Skip to main content
Free trial

Ask AI to review this stock

Copy the prompt to get recent events, historical reactions, and a short research brief.

Tell your AI:

Help me set up FinLab and analyze 3533 嘉澤. Please read: https://finlab.finance/en/setup?stock=3533

3533

嘉澤 Electronic Components As of 2026-06-17 Updated daily after market close

strong quality, fair valuation, no active signal today.

This page refreshes daily after the close. Add 嘉澤 to your watchlist to track signal changes on your next visit.
Valuation 30Quality 90Growth 49Momentum 72Chips 93

Is 嘉澤 a buy? Fundamentals and valuation first

  • Top-tier profitability.Operating margin beats ~92% of the market. Latest quarterly ROE ≈ 5.8%.

  • Valuation is neutral.PE is in the 64th market percentile; within its own history, PE is around the 97th percentile.

  • Foreign investors are net buying.Foreign 20-day net flow ≈ 12.7% of volume (net buy).

Chip flow

Bearish -0.33
BearishNeutralBullish

Distribution, concentration falling — bearish

  • Distribution, concentration falling — bearish
  • 台灣摩根士丹利 holds 4,492 lots
  • sellers have 1,655 lots left, ~30 sessions to clear

Window from 2026-01-17 · Price +61% · branches to 2026-06-17· TDCC to 2026-06-12

  • Largest holder 台灣摩根士丹利 has accumulated 4,492 lots since the start (sold 0%), still adding.
  • Main distributor 元大證券 peaked at 1,928 lots, has sold 46%, 1,034 left, ~22 sessions to clear at the recent pace.
  • TDCC big holders (absolute) >1000-lot holders -4.43pp, mid holders +7, retail -0.25pp — little change.
  • Foreign hedge-