有砍 capex（≥15%）？Samsung / SK hynix / Micron / TSMC calls & board decisions三星／SK海力士／美光／台積電法說與董事會決議Green綠SK hynix ₩54T new fabs (08-07); TSMC equipment purchases ~1.9× plan (09-02); Micron 09-30 pending.SK 海力士 ₩54兆新廠（08-07）；台積電設備採購約 1.9× 計畫（09-02）；美光 09-30 待公布。
2. Have the company's own near-term numbers weakened?2. 公司自己的近端數字有沒有轉弱？Results, guidance, next-quarter consensus財報、指引、下一季共識Green綠FQ3 beat; FQ4 guide +12.5% QoQ; Q4 consensus $4.06 > $4.02 guide ⚠.FQ3 超標；FQ4 指引季增 12.5%；Q4 共識 $4.06 高於指引 $4.02 ⚠。
3. Which way is the industry's equipment forecast moving?3. 產業設備預測方向？SEMI / TrendForce WFE forecasts, lead timesSEMI／TrendForce WFE 預測、交期Green綠SEMI 2028 WFE raised to $220B (09-01); lead times 12-24 months (08-19).SEMI 2028 WFE 上修至 $220B（09-01）；交期 12-24 個月（08-19）。
4. Has China / export control newly deteriorated?4. 中國／出口管制有沒有新惡化？BIS notices, VEU list, company callsBIS 公告、VEU 名單、公司法說Yellow黃No new action; FY26 hit ~$600M sits at the line; China 26% and falling.無新行動；FY26 衝擊約 $600M 在線上；中國占比 26% 且續降。
5. Has the gross-margin structure broken?5. 毛利率結構有沒有破？Quarterly gross margin, guidance季報毛利率、指引Yellow黃FQ4 guided flat at 50.4%; the 13-quarter YoY streak is intact but sequential expansion paused.FQ4 指引持平於 50.4%；連續 13 季年增紀錄未斷，但季增擴張暫停。
6. Is the decline AMAT-specific?6. 跌勢是不是 AMAT 獨有？1-month AMAT vs LRCX / KLAC / SOXX一個月 AMAT vs LRCX／KLAC／SOXXYellow黃-14% vs -2% / -6% / -0.7% ⚠ single source; no datapoint explains it — the one place on this list where the market may know something the numbers don't.-14% vs -2%／-6%／-0.7% ⚠ 單一來源；沒有數據解釋 — 這張清單裡唯一一項，市場可能知道數字還沒顯現的事。

0 red, 3 yellow, 3 green → per the rule this page wrote down on 07-23, what fell is the valuation, not the thesis; the price condition is met; the three yellows are the reason to stage the entry, not lump it. Any one turning red → stop and re-review. ◆

0 紅、3 黃、3 綠 → 依本頁 07-23 寫下的規則，跌的是評價、不是論點；價格條件成立；三個黃是「分批」而不是「一次」的理由。任何一項轉紅 → 先停手、重審。 ◆

Standing instruction: re-run this table on the next 10% drawdown, and don't re-read the news to do it.

股價再跌 10% 時，重走一次這張表，不要重看新聞。

Catalysts — threshold → reaction催化劑 — 門檻 → 反應

When時間Event事件Threshold → reaction門檻 → 反應Status狀態
2026-08-13FQ3 FY26 resultsFQ3 FY26 財報Rev ≥ $8.95B mid & FQ4 guide accelerating → base confirmed; below $8.45B floor → trim per playbook
Actual: $9,115M (+1.8% vs mid), EPS $3.50 (+4.2%), FQ4 guide $10.25B = +12.5% QoQ. Base confirmed; no trim. Upside invalidation (FQ4 >$10B) also fired → entry zone re-derived.營收 ≥ 中值 $8.95B 且 FQ4 指引續增 → 基準確認；低於下緣 $8.45B → 依框架減碼
實際：$9,115M（高於中值 1.8%）、EPS $3.50（高於 4.2%）、FQ4 指引 $10.25B ＝ 季增 12.5%。基準確認，不減碼。上行失效條件（FQ4 >$10B）同步觸發 → 進場區已重新推導。		✓ HIT已達標
2026-08 (late)2026-08 下旬Post-print FY27 consensus reset財報後 FY27 共識重設FY27E non-GAAP EPS revised to ≥ ~$17.6 → the 29× leg of the entry zone reaches $510 → price condition met at the current price; below that → keep waiting for < ~$496FY27E 非 GAAP 每股盈餘上修至 ≥ 約 $17.6 → 進場區 29× 那一腳到 $510 → 價格條件在現價即成立；低於此 → 繼續等 < 約 $496✗→ closed 09-03已結案 09-03
FY27E $18.44 (MarketScreener/S&P Global, 09-03); like-for-like ≈$17.9 — both readings ≥ $17.6 → the 29× leg rises to ≥ $519 → the entry zone is now set by the 40× FY26E leg at $510.FY27E $18.44（MarketScreener／S&P Global，09-03）；同口徑約 $17.9 — 兩種讀法都 ≥ $17.6 → 29× 那一腳升到 ≥ $519 → 進場區改由 40× FY26E 的 $510 決定。
2026-09-08 / 2026-09-09Citi Global TMT (CFO Hill) / Goldman Sachs Communacopia + Technology (CEO Dickerson)Citi Global TMT（財務長 Hill）／高盛 Communacopia + Technology（執行長 Dickerson）Threshold: any change in tone on FQ4 guidance or the calendar-2026 framework → unchanged = information event; downgraded = invalidation (1) triggers early門檻：FQ4 指引或 2026 曆年框架有無改口 → 維持＝資訊事件；下修＝失效條件 (1) 提前觸發scheduled已排定
2026-09-30Micron FQ4 FY26 results (first FY27 capex language)美光 FQ4 FY26 財報（首度 FY27 capex 語言）Threshold: FY27 capex ≥ FY26's ~$27B → the memory leg is supported; cut ≥15% → invalidation (2)門檻：FY27 capex ≥ FY26 的約 $27B → 記憶體那一腳有撐；砍 ≥15% → 失效條件 (2)scheduled已排定
2026-10TSMC Q3 call台積電 Q3 法說First 2027 capex language ≥ 2026's $60-64B → FY27E supported; "digestion" language → FY27E at risk2027 capex 初步語言 ≥ 2026 的 $60-64B → FY27E 有撐；出現「消化」字眼 → FY27E 有險scheduled已排定
2026-11-12FQ4 FY26 results + first FY27 framingFQ4 FY26 財報＋首度 FY27 說法Rev ≥ $9.75B guide floor & 2027 quantified → the FY27 leg stops being an assumption; below the floor or deceleration language → trim per playbook營收 ≥ 指引下緣 $9.75B 且 2027 被量化 → FY27 那一腳不再只是假設；低於下緣或出現減速語言 → 依框架減碼scheduled已排定 (TipRanks marks it confirmed ⚠TipRanks 標示已確認 ⚠)
2026-12SEMI year-end WFE forecast; China VEU annual reviewsSEMI 年末 WFE 預測；中國 VEU 年審2027 WFE revised above $135B → bull tilt; license tightening → China-hit > $600M revisit2027 WFE 上修破 $135B → 偏多；許可收緊 → 中國衝擊 >$600M 重估scheduled已排定
ongoing持續Memory-maker capex guidance (Samsung/SK hynix/Micron)記憶體三雄 capex 指引Any cut ≥15% → invalidation row (2) fires任一下修 ≥15% → 失效條件 (2) 觸發monitoring監測中
2H26+EssilorLuxottica AR-opticsPure information event — optionality, no threshold純資訊事件 — 選擇權價值，無門檻watching觀察

Update log — material news since publish更新紀錄 — 發布後的重要消息

Dated additions, newest first. The core thesis above stays as-is unless an entry says otherwise.

逐則標日期、最新在上。除非條目另外註明，上方核心論點維持不變。

✓ 2026-09-03 -40% from the high, and now inside the page's own entry zone — the price condition is met, the evidence condition re-checked green, the multiple at ~34× 距高點 -40%、跌進本頁自己畫的進場區 — 價格條件成立、證據條件重查仍綠、本益比約 34 倍

一句話：股價從 08-13 的 $534.54 跌到 09-02 的 $438.46（-18%），跌破本頁 08-14 寫下的進場區上緣 $496；這三週沒有任何客戶砍 capex、沒有新的出口管制、FY27 共識反而上修 — 所以本頁依 07-23 就寫好的規則，把「未持有」從「不追」改成「分批進場」，把「已持有」的加碼階梯第一階判定為成立。

One line: the price fell from the 08-13 close $534.54 to the 09-02 close $438.46 (-18%), breaking below the entry-zone top this page wrote down on 08-14 ($496); in those three weeks no customer cut capex, no new export control landed, and the FY27 consensus was revised up instead — so per the rule this page wrote down on 07-23, "if not holding" changes from "don't chase" to "stage an entry," and the first rung of the "if holding" add ladder is judged satisfied.

1. What happened (the tape)

  • 08-13 $534.54 → 09-02 $438.46, -18.0% over 15 sessions; 09-02 intraday $433.57 set a fresh 13-week low. ✓ IBKR
  • 08-14's close $507.18 was already below the current top $510; 08-21's $492.32 was the first close inside the 08-14 zone (< $496).
  • Two dated causes: 08-14 -5.1% (flat margin guide); 08-28 -4.3% (CFO's Form 4 sale of 7,500 sh, ~$3.59M ✓ Form 4 · 08-25, plus sector profit-taking). Peers over the same span: LRCX -2%, KLAC -6%, SOXX -0.7% ⚠ single source.

1. 盤面

  • 08-13 $534.54 → 09-02 $438.46，15 個交易日 -18.0%；09-02 盤中 $433.57 創 13 週新低。✓ IBKR
  • 08-14 收盤 $507.18 已低於現在的上緣 $510；08-21 收 $492.32 首次收進 08-14 版區間（< $496）。
  • 兩個標日期的跌因：08-14 -5.1%（毛利率指引持平）；08-28 -4.3%（財務長 Form 4 賣 7,500 股約 $3.59M ✓ Form 4 · 08-25，加上族群獲利了結）。同期同業 LRCX -2%／KLAC -6%／SOXX -0.7% ⚠ 單一來源。

2. Fundamentals in these three weeks

  • TSMC equipment purchases ~1.9× plan (09-02) ✓ Bloomberg／Taipei Times; SK hynix ₩54T new fabs (08-07) ✓ CNBC; SEMI raised 2028 WFE to $220B (09-01) ✓ SEMI; lead times 12-24 months ✓ TrendForce · 08-19.
  • Consensus revised up: FY26 EPS $12.79 ✓ stockanalysis · 09-01; FY27 EPS $18.44 ✓ MarketScreener · 09-03 (like-for-like ≈$17.9, basis explained in the numbers section); sell-side average target $640.89.
  • 08-31 Seoul briefing: 3D scaling, hybrid bonding, HBM thermal materials, CPO waveguide materials ✓ THE ELEC／ZDNet Korea／Korea Times — technical explainer in the 3D-scaling section of the full case.

2. 基本面這三週

  • 台積電設備採購約 1.9× 計畫（09-02）✓ Bloomberg／Taipei Times；SK 海力士 ₩54兆 新廠（08-07）✓ CNBC；SEMI 2028 WFE 上修至 $220B（09-01）✓ SEMI；設備交期 12–24 個月 ✓ TrendForce · 08-19。
  • 共識上修：FY26 EPS $12.79 ✓ stockanalysis · 09-01；FY27 EPS $18.44 ✓ MarketScreener · 09-03（同口徑約 $17.9，口徑說明見數字段）；賣方平均目標價 $640.89。
  • 08-31 首爾說明會：3D scaling、hybrid bonding、HBM 導熱材料、CPO 光波導材料 ✓ THE ELEC／ZDNet Korea／Korea Times，技術解釋見完整論證的 3D scaling 段。

3. What weighs against it

  • The China hit, ~$600M, sits right on this page's >$600M revisit line.
  • The -14% one-month move is AMAT-specific, with no data explaining it.
  • 34× is still 1.5-2x the 2019-2024 trailing band. ✓ companiesmarketcap · 09-03

3. 調降的一面

  • 中國衝擊約 $600M，正好壓在本頁 >$600M 的重估線上。
  • 一個月 -14% 是 AMAT 獨有，找不到數據解釋。
  • 34 倍仍是 2019–2024 歷史帶的 1.5–2 倍 ✓ companiesmarketcap · 09-03。

4. What strengthens it

  • Zero customer capex cuts; three of four invalidation lines moved favorably.
  • FY27 consensus revised up; Q4 consensus $4.06 above the $4.02 guide. ⚠ single source

4. 強化的一面

  • 零客戶砍單；四條失效線裡三條朝有利方向。
  • FY27 共識上修；Q4 共識 $4.06 高於指引 $4.02 ⚠ 單一來源。

5. The rule, followed

  • 07-23 entry rule (price + evidence) both green → if not holding: staged entry, tranches tied to 09-30 / October / 11-12.
  • 07-23 add rule (< ~$490 and FQ3 confirmed) → if holding: rung 1 satisfied.
  • 08-14 catalyst rule (FY27E ≥ ~$17.6 → re-derive) → HIT, entry zone < ~$510.
  • -40% checklist: 0 red / 3 yellow / 3 green. Unchanged: no target price, 25/50/25, trim tripwires.

5. 規則照做

  • 07-23 進場規則（價格＋證據）兩半皆綠 → 未持有：分批進場，各批綁 09-30／10 月／11-12。
  • 07-23 加碼規則（< 約 $490 且 FQ3 確認）→ 已持有：第一階成立。
  • 08-14 催化劑規則（FY27E ≥ 約 $17.6 → 重推導）→ HIT，進場區 < 約 $510。
  • -40% 檢查清單 0 紅／3 黃／3 綠。沒改的：無目標價、25/50/25、減碼絆線。

6. Data snapshot

Close $438.46 (09-02) · YTD +71.0% · off high -40.7% · market cap $348.0B · fwd P/E ~34×/~24× · trailing P/E ~38× · entry zone < ~$510 (14% below the top).

6. 數據快照

收盤 $438.46（09-02）· YTD +71.0% · 距高點 -40.7% · 市值 $348.0B · 預估本益比 ~34×／~24× · 歷史本益比 ~38× · 進場區 < 約 $510（低於上緣 14%）。

Bottom line: view — no-target-price qualitative frame (unchanged) | entry zone < ~$510 (was < ~$496) | price $438.46, 14% below the zone top | if holding: hold, add-ladder rung 1 satisfied; if not holding: staged entry. ◆ framing = judgment

結論：觀點 — 無目標價定性框架（不變）｜進場區 < 約 $510（原 < 約 $496）｜現價 $438.46，低於進場區上緣 14%｜已持有：續抱＋加碼階梯第一階成立；未持有：分批進場。 ◆ 定位＝判斷

Earlier entries (2026-08-14, 07-23, 06-30, 06-16)先前條目（2026-08-14、07-23、06-30、06-16）

✓ 2026-08-14 FQ3 beat, FQ4 guided to $10.25B — and the stock fell 5%. The tripwire resolved HIT; the entry zone moved up as pre-committed FQ3 超標、FQ4 指引 $10.25B — 股價卻跌 5%。絆線判定為 HIT；進場區依 07-23 的預先承諾上移

The print (FQ3 FY2026, reported after the 08-13 close). Revenue $9,115M, +25% YoY and +15% sequentially — management calls it the highest sequential revenue growth in the company's history — against a guide midpoint of $8,950M. Non-GAAP EPS $3.50 (+41% YoY) vs the $3.36 guide; GAAP EPS $3.17 (+43%). GAAP gross margin 50.3% / non-GAAP 50.4%, the 13th consecutive quarter of year-over-year gross-margin expansion. Non-GAAP operating margin 34.0% (+330bp YoY). Operating cash flow $3,037M, free cash flow $2,330M, capex $707M; $860M returned ($440M buybacks + $420M dividends). ✓ Q3 FY26 release / 8-K · 08-13Q3 FY26 財報／8-K · 08-13

財報內容（FQ3 FY2026，08-13 盤後公布）。營收 $9,115M，年增 25%、季增 15% — 經營層稱這是公司史上最高的單季環比成長 — 指引中值為 $8,950M。非 GAAP 每股盈餘 $3.50（年增 41%），指引為 $3.36；GAAP 每股盈餘 $3.17（年增 43%）。GAAP 毛利率 50.3%／非 GAAP 50.4%，為連續第 13 季毛利率年增。非 GAAP 營益率 34.0%（年增 3.3 個百分點）。營運現金流 $3,037M、自由現金流 $2,330M、資本支出 $707M；回饋股東 $860M（庫藏股 $440M＋股利 $420M）。✓ Q3 FY26 財報／8-K · 08-13

Segments and mix. Semiconductor Systems $7,040M (+27% YoY, +18% QoQ); Applied Global Services $1,781M (+22% YoY); Other, which now carries Display, $294M with a $118M operating loss. Semiconductor Systems mix: foundry/logic 67% · DRAM 26% · flash 7%. China fell to 26% of Semiconductor Systems + AGS revenue, from 27% in FQ2 ✓ release + call · 08-13財報＋法說 · 08-13. Nine months of FY26: revenue $24,037M, non-GAAP EPS $8.73.

各部門與產品組合。半導體系統 $7,040M（年增 27%、季增 18%）；AGS 服務 $1,781M（年增 22%）；其他（現含顯示器）$294M，營業損失 $118M。半導體系統組合：晶圓代工／邏輯 67% · DRAM 26% · 快閃記憶體 7%。中國占半導體系統＋AGS 營收比重降至 26%（FQ2 為 27%）✓ 財報＋法說 · 08-13。FY26 前三季合計：營收 $24,037M、非 GAAP 每股盈餘 $8.73。

The guide, and the sentence the market actually traded on. FQ4: revenue $10.25B ±$0.5B (+12.5% QoQ, +51% YoY) and non-GAAP EPS $4.02 ±$0.20 (+85% YoY) — an acceleration, not a fade. But non-GAAP gross margin is guided to 50.4%, flat sequentially, on ramp costs (hiring, added resources) and a heavier display mix, with non-GAAP opex ~$1.58B. CFO Brice Hill also modelled the non-GAAP tax rate at ~11% in FQ4 stepping to ~13% in 2027 as the global minimum tax lands. Bernstein pressed directly on the flat margin; UBS pressed on whether growth decelerates into year-end. The stock closed 08-13 at $534.54 (-2.5% on the day, before the print) and traded down ~5% after hours. ✓ guidance + call transcript · 08-13指引＋法說逐字稿 · 08-13

指引，以及市場真正在交易的那句話。FQ4：營收 $10.25B ±$0.5B（季增 12.5%、年增 51%）、非 GAAP 每股盈餘 $4.02 ±$0.20（年增 85%）— 是加速，不是轉弱。但非 GAAP 毛利率指引 50.4%、與前季持平，原因是產能拉升成本（招募、資源投入）與顯示器占比提高，非 GAAP 營業費用約 $1.58B。財務長 Brice Hill 另外把非 GAAP 稅率設在 FQ4 約 11%、2027 年約 13%（全球最低稅負制上路）。Bernstein 直接質疑毛利率為何持平，UBS 追問下半年成長是否減速。08-13 收盤 $534.54（當日 -2.5%，尚在財報前），盤後再跌約 5%。✓ 指引＋法說逐字稿 · 08-13

What management put on the record. Gary Dickerson: "Applied Materials delivered another record-breaking quarter, including the highest sequential revenue growth in the company's history. As rapid global AI adoption drives demand for our materials engineering solutions, we are further raising our Semiconductor Systems revenue expectations for calendar 2026." On the shape of the cycle: "Leading-edge foundry logic, DRAM, and advanced packaging have the greatest impact on AI computing performance, power efficiency, and cost. Together, we expect these areas to represent around 80% of wafer fab equipment growth in 2026 and 2027." Hill on 2027: "When we look into 2027, we expect this whole demand function led by AI to continue… we expect another strong year." Calendar-2026 growth framing: advanced packaging >70%, process diagnostics and control >50%, AGS >20%; manufacturing capacity doubling by 2028. ✓ PR + prepared remarks + transcript · 08-13新聞稿＋書面談話＋逐字稿 · 08-13

經營層說了什麼。執行長 Gary Dickerson：「應材再次交出破紀錄的一季，包含公司史上最高的單季環比營收成長。隨著全球 AI 快速普及帶動我們材料工程解決方案的需求，我們再次上修 2026 曆年的半導體系統營收預期。」關於這輪循環的形狀：「先進邏輯、DRAM 與先進封裝，對 AI 運算的效能、能效與成本影響最大。我們預期這三塊合計將占 2026 與 2027 年晶圓廠設備（WFE）成長的約 80%。」財務長 Hill 談 2027：「看向 2027，我們預期這整個由 AI 帶動的需求會延續……會是又一個強勁的年份。」曆年 2026 成長框架：先進封裝 >70%、製程檢測與控制 >50%、AGS >20%；製造產能將於 2028 年前倍增。✓ 新聞稿＋書面談話＋逐字稿 · 08-13

Decision layer — executed, not re-argued. The 07-23 catalyst read "rev ≥ $8.95B mid & FQ4 guide accelerating → base confirmed." Actual $9.115B and a +12.5% QoQ guide: resolved HIT, base confirmed. The upside invalidation read "FQ4 guide >$10B → the entry zone revises up." $10.25B cleared it, so the zone is re-derived from the unchanged 40× FY26E / 29× FY27E framework: 40 × $12.75 = $510, 29 × $17.12 = $496. The tighter leg binds, so the zone becomes < ~$496 (was < ~$490). This matters procedurally: the raise was written down on 07-23, before the print — a mechanical consequence of higher EPS, not a goalpost moved to justify buying. The FY27 denominator is still the pre-print consensus and will likely be revised up; a new catalyst now tracks that explicitly.

決策層 — 執行，不是重新辯論。07-23 的催化劑寫的是：「營收 ≥ 中值 $8.95B 且 FQ4 指引續增 → 基準確認。」實際 $9.115B、FQ4 指引季增 12.5%：判定 HIT，基準確認。上行失效條件寫的是：「FQ4 指引 >$10B → 進場區上修。」$10.25B 已越過，因此依未變動的 40× FY26E／29× FY27E 框架重新推導：40 × $12.75 = $510、29 × $17.12 = $496。取較嚴的一腳，進場區成為 < 約 $496（原為 < 約 $490）。這在程序上很重要：上修是 07-23 財報前就白紙黑字寫下的，是 EPS 變高的機械結果，不是為了合理化買進而事後移動的球門。FY27 分母仍是財報前共識、預期會被上修；已新增一條催化劑專門追蹤這件事。

Numbers refreshed. Price $560.00 → $534.54 (08-13 close; pre-market 08-14 ~$507.55, ~-5.0%) ✓ IBKR; YTD +118.2% → +108.3%; drawdown from $739.67 -24.3% → -27.7% (-31.4% at the pre-market mark); market cap $444.6B → $424.4B (793,959,430 shares — the FQ3 10-Q was not yet filed, so the 04-26 cover count still stands) ✓ SEC / stockanalysis; forward P/E ~46× → ~42×, and the FY26E denominator moves from consensus $12.26 to the company's own $12.75 (9M actual $8.73 + FQ4 guide mid $4.02) ✗→ basis upgraded from consensus to reported+guide口徑由共識升級為實績＋指引; FY27 consensus EPS $16.75 → $17.12 and revenue $42.59B → $43.36B (stockanalysis 39 analysts, 08-13, pre-print); street average target $623.06 → $634.94 (range $358–900, opinions) ✓ stockanalysis · 08-13; trailing P/E ~52× → ~46× once FQ3's $3.17 replaces the year-ago quarter in the TTM. FY26E revenue is now $34.29B (+20.9% YoY) on the same reported-plus-guide basis.

數字已更新。股價 $560.00 → $534.54（08-13 收盤；08-14 盤前約 $507.55，約 -5.0%）✓ IBKR；年初至今 +118.2% → +108.3%；距 $739.67 高點 -24.3% → -27.7%（以盤前價計 -31.4%）；市值 $444.6B → $424.4B（793,959,430 股 — FQ3 的 10-Q 尚未申報，仍沿用 04-26 封面股數）✓ SEC／stockanalysis；預估本益比 ~46× → 約 42×，且 FY26E 分母由共識 $12.26 改為公司自己的 $12.75（前三季實績 $8.73＋FQ4 指引中值 $4.02）✗→ 口徑由共識升級為實績＋指引；FY27 共識每股盈餘 $16.75 → $17.12、營收 $42.59B → $43.36B（stockanalysis 39 位分析師，08-13，財報前）；賣方平均目標價 $623.06 → $634.94（區間 $358–900，屬意見）✓ stockanalysis · 08-13；歷史本益比 ~52× → 約 46×（把 FQ3 的 $3.17 換掉去年同季後的 TTM）。FY26E 營收以同一口徑為 $34.29B（年增 20.9%）。

How it bears on the thesis. The near-term bear leg is dead: there was no miss, and the guide accelerated. What replaces it is narrower and more honest — a margin question (why does record revenue buy zero sequential gross-margin expansion?) and a 2027 quantum question (management has given direction and an 80%-of-WFE-growth shape, but no number, and a ~2pt tax step-up quietly works against FY27 EPS). Meanwhile the multiple did the thing this page kept saying had to happen: it compressed again, to ~42×, while earnings were marked up. The headline finding stands as written — the re-rating did largely happen, and it has since partly reversed — but "priced for perfection" is retired as the operative description. This update also scores the FQ3 catalyst, adds the first revision and scorecard tables (the model file now has a prior snapshot to compare against), and collapses the narrative sections behind 完整論證 per the standing page format. ◆ framing = judgment定位＝判斷

對論點的影響。近端的空方腳已經死了：沒有 miss，指引還加速。取而代之的問題更窄、也更誠實 — 一個是毛利率問題（為什麼創紀錄的營收換不到任何季增的毛利率？），一個是2027 的量問題（經營層給了方向、也給了「占 WFE 成長 80%」的形狀，但沒給數字；而稅率約 2 個百分點的墊高正悄悄扣 FY27 的每股盈餘）。同時，本益比做了本頁一直說必須發生的事：再次收縮到約 42 倍，而獲利被往上修。開頭的結論維持原文 — 重估確實大半發生過，之後又部分回吐 — 但「完美定價」這個描述正式退休。本次更新同時判定 FQ3 催化劑、新增第一版預估調整表與計分卡（模型檔已有前期快照可比），並依標準頁面格式將論證段落收合進「完整論證」。◆ 定位＝判斷

✓ 2026-07-23 -24% in three weeks: the "any wobble" bear leg fired — on sentiment, not on AMAT's numbers 三週 -24%：「一有閃失」的空方句兌現了 — 但閃失來自情緒，不是 AMAT 的數字

The tape. $723.00 all-time closing high 06-30 → $739.67 intraday ~07-01 → -7.4% on 07-02 (report that SK Hynix is slowing part of its HBM4 expansion) → ~-10% on 07-07 (Samsung's record ₩89.4T preliminary OP still missed the "AI bar") → $529.66 trough 07-17 (~-27% off the closing high) → rebound after CEO Gary Dickerson described customer equipment forecasts extending "to 2030" → $560.00 close 07-22 ✓ IBKR · 07-22 + CNBC/MacroTrends. Michael Burry disclosed a short position at ~$729 on 06-30 ⚠ single outlet單一來源.

盤面。06-30 收盤歷史高 $723.00 → 約 07-01 盤中 $739.67 → 07-02 -7.4%（SK海力士(000660.KS) 傳出放緩部分 HBM4 擴產）→ 07-07 約 -10%（三星(005930.KS) 初步營益 ₩89.4兆 創紀錄仍不及「AI 標準」）→ 07-17 低點 $529.66（距收盤高點約 -27%）→ CEO Gary Dickerson 稱客戶設備需求預測「看到 2030」後反彈 → 07-22 收 $560.00 ✓ IBKR · 07-22 + CNBC/MacroTrends。Michael Burry 於 06-30 揭露在約 $729 的空頭部位 ⚠ 單一來源。

What strengthened while the stock fell. FQ2 FY26 (reported 05-14): record revenue $7.91B (+11% YoY), non-GAAP EPS $2.86 (+20%), China down to 27% of revenue; FQ3 guide $8.95B ±$0.5B / EPS $3.36 ±$0.20 — implying ~+13% QoQ ✓ GlobeNewswire 05-14. TSMC(TSM.N) raised 2026 capex to $60-64B from $52-56B (07-16); Micron(MU.O) lifted FY26 capex to ~$27B; SK hynix raised $26.5B on NASDAQ earmarked for fabs + EUV ✓ transcripts/filings 06-24~07-16. FQ3 EPS revisions past 90 days: 25 up, 0 down ✓ Investing.com 07-21. July target hikes into the selloff: TD Cowen $700 · UBS $705 · Stifel $650 · GS $645; street average $623.06 (39 analysts) ✓ stockanalysis · 07-15.

股價下跌的同時，基本面反而變強。FQ2 FY26（05-14 公布）：營收 $7.91B 創紀錄（YoY +11%）、非 GAAP EPS $2.86（+20%）、中國占比降至 27%；FQ3 指引 $8.95B ±$0.5B／EPS $3.36 ±$0.20 — 隱含 QoQ 約 +13% ✓ GlobeNewswire 05-14。台積電(TSM.N) 2026 capex 自 $52-56B 上修至 $60-64B（07-16）；美光(MU.O) FY26 capex 升至約 $27B；SK海力士在 NASDAQ 募得 $26.5B 全數投入廠房與 EUV ✓ 法說/申報 06-24~07-16。90 天內 FQ3 EPS 修正：上修 25 次、下修 0 次 ✓ Investing.com 07-21。7 月逆勢調升目標價：TD Cowen $700 · UBS $705 · Stifel $650 · 高盛 $645；賣方平均 $623.06（39 位）✓ stockanalysis · 07-15。

Numbers refreshed. Price $713.00 → $560.00 (-21.5% vs the 06-30 build; IBKR 07-22); YTD +177.8% → +118.2%; market cap ~$560B → $444.6B (793,959,430 shares, 10-Q 04-26 ✓ SEC); forward P/E ~55× → ~46× — and the FY26 consensus EPS denominator is corrected to $12.26 (stockanalysis, 39 analysts, 07-15) ✗→ replaces the ~$13.0 secondary figure取代 ~$13.0 次級數字; FY27 consensus EPS $16.75 (+37%) → ~33× at the live price. CEO Dickerson sold 78,321 shares for ~$55.5M at ~$700-736 on 06-29/30 — at the top, filed 07-01 ✓ SEC Form 4.

數字已更新。股價 $713.00 → $560.00（較 06-30 版 -21.5%；IBKR 07-22）；YTD +177.8% → +118.2%；市值約 $560B → $444.6B（793,959,430 股，10-Q 04-26 ✓ SEC）；預估本益比 ~55× → 約 46× — 且 FY26 共識 EPS 分母修正為 $12.26（stockanalysis，39 位，07-15）✗→ 取代 ~$13.0 次級數字；FY27 共識 EPS $16.75（+37%）→ 以現價計約 33 倍。CEO Dickerson 於 06-29/30 以約 $700-736 賣出 78,321 股、約 $55.5M — 正在頂點，07-01 申報 ✓ SEC Form 4。

How it bears on the thesis. The 06-30 warning — priced for perfection, any wobble compresses — executed to the letter, except the wobble was the customers' capex sentiment, not AMAT's own numbers. The de-rating the finding said was needed has begun (~55×→~46×). The live question is now FY27: consensus embeds +37% EPS growth that assumes the 2027 memory-capex wave lands intact — the same wave whose durability is the open question across this desk's memory coverage. This update also adds the playbook section below (the page predates the decision-layer format). ◆ framing = judgment定位＝判斷

對論點的影響。06-30 的警告 — 完美定價、一有閃失就收縮 — 逐字兌現，只是閃失來自客戶端的資本支出情緒，不是 AMAT 自己的數字。結論所說「需要的去評價」已經開始（~55×→~46×）。現在的活問題是 FY27：共識隱含 EPS +37%，前提是 2027 記憶體資本支出波完整落地 — 而那波的持續性，正是本站記憶體研究裡懸而未決的同一個問題。本次更新同時補上文末操作框架（本頁早於 decision-layer 格式）。◆ 定位＝判斷

✓ 2026-06-30 Numbers refreshed — the re-rating kept going, now on rising estimates 數字已更新 — 評價重估還在續，且這次是靠獲利上修

Re-pulled live: $713.00, +177.8% YTD, a fresh 52-week high ✓ IBKR · 06-30 — up another ~21% from the $589 build price. Yet the forward P/E fell to ~55× (vs ~57× at build), because consensus FY26 EPS was revised up to ~$13.0 (from ~$10.4) and the FY26 revenue forecast lifted to ~$33.3B (from ~$31.6B) ✓ Yahoo / Simply Wall St · ~06-26. Driver: a high-conviction product showcase plus a wave of Wall-Street target upgrades.

即時重抓：$713.00、年初至今 +177.8%、創 52 週新高 ✓ IBKR · 06-30 — 比生成時的 $589 又漲約 21%。但預估本益比反而降到約 55 倍（生成時約 57 倍），因為市場把 FY26 每股盈餘共識上修到約 $13.0（原約 $10.4）、FY26 營收預估也升到 約 $33.3B（原約 $31.6B）✓ Yahoo／Simply Wall St · 約 06-26。推力：一場高聲量的產品發表，加上多家華爾街調升目標價。

How it changes the finding. The headline above said the re-rating was largely done and a further double would need earnings growth (engine 1), not multiple expansion. That is now visibly happening: the multiple compressed while the price rose, because the E got marked up — engine 1 is firing, partly validating the variant bull. The flip side hardens too: at +178% YTD and a fresh high (~76× trailing on FY25 EPS; TTM since risen), the "priced-for-perfection, any-wobble-compresses" bear is sharper, and "edge after a double" now means going from ~$560B to ~$1.1T+. Net: not more bullish, not more bearish — the core tension is resolving in real time, with estimates doing the work the thesis said they would have to. ◆ framing = judgment定位＝判斷

對結論的影響。上方結論說評價重估大半已發生，要再翻倍得靠獲利成長（引擎一）、而非評價擴張。現在這件事正在發生：股價漲、本益比反而收縮，因為E 被上修了 — 引擎一正在點火，部分驗證變異多方。但另一面也更硬：在年初至今 +178%、又創新高（以 FY25 EPS 算約 76 倍歷史本益比，惟 TTM 已上升）之下，「完美定價、一有閃失就收縮」的空方更尖銳；而「翻倍後還有沒有超額」如今是指從約 $560B 翻到約 $1.1 兆以上。淨結論：沒有更偏多、也沒有更偏空 — 核心張力正即時化解，且正是靠獲利上修在做事。◆ framing = judgment定位＝判斷

✓ 2026-06-16 AMAT × EssilorLuxottica — a new AR-optics growth vector AMAT × EssilorLuxottica — 一條新的 AR 光學成長軸

On 2026-06-16, EssilorLuxottica (the Ray-Ban / Oakley owner) and Applied Materials signed a long-term joint development agreement for next-generation intelligent optical systems in AR / AI smart glasses. It pairs EssilorLuxottica's lenses-frames-eyewear scale with AMAT's materials engineering and waveguide expertise; scope covers waveguides, adaptive (auto-dimming) lenses, materials innovation and advanced lens packaging, with a dedicated joint lab on AMAT's Silicon Valley campus. ✓ GlobeNewswire · 06-16

2026 年 6/16，EssilorLuxottica（Ray-Ban／Oakley 母公司）與應用材料簽署長期共同開發合約，鎖定 AR／AI 智慧眼鏡的次世代智慧光學系統。結合 EssilorLuxottica 的鏡片－鏡框－眼鏡規模，與 AMAT 的材料工程與導波路（waveguide）技術；範圍涵蓋導波路、適應型調光鏡片、材料創新、先進鏡片封裝，並在 AMAT 矽谷園區設專用聯合實驗室。✓ GlobeNewswire · 06-16

How it bears on the thesis. This is optionality, not a near-term number. It supports the variant view that AMAT's materials-engineering moat is broader than core semi WFE — extending into a large new AR-optics TAM. But it is an early-stage R&D agreement (a joint lab, no disclosed revenue or timeline), so it does not change the headline finding above: at ~57× forward P/E the re-rating is largely done, and edge-after-a-double still depends on earnings growth (engine 1), not this. Treat it as a multi-year call option on AMAT's TAM, not a 2026 catalyst. ◆ framing = judgment定位＝判斷

對論點的意義。這是選擇權式的想像空間（optionality），不是近期數字。它支持「AMAT 的材料工程護城河不只在核心半導體 WFE、能延伸到一塊很大的新 AR 光學 TAM」這個變異觀點。但它是早期 R&D 合約（只設聯合實驗室，未揭露營收或時程），所以不改變上方的結論：約 57 倍預估本益比下，評價重估大半已發生，翻倍後還有沒有超額報酬，仍取決於獲利成長（引擎一），而非此案。把它當成 AMAT 市場規模（TAM）的「多年期買權」，而不是 2026 年的催化劑。◆ framing = judgment定位＝判斷

Full case — business, bull, bear, variant perception, management完整論證 — 本業、多方、空方、變異觀點、經營層

Business summary本業概要

A single chip growing upward into an intricate vertical city of stacked micro-architecture on a circuit-board base
More AI = more complexity stacked per chip = more process steps AMAT sells into.AI 越多 = 每顆晶片堆疊的複雜度越高 = AMAT 能賣進去的製程步驟越多。

AMAT is the world's broadest supplier of materials-engineering equipment for making chips — the deposition, etch, CMP, implant and metrology tools from the decoder above. As chips move to GAA transistors, backside power and HBM stacking, each wafer needs more steps and higher-value tools. AMAT sells the steps.

AMAT 是全球產品線最廣的材料工程設備供應商 — 也就是上面解碼器裡的沉積、蝕刻、CMP、佈植、量測機台。當晶片走向 GAA 電晶體、晶背供電、HBM 堆疊，每片晶圓需要的步驟更多、機台單價更高。AMAT 賣的正是這些步驟。

How it makes money怎麼賺錢 FY2025 ✓ 10-K

  • Semiconductor Systems — $20.8B (73%). Capital tools to TSMC, Samsung, SK Hynix, Intel.
  • AGS — $6.4B (23%). High-margin recurring service on the installed base.
  • Display & adjacent — $1.06B (~4%). Small, cyclical.
  • 半導體系統 — $20.8B（73%）。賣機台給台積電、三星、SK 海力士、Intel。
  • AGS 服務 — $6.4B（23%）。對已裝機機台的高毛利重複性服務。
  • 顯示器與其他 — $1.06B（約 4%）。小、且循環。
Total revenue $28,368M · GAAP EPS $8.66 · non-GAAP EPS $9.42.總營收 $28,368M · GAAP 每股盈餘 $8.66 · 非 GAAP $9.42。

Why it matters為何重要

Rising chip complexity = more materials-engineering steps per wafer. AMAT is the diversified toll-taker on that complexity, not a bet on one node or customer.

晶片複雜度上升 = 每片晶圓的材料工程步驟更多。AMAT 是這個複雜度的「多元收費員」，不是押注單一節點或客戶。

Gross margin 48.7% GAAP / 48.8% non-GAAP, +1.2pt YoY毛利率 48.7%（GAAP）／ 48.8%（非 GAAP），年增 1.2 個百分點 ✓ 10-K. corrected the draft said "low-50s" — never happened.記憶版寫「50 出頭」— 並未發生。

3D scaling — why every stacked layer is a materials-engineering toll booth3D scaling（立體堆疊）— 為什麼每多疊一層，都要經過 AMAT 的收費站

Moving a bit of data across a package costs energy. As AI chips move more bits per second than they compute, the wire length between logic and memory becomes the power bill. Applied's Mukund Srinivasan, at the 08-31 Seoul briefing: "The solution to the industry's key challenge of minimizing power consumption per bit of data is 3D scaling." Hybrid bonding, he said, "appears essential across all 3D stacking." ✓ THE ELEC／ZDNet Korea／Korea Times · 08-31

把一個位元的資料搬過封裝，是要耗能的。AI 晶片每秒搬動的位元數，遠超過它運算的量，於是邏輯與記憶體之間的導線長度，就變成那筆電費帳單。應材 Mukund Srinivasan 在 08-31 首爾說明會上：「解決產業對每位元資料耗能最小化這個關鍵挑戰的方案，就是 3D scaling（立體堆疊）。」他也說，混合鍵合「在所有 3D 堆疊裡看起來都是必要的」。✓ THE ELEC／ZDNet Korea／Korea Times · 08-31

Two ways to stack兩種堆疊方式

Microbump stacking joins dies with tiny solder balls plus underfill resin, at a pitch of tens of microns ⚠ industry-typical value. Hybrid bonding polishes both die surfaces dead flat and presses them together so copper pads fuse directly to copper pads, and the surrounding oxide fuses to oxide — no solder, a pitch of a few microns down to roughly 10 µm, a much shorter wire, a better heat path, and a thinner stack ⚠ industry-typical value. Background context from this research, not from the briefing: HBM4 remains on microbumps; hybrid bonding is slated for the HBM4E generation ⚠ single source.

微凸塊（microbump）堆疊，是用細小的焊球加上底部填充膠把晶粒接起來，間距在數十微米等級 ⚠ 業界典型值。混合鍵合（hybrid bonding）則是把兩片晶粒的表面拋得極平，直接壓合，讓銅墊片對銅墊片直接熔接、周圍的氧化層對氧化層也直接熔接 — 沒有焊料，間距在數微米到約 10 微米之間，導線短很多、散熱路徑更好、堆疊也更薄 ⚠ 業界典型值。以下為本研究的背景補充，非說明會原文：HBM4 目前仍用微凸塊；混合鍵合預計要到 HBM4E 世代才會導入 ⚠ 單一來源。

Microbump stacking Hybrid bonding solder balls + underfill · pitch tens of µm Cu–Cu + oxide–oxide direct bond no solder · pitch a few µm · shorter wires, better heat path Schematic, not to scale; pitch figures are industry-typical values
Microbump stacking (left) vs hybrid bonding (right): the same die pair, joined two different ways.微凸塊堆疊（左）vs 混合鍵合（右）：同一對晶粒，兩種不同的接合方式。

Where AMAT's tools sit in that stackAMAT 的機台，卡在這個堆疊的哪裡

The product set Applied showed at the 08-31 briefing is its existing 06-25 "DRAM & Advanced Packaging Master Class" line-up: deposition (Producer Avila 2, Nokota VMax 2), epitaxy (Centura Prime Epi), the Kinex die-to-wafer hybrid bonder, CMP polishing (Opta Quad), and inspection/metrology (VeritySEM 7AP, SEMVision G7AP) ✓ THE ELEC／ZDNet Korea／Korea Times · 08-31. Bonding needs CMP because the two die surfaces must be flat to the nanometre — a single stray particle or a nanometre of bow on either surface ruins the bond, so the polish step (Opta Quad) and the inspection step (VeritySEM 7AP, SEMVision G7AP) are not optional add-ons, they are what makes hybrid bonding possible at all. This is the toll-booth logic: every added layer repeats deposit → polish → bond → inspect, and Applied sells a tool at each stop.

應材在 08-31 說明會上展示的產品線，其實是既有的 06-25「DRAM 與先進封裝大師班」機台組合：沉積（Producer Avila 2、Nokota VMax 2）、磊晶（Centura Prime Epi）、die-to-wafer 混合鍵合機（Kinex）、CMP 拋光（Opta Quad）、檢測與量測（VeritySEM 7AP、SEMVision G7AP）✓ THE ELEC／ZDNet Korea／Korea Times · 08-31。鍵合需要 CMP，是因為兩片晶粒的表面必須平整到奈米等級 — 任何一顆微粒，或任一面翹曲了一奈米，這個鍵合就毀了，所以拋光（Opta Quad）與檢測（VeritySEM 7AP、SEMVision G7AP）不是可有可無的附加步驟，而是讓混合鍵合這件事得以成立的前提。這就是收費站的邏輯：每多疊一層，就重複一次「沉積 → 拋光 → 鍵合 → 檢測」，而應材在每一站都賣一台機器。

Two new materials兩項新材料

The 08-31 briefing newly mentioned two materials on top of that product set. Thermally conductive insulating materials for HBM: heat generated inside a 12- or 16-high DRAM stack has to leave through the stack itself, so the material between the layers has to conduct heat while still insulating electrically. Optical waveguide materials for co-packaged optics (CPO): light, instead of copper, carries the signal for the last centimetres to the switch chip. No numbers were given for either, so none are quoted here.

08-31 說明會在既有機台組合之外，新提到兩項材料。HBM 用的導熱絕緣材料：12 層或 16 層堆疊的 DRAM 內部產生的熱，得靠堆疊本身往外散，所以層與層之間的材料，既要能導熱、又要維持電氣絕緣。共同封裝光學（CPO，co-packaged optics）用的光學導波路材料：訊號傳到交換晶片的最後幾公分，改用光而不是銅來傳輸。兩者都沒有給出具體數字，所以這裡也不引用任何數字。

Isometric illustration: two silicon dies being pressed together with a glowing seam of copper pads fusing directly, layered materials and inspection beams visible around the stack
3D scaling in one image: each stacked layer repeats deposit, polish, bond, inspect.3D scaling 濃縮成一張圖：每多疊一層，就重複一次沉積、拋光、鍵合、檢測。

What it means for the thesis對論點的意義

The 08-31 briefing changes no estimate on this page — it reprises the June product set. It matters because it is the mechanism behind management's ">70% advanced packaging growth in calendar 2026" claim (already on the page) and behind the ~80%-of-WFE-growth shape. It is why the compounder reading of AMAT — a materials-engineering toll booth sitting on every stacking roadmap — survives a -40% tape. It is also why the cyclical reading is not dead: a toll booth still collects less when traffic drops. ◆ framing = judgment定位＝判斷

08-31 的說明會沒有改變本頁任何一個預估數字 — 展示的仍是 6 月那組產品線。它重要的地方在於：它就是經營層「2026 曆年先進封裝成長 >70%」（本頁已有）與「約 80% WFE 成長」這個形狀背後的機制。它也是 AMAT 的「複利型」讀法 — 每一種堆疊路線圖上都坐著一座材料工程收費站 — 為什麼能撐過 -40% 的走勢的原因。但這同時也是「循環型」讀法沒有死透的原因：車流變少時，收費站收到的錢也會變少。◆ 定位＝判斷

Bull case — what could go right多方論點 — 哪裡可能對

AI-driven structural demandAI 帶動的結構性需求

GAA transistors, backside power and HBM/advanced packaging each add deposition, etch, CMP and metrology steps — AMAT's strongest areas. AI capex could stay high for years.

GAA 電晶體、晶背供電、HBM／先進封裝，每一項都增加沉積、蝕刻、CMP、量測步驟 — 都是 AMAT 最強的領域。AI 資本支出可能維持高檔好幾年。

Mix upgrade, not just a headwind不只是逆風，是品質升級

China fell from 37.2% of revenue in FY24 to 30.1% in FY25 ✓ 10-K. The bull read: mix shifts toward higher-ASP leading-edge tools for TSMC/Samsung/SK Hynix — better margins, steadier demand.

中國營收占比從 FY24 的 37.2% 降到 FY25 的 30.1% ✓ 10-K。多方解讀：產品組合轉向賣給台積電／三星／SK 海力士的高單價先進機台 — 毛利更好、需求更穩。

AGS annuity compoundingAGS 年金式複利

$6.4B of recurring, high-margin service grows with the installed base and smooths the WFE cycle — a layer the market underweights vs. lumpy systems sales.

$6.4B 的重複性高毛利服務，隨已裝機規模成長、平滑 WFE 循環 — 相對忽高忽低的機台銷售，這塊被市場低估。

Optionality not in the price股價未反映的選擇權

Leadership in EPI for GAA and CMP for new materials; possible expansion into power semis and process-control software; CHIPS-Act onshoring as a 2026-27 tailwind. R&D $3.57B/yr funds the edge ✓ 10-K.

在 GAA 用的 EPI、新材料用的 CMP 領先；可能擴張到功率半導體與製程控制軟體；CHIPS 法案在地化是 2026–27 的順風。研發支出每年 $3.57B 支撐技術領先 ✓ 10-K。

What the market may be missing:市場可能漏看的： the duration and breadth of AI-complexity intensity.AI 複雜度需求的持續時間與廣度。 but see the headline — much of this is now in the price.但見開頭結論 — 這大半已反映在股價裡。

Bear case — what could go wrong空方論點 — 哪裡可能錯

Valuation is the bear case now估值本身就是空方理由

At ~34× forward / ~38× trailing after a -40.7% drawdown (was ~42×/~46× on 08-14, ~46×/~52× on 07-23, ~55×/~76× at the 06-30 high), any growth wobble = multiple compression — the July selloff was this card executing. It executed again on 08-13: a beat-and-raise still sold off ~5%, on a gross margin guided flat. The stock still prices in durable, low-cyclicality growth. live · 09-02

在 -40.7% 回檔後仍有預估 34 倍／歷史 38 倍（08-14 為 ~42×／~46×，07-23 為 ~46×／~52×，06-30 高點為 ~55×／~76×）；成長一有閃失就會本益比收縮 — 7 月的殺盤就是這張卡片的兌現。08-13 又兌現了一次：超標又上修指引，仍然因為毛利率指引持平而跌約 5%。股價仍把「持久、低循環」的成長反映在價內。即時 · 09-02

WFE is cyclical and lumpyWFE 是循環且忽高忽低

If hyperscalers digest AI capacity, equipment orders can drop sharply — history says so. A "super-cycle" is itself an assumption.

若雲端大廠消化既有 AI 產能，設備訂單可能急跌 — 歷史有前例。「超級循環」本身就是一個假設。

China / geopolitics中國 / 地緣政治

30.1% of FY25 revenue, down to 27% in FQ2 FY26 and 26% in FQ3 (share of Semiconductor Systems + AGS revenue) ✓ call · 08-13法說 · 08-13. AMAT guided a $600M FY2026 hit from expanded U.S. export curbs ✓ Reuters/TF; domestic substitution (NAURA) and Taiwan tail-risk add more.

FY25 占營收 30.1%，FQ2 FY26 降至 27%、FQ3 再降至 26%（占半導體系統＋AGS 營收）✓ 法說 · 08-13。AMAT 預估美國擴大出口管制將造成 FY2026 約 $600M 衝擊 ✓ Reuters/TF；中國本土替代（NAURA 北方華創）與台海尾端風險還會再加。

Share pressure at the edges邊緣領域的市佔壓力

Lam leads etch, KLA leads metrology, ASML owns EUV litho. AMAT must keep winning deposition/CMP while defending elsewhere.

Lam 主導蝕刻、KLA 主導量測、ASML 獨佔 EUV 微影。AMAT 得在沉積/CMP 持續贏，同時守住其他領域。

Must hold for the thesis to fail:論點要被推翻，需同時成立： sustained high AI capex from a concentrated customer set, on-time node transitions, no new export shocks, continued pricing/mix gains. Break one and the multiple unwinds.集中客戶群的 AI 資本支出持續高檔、節點轉換準時、沒有新的出口管制衝擊、定價/組合持續改善。任一條斷掉，本益比就會回吐。 Two additions from the 08-13 call: FQ4 gross margin is guided flat sequentially on ramp costs and display mix — the 13-quarter expansion streak pauses exactly when revenue is at a record; and the non-GAAP tax rate steps from ~11% to ~13% in 2027 as the global minimum tax lands, a quiet ~2-point drag on the FY27 EPS the whole bull case leans on.08-13 法說新增兩點：FQ4 毛利率指引與前季持平，原因是產能拉升成本與顯示器組合 — 連續 13 季的擴張正好在營收創新高時停下；另外非 GAAP 稅率將自約 11% 墊高到 2027 年約 13%（全球最低稅負制），對整個多方論點所倚賴的 FY27 每股盈餘，是一個約 2 個百分點的隱形拖累。 ✓ guidance + call · 08-13指引＋法說 · 08-13

Variant perception變異觀點 — the reason to read— 值得讀的地方

WALL STREET BELIEVES High-quality cyclical riding the AI upcycle. China = a headwind being de-risked. Implied: re-rating still ahead. THE VARIANT VIEW A secular, multi-year shift to equipment- intensive chipmaking. China decline = a quality upgrade. Earnings durable, cycle shallower. WHAT THE PRICE DID +71% YTD -40.7% off the 07-01 high Fwd P/E ~34× · ~38× ttm. The variant view is now largely consensus — and largely paid for. New edge needed.
Consensus → variant → what the tape already reflects. The variant thesis was right enough that the market adopted it; the +71% YTD (after a -40.7% give-back from the 07-01 high) is the receipt — drawdown included. The 08-13 print started the sharpest version yet of the third panel: record revenue, a raised guide, and the stock down 5% that day, then down 18% more by 09-02.共識 → 變異觀點 → 股價已反映的。變異論點夠對，所以市場已採納；+71% YTD（自 07-01 高點回吐 -40.7% 之後）就是收據 — 連同回檔一起。08-13 的財報開啟了第三格最尖銳的版本：營收創紀錄、指引上修，股價當天跌 5%，到 09-02 又再多跌 18%。

Management & insiders經營層與內部人

Track record經營紀錄

CEO Gary Dickerson ✓ public — long tenure, navigated multiple cycles, positioned AMAT for advanced nodes and AI. High-quality capital allocation: heavy R&D ($3.57B), buybacks, growing dividend.

執行長 Gary Dickerson ✓ 公開 — 任期長、歷經多次循環、把 AMAT 卡位在先進節點與 AI。資本配置品質高：重研發（$3.57B）、買回庫藏股、配息成長。

Insider signal — net selling內部人訊號 — 淨賣出

Insiders own just 0.3% (~$1.1B) ✓ SWS. CEO Gary Dickerson sold 78,321 shares for ~$55.5M at ~$700–736 on 2026-06-29/30 — the all-time-high days — filed 07-01; 1,599,843 shares held after ✓ SEC Form 4. President Prabu Raja sold 50,000 shares on 2026-06-04 for $25,264,197 ($503.99–507.23, via living trust) ✓ SEC Form 4. Sales at highs are common; the cluster of them AT the top, with zero buying, is a data point the bear column gets to keep.

內部人僅持股 0.3%（約 $1.1B）✓ SWS。執行長 Gary Dickerson 於 2026-06-29/30 — 正是歷史高點那兩天 — 以約 $700–736 賣出 78,321 股、約 $55.5M，07-01 申報，餘 1,599,843 股 ✓ SEC Form 4。總裁 Prabu Raja 於 2026-06-04 賣出 5 萬股、$25,264,197（$503.99–507.23，透過生前信託）✓ SEC Form 4。高檔賣股常見；但成群出現在頂點、且無人買進，是空方欄位可以留著的一個數據點。

Jargon decoder — what the process steps actually are術語解碼器 — 那些製程步驟到底是什麼

Jargon decoder — read this first術語解碼器 — 先看這個

AMAT sells the steps that turn a blank silicon wafer into a chip. To judge the thesis you have to know what those steps are. Here is the wafer's journey, then a plain-language glossary.

AMAT 賣的是把一片空白矽晶圓變成晶片的那些製程步驟。要判斷這個論點，得先知道這些步驟是什麼。先看晶圓的旅程，再看白話術語表。

ONE WAFER · HUNDREDS OF PROCESS STEPS · AMAT LEADS MOST 1 · Deposit CVD / ALD / EPI add a new layer 2 · Pattern Litho ASML's step, not AMAT 3 · Etch Etch carve a pattern 4 · Implant Ion implant fire in ions 5 · Polish CMP flatten to mirror 6 · Measure Metrology check, then repeat repeat the loop hundreds of times — more layers = more AMAT tools sold
The whole bull case in one picture: AI chips need more layers, so the loop runs more times, so AMAT sells more tools. AMAT leads every step here except lithography (step 2 — that's ASML's EUV).一張圖看懂多方論點：AI 晶片需要更多層 → 循環跑更多次 → AMAT 賣出更多機台。圖中每一步幾乎都是 AMAT 主場，只有微影（第 2 步）是 ASML 的 EUV。

And here is what those steps physically do to the silicon — building one metal wire, in cross-section (the real copper "damascene" process):

再用剖面圖看這些步驟對矽晶到底做了什麼 — 以做出一條金屬線為例（這就是真實的銅「鑲嵌／大馬士革」製程）：

silicon 1 · Deposit lay a dielectric film 2 · Pattern + etch open a trench 3 · Deposit metal fill + overflow 4 · CMP polish flat — wire embedded
This loop builds just one wiring layer. An AI chip stacks 10–15+ of them — so the loop runs 10–15+ times, and AMAT sells a tool for nearly every pass. More complexity = more passes = more AMAT.這套循環只做出一層線路。一顆 AI 晶片要疊 10–15 層以上 — 循環就跑 10–15 次以上，幾乎每一輪 AMAT 都賣得到機台。越複雜 = 跑越多次 = AMAT 賣越多。

Step 1 (deposit) isn't one thing — it's a ladder of tools for ever-harder 3D structures: PVD → CVD → ALD. The more vertical the chip, the more you need the precise (slower, pricier) end. This is AMAT's backyard.

第 1 步（沉積）不是單一技術，而是一排對應越來越難 3D 結構的工具：PVD → CVD → ALD。晶片越立體，越需要精準（但更慢、更貴）的那一端。這正是 AMAT 的主場。

Illustration of PVD: a solid target sputtered into atoms that fly straight down onto the wafer
PVDphysical · sputter物理氣相沉積
Blast a solid target into atoms that fly straight down. Fast, high throughput — great for metal layers, barriers, liners. Struggles to reach the bottom of deep 3D structures.把固態靶材打成原子，直線飛濺到晶圓上。快、吞吐量高 — 金屬層、阻障層（barrier）、襯墊層（liner）的主力。但難顧到深 3D 結構的洞底與側壁。
Illustration of CVD: a cloud of gas particles settling onto a wafer and forming an even film
CVDchemical · grows化學氣相沉積
Gaseous precursors react on the surface — the film "grows." Better coverage and uniformity than PVD; the production workhorse. Hits its precision limit as structures get more 3D.氣態前驅物在表面反應，薄膜是「長出來」的。覆蓋性與均勻性優於 PVD，量產主力。結構越 3D，會碰到精度極限。
Illustration of ALD: a uniform thin skin coating every contour of a 3D trenched structure
ALDatom by atom原子層沉積
Self-limiting — each cycle grows exactly one atomic layer, almost regardless of shape. Indispensable for FinFET, GAA, high-aspect-ratio. Slow and pricey, so used only where nothing else works. Key for GAA.自限制反應 — 每循環只長一層原子厚度，幾乎不受結構形狀影響。FinFET、GAA、高深寬比必備。慢又貴，只用在非用不可處。GAA 關鍵。

AMAT is the deposition leader — #1 in PVD and a top player in CVD and ALD. (EPI — growing a single crystal for the transistor channel — is a fourth AMAT strength.) Illustrations are conceptual.AMAT 是沉積龍頭 — PVD 市占第一，CVD 與 ALD 也是要角。（另有磊晶 EPI，長單晶當電晶體通道，也是 AMAT 強項。）圖為示意。

Whatever the method, deposition is judged on three things — and 3D chips make all three harder:

不論用哪種方法，沉積的好壞看三件事 — 而 3D 晶片讓這三件事都更難：

Uniformity Conformality Surface smoothness Poor Good
Every method chases the same three: even thickness everywhere (uniformity), full coverage down into 3D trenches (conformality), a smooth top (surface smoothness). The deeper and narrower the structure, the harder all three get — which is why the precise tools (ALD) keep gaining share. Recreated from a vocus.cc explainer.每種方法追求的都是同三件事：到處厚度一致（均勻性）、連 3D 深溝底部都完整包覆（覆蓋性）、表面平整（表面平整度）。結構越深越窄，這三件事越難 — 這正是精準工具（ALD）市占持續成長的原因。圖表重製自 vocus.cc 解說。
The process steps + 2 key terms製程步驟 ＋ 2 個關鍵詞
Deposition薄膜沉積（Deposition）
Lay an ultra-thin film onto the wafer. Four tools, by rising precision: PVD (sputter a solid, fast), CVD (gas grows a film, the workhorse), ALD (one atom-layer at a time, for 3D), EPI (grow a single crystal). See the diagrams above.
在晶圓表面鋪一層極薄的膜。依精準度由低到高有四種：PVD（濺鍍固態靶材，快）、CVD（氣體長膜，主力）、ALD（一次一原子層，攻 3D）、EPI（長單晶）。見上方圖解。
AMAT — its strongest area: #1 in PVD, a leader in ALD/EPI for 2nm GAA.— AMAT 最強的領域：PVD 市占第一，ALD／EPI 在 2 奈米 GAA 領先。
Litho (Pattern)微影（Litho，Lithography／黃光）
Shine light through a stencil (a "mask") to print the circuit pattern onto a light-sensitive layer — this decides where the next etch happens.
讓光透過光罩（mask）把電路圖案「印」到感光的光阻層上 — 決定接下來要在哪裡蝕刻。
Led by ASML's EUV — the one major step where AMAT is not the player.由 ASML 的 EUV 主導 — 少數 AMAT 不是主角的關鍵步驟。
Etch蝕刻（Etch）
Like carving: chemically or with plasma, bite away the material you don't want, leaving the pattern behind.
像雕刻：用化學或電漿把不要的材料咬掉，留下想要的圖案。
Led by Lam Research — a key AMAT competitor here.由 Lam Research（科林研發）主導 — 此領域 AMAT 的主要對手。
Ion implantation離子佈植（Ion Implantation）
Fire high-speed ions into the silicon to change how it conducts electricity (called "doping").
把高速離子「打」進矽裡，改變它導電的方式（稱為「摻雜 / doping」）。
AMAT — a leadership area.— AMAT 領先的領域。
CMP化學機械研磨（CMP，Chemical-Mechanical Polishing）
Spin the wafer against a pad while spraying chemical slurry — grind the surface mirror-flat so the next layer sits cleanly.
晶圓貼著墊布旋轉、同時噴含研磨粒子的化學液，把表面磨到鏡面平整，下一層才鋪得乾淨。
AMAT — a leadership area.— AMAT 領先的領域。
Metrology量測檢測（Metrology）
After each step, measure dimensions and hunt for defects to protect yield (the % of good chips).
每一步之後量尺寸、抓缺陷，保護良率（合格晶片的比例）。
Led by KLA — AMAT competes here.由 KLA（科磊）主導 — 此處 AMAT 為競爭者。
Node製程節點（Node）
The generation label as transistors shrink (e.g. "2nm"). Smaller node = more, harder process steps.
電晶體越做越小的世代代號（例如「2 奈米」）。節點越小 = 步驟越多、越難。
More steps per node = more tools sold.每個節點步驟越多 = 賣出的機台越多。
Yield良率（Yield）
The % of chips on a wafer that come out good. As complexity rises, keeping yield high is the whole game.
一片晶圓上做出來「良品」的比例。複雜度越高，把良率守住就是整場競賽的核心。
Every extra step risks losing yield — which is why metrology matters.每多一道步驟就多一次掉良率的風險 — 這正是量測檢測重要的原因。
The advanced tech driving demand帶動需求的先進技術
GAA環繞閘極（GAA，Gate-All-Around）
A new transistor shape where the "gate" wraps around the channel on all sides — replaces FinFET at 2nm and below.
新一代電晶體結構，閘極四面包住通道 — 在 2 奈米以下取代 FinFET（鰭式電晶體）。
Needs far more deposition/etch steps → AMAT benefits.需要更多沉積/蝕刻步驟 → AMAT 受惠。
HBM高頻寬記憶體（HBM，High Bandwidth Memory）
Memory chips stacked tall with a very wide data pipe, used to feed AI accelerators fast enough.
把記憶體晶片疊高、資料通道拉很寬，才餵得動 AI 加速晶片。
Core of AI demand; drives advanced-packaging tools.AI 需求的核心；帶動先進封裝設備。
Backside power晶背供電（Backside Power Delivery）
Move the power wiring to the back of the chip, freeing the front for more signal routing.
把供電線路移到晶片背面，正面才有更多空間走訊號線。
A new technique = more new process steps.新技術 = 更多新的製程步驟。
Advanced packaging先進封裝（Advanced Packaging）
Stack and join several chips into one high-performance module (how HBM + an AI GPU come together).
把多顆晶片堆疊/併接成一個高效能模組（HBM 與 AI GPU 就是這樣結合的）。
A fast-growing tool market AMAT is pushing into.AMAT 正在切入、成長很快的設備市場。
Industry & money words產業與財務名詞
WFE晶圓廠設備（WFE，Wafer Fab Equipment）
The umbrella term for all the machines that make chips. The market AMAT lives in.
製造晶片所有機台的總稱。AMAT 所在的市場。
Notoriously cyclical and lumpy.出了名的景氣循環、訂單忽高忽低。
Foundry / IDM晶圓代工 / 整合元件製造商（Foundry / IDM）
Foundry = makes chips for others (TSMC). IDM = designs and makes its own (Intel, Samsung memory).
晶圓代工 = 幫別人製造晶片（台積電）。IDM = 自己設計也自己製造（Intel、三星記憶體）。
Both are AMAT's customers.兩者都是 AMAT 的客戶。
AGSApplied Global Services（應用全球服務）
AMAT's after-sales arm: spares, service, upgrades and software on the installed base.
AMAT 的售後事業：對已安裝機台提供備品、維修、升級與軟體。
High-margin, recurring — an "annuity" that smooths the cycle.高毛利、可重複 — 像「年金」般平滑景氣循環。
P/E · ASP · moat本益比 · 平均售價 · 護城河
P/E = price ÷ earnings (how expensive). ASP = average selling price per tool. Moat = a hard-to-cross competitive edge.
本益比（P/E）= 股價 ÷ 每股盈餘（貴不貴）。ASP（平均售價）= 每台機台平均賣多少。護城河（moat）= 對手難跨越的競爭優勢。
P/E now ~34× = the bear case.現在本益比約 34 倍 = 空方的理由。

Data provenance — every number, sourced資料來源 — 每個數字都有出處

Show the provenance table (46 rows)展開來源表（46 列）
Claim主張 Verdict判定 Value · source · date數值 · 來源 · 日期
FY2025 revenueFY2025 營收✓$28,368M, +4% (FY25 Q4 release 2025-11-13)
Semi Systems✓$20.8B / 73% (10-K)
AGS✓$6.4B / 23% (10-K)
Display✓$1.06B (10-K)
Gross margin毛利率✗→48.7% GAAP / 48.8% non-GAAP (10-K). draft "low-50s" corrected記憶版「50 出頭」已修正
EPS FY25✓GAAP $8.66 · non-GAAP $9.42 (release)
China mix中國占比✓FY24 37.2% ($10.12B) → FY25 30.1% ($8.53B) (10-K)
2026 China hit2026 中國衝擊✓$600M FY2026 (Reuters / TrendForce, 2025-11)
Price股價✓$438.46 close 2026-09-022026-09-02 收盤, +71.0% YTD, -40.7% off intraday high $739.67距盤中高點 $739.67, -18.0% since the 08-13 close較 08-13 收盤 (IBKR); 09-02 intraday low $433.57 undercut the 07-29 intraday low $436.33 — a fresh 13-week low09-02 盤中低點 $433.57 跌破 07-29 盤中低點 $436.33 — 創 13 週新低
Trailing P/E歷史本益比✗→~38× (37.9), same derivation as 08-14（37.9），推導方式與 08-14 版相同 (TTM GAAP EPS ~$11.58 — FY25 Q3's ~$2.22 swapped for FQ3's $3.17TTM GAAP 每股盈餘約 $11.58 — 把 FY25 Q3 的約 $2.22 換成 FQ3 的 $3.17)
Forward P/E預估本益比✗→~34× (34.4) ($438.46 ÷ FY26E non-GAAP EPS $12.75 — unchanged company arithmetic since 08-14$438.46 ÷ FY26E 非 GAAP 每股盈餘 $12.75 — 自 08-14 以來的公司算式未變); ~24× (23.8) on FY27E $18.44 (MarketScreener/S&P Global, 09-03); ~26× (25.6) on the old $17.12; basis caveat applies — see Consensus row below以 FY27E $18.44 計（MarketScreener／S&P Global，09-03）；以舊的 $17.12 計約 26 倍（25.6）；口徑但書見下方共識列
Market cap市值✓$348.0B ($438.46 × 793,597,443 sh, 10-Q cover as of 2026-07-26, filed 2026-08-20$438.46 × 793,597,443 股，10-Q 封面日 2026-07-26，2026-08-20 申報)
FQ2 FY26 (2026-05-14)✓rev $7.91B +11% YoY · GAAP EPS $3.51 · non-GAAP $2.86 · GM ~50% · China 27% (GlobeNewswire/10-Q)
FQ3 FY26 (2026-08-13)✓rev $9,115M +25% YoY / +15% QoQ · GAAP EPS $3.17 · non-GAAP EPS $3.50 · GM 50.3% GAAP / 50.4% non-GAAP · non-GAAP op margin 34.0% · OCF $3,037M · FCF $2,330M · capex $707M · returns回饋股東 $860M ($440M buyback庫藏股 + $420M dividends股利) (release / 8-K財報／8-K)
FQ3 segments & mixFQ3 部門與組合✓Semi Systems $7,040M (+27% YoY) · AGS $1,781M (+22%) · Other/Display $294M (op loss營業損失 $118M); Semi Systems mix foundry-logic 67% / DRAM 26% / flash 7%; China 26% of Semi Systems + AGS半導體系統組合 代工邏輯 67%／DRAM 26%／快閃 7%；中國占半導體系統＋AGS 之 26% (release + call財報＋法說)
9M FY26 cumulativeFY26 前三季累計✓rev $24,037M · GAAP EPS $9.22 · non-GAAP EPS $8.73 (8-K)
FQ4 FY26 guideFQ4 指引✓rev $10.25B ±$0.5B (+12.5% QoQ) · non-GAAP EPS $4.02 ±$0.20 · non-GAAP GM ~50.4% (flat QoQ, ramp costs + display mix季持平，產能拉升成本＋顯示器組合) · non-GAAP opex ~$1.58B · tax ~11%, modelled ~13% in 2027 (global minimum tax)稅率約 11%，2027 年設定約 13%（全球最低稅負制） (guidance + prepared remarks指引＋書面談話)
Calendar-2026 growth framing2026 曆年成長框架✓Semi Systems >30% and raised again ("further raising" — Dickerson PR; "greater than that at this point" — Hill) · advanced packaging >70% · PDC >50% · AGS >20%; leading-edge foundry-logic + DRAM + advanced packaging ≈80% of WFE growth in 2026 and 2027; capacity doubling by 2028半導體系統 >30% 且再度上修（Dickerson 新聞稿「再次上修」；Hill「目前看是高於那個數」）· 先進封裝 >70% · 製程檢測控制 >50% · AGS >20%；先進邏輯＋DRAM＋先進封裝約占 2026 與 2027 年 WFE 成長 80%；產能 2028 年前倍增 (PR + prepared remarks + call新聞稿＋書面談話＋法說)
2027 outlook2027 展望⚠"Another strong year" driven by continued AI-led demand (CFO Brice Hill) — directional only; no revenue or EPS number was given「又一個強勁的年份」，由 AI 帶動的需求延續（財務長 Brice Hill）— 僅為方向性說法，未給出營收或每股盈餘數字
Consensus FY26 (post-print)共識 FY26（財報後）✓rev $34.25B / EPS $12.79 (stockanalysis, 39 analysts, 2026-09-01); avg PT平均目標價 $640.89, "Strong Buy強力買進" (range $358–900; opinions; was $634.94 at 08-13)（區間 $358–900；屬意見；08-13 時為 $634.94）
Consensus FY27/FY28共識 FY27/FY28⚠FY27 rev $46,132M / EPS $18.44; FY28 rev $55,445M / EPS $23.92 (MarketScreener, S&P Global data, 39 analysts, fetched擷取日 2026-09-03). Basis caveat: this series' own FY26 EPS is $13.19 vs stockanalysis's $12.79 (+3.1%) — not the same yardstick as the 08-14 figures; on the stockanalysis yardstick FY27 ≈ $17.9. stockanalysis now paywalls FY27.口徑但書：此系列自己算出的 FY26 EPS 為 $13.19，較 stockanalysis 的 $12.79 高 3.1% — 與 08-14 所用的並非同一把尺；以 stockanalysis 的口徑計，FY27 約 $17.9。stockanalysis 目前已將 FY27 鎖付費牆。
FY2026E (this page)FY2026E（本頁）✓rev $34.29B · non-GAAP EPS $12.75 — 9M actual + FQ4 guide midpoint, not a forecast; unchanged since 08-14前三季實績＋FQ4 指引中值，非預測；自 08-14 未變
Entry-zone derivation進場區推導◆40× FY26E $12.75 = $510; 29× FY27E $18.44 = $535 (or ~$17.9 like-for-like = $519); the FY26 leg now binds → < ~$510. Framework unchanged since 07-23; only the EPS inputs moved40× FY26E $12.75 ＝ $510；29× FY27E $18.44 ＝ $535（或同口徑約 $17.9 ＝ $519）；現在由 FY26 那一腳決定 → < 約 $510。框架自 07-23 未變，變的只有 EPS 輸入值
Gary Dickerson Form 4✓78,321 sh, 2026-06-29/30, ~$55.5M @ ~$700–736, filed 07-01 (SEC Form 4 via secform4/StockTitan)
Insider ownership內部人持股✓0.3% (~$1.1B, Simply Wall St); Vanguard 7.49% 13G 2026-03-31
Prabu Raja✓50,000 sh, 2026-06-04, $25,264,197 @ $503.99–507.23 (SEC Form 4)
R&D (RD&E)✓$3,570M FY25 (vs $3,233M FY24) (10-K)
CEO✓Gary Dickerson (public record公開資料)
AR-optics JDAAR 光學合作✓EssilorLuxottica × AMAT long-term JDA, 2026-06-16 (GlobeNewswire / Euronext press release)
Probability split機率分布◆25/50/25 — model judgment, not data模型判斷，非資料
10-Q (2026-08-20)✓9M FY26 buybacks $1,137M (3M shares); authorization remaining $12.8B as of 2026-07-26FY26 前三季庫藏股 $1,137M（3M 股）；截至 2026-07-26 剩餘授權額度 $12.8B
Brice Hill Form 4✓7,500 sh, ~$3.59M @ avg $479.25, filed 2026-08-25
Board addition新任董事✓Akash Palkhiwala (Qualcomm CFO/COO) appointed to AMAT board + audit committee, effective 2026-08-27 (GlobeNewswire)
Seoul briefing首爾說明會✓2026-08-31, 3D scaling / hybrid bonding / HBM & CPO materials (THE ELEC／ZDNet Korea／Korea Times)
Six Five Summit✓2026-08-27 keynote, ~80% of WFE growth AI-tied (逐字稿 via Investing.com／Yahoo Finance)
SEMI 2028 WFE forecastSEMI 2028 WFE 預測✓$220B, raised from ~$200B in July (DigiTimes／Taipei Times, 2026-09-01)
TSMC (Cliff Hou)✓2026-09-02, equipment purchases ~1.9× Dec-2025 plan (Bloomberg／Taipei Times)
Equipment lead times設備交期✓up to 24 months packaging/test, 12 months standard etch/deposition (TrendForce, 2026-08-19)
SK hynix capex✓₩54T (~$38B) new Yongin DRAM + Cheongju NAND fabs (CNBC／24/7 Wall St, 2026-08-07)
Micron FQ4 FY26 date✓2026-09-30 (GlobeNewswire, 2026-08-26)
Peer 1-month moves同業一個月漲跌⚠AMAT -14% vs LRCX -2%, KLAC -6%, SOXX -0.7% — single sourceAMAT -14% vs LRCX -2%、KLAC -6%、SOXX -0.7% — 單一來源 (24/7 Wall St, 2026-09-02)
Historical P/E band歷史本益比帶✓2019 20.0× · 2020 21.0× · 2021 23.5× · 2022 12.7× · 2023 19.6× · 2024 18.6× · 2025 29.5× · current 41.0× on their TTM basis以其口徑計現值 41.0× (companiesmarketcap.com, fetched 2026-09-03)
Sell-side target moves賣方目標價調整✓JPMorgan $515→$660 OW · BofA $720→$650 Buy · Bernstein $675→$700 OP · MS $646→$642 EW · Craig-Hallum $530→$585 Buy · Berenberg $575→$675 Buy · Mizuho $650→$590 OP · UBS $675→$695 Buy · Goldman $670 Buy (MarketBeat／TipRanks, 2026-08-14~09-01)
FQ4 FY26 date⚠2026-11-12, after close (TipRanks, marked confirmed but not cross-checked on IR)
Investor conferences投資人法說✓2026-09-08 Citi Global TMT (CFO Hill) / 2026-09-09 Goldman Sachs Communacopia + Technology (CEO Dickerson) (GlobeNewswire, 2026-08-14)
07-29 -8.4% session07-29 -8.4% 那一天✓close $436.45, intraday low $436.33 (IBKR). Cause on the record:收 $436.45，盤中低點 $436.33（IBKR）。記錄在案的成因： ◆ press report Shanghai Aishengna reached DUV lithography mass production (Reuters／TrendForce, 2026-07-27/28) — causal link is press attribution, not company disclosure媒體報導 Shanghai Aishengna（上海一家 DUV 微影設備商）已量產 DUV 微影（Reuters／TrendForce，2026-07-27/28）— 因果連結為媒體歸因，非公司揭露
← Learn — jazzlien.com/learn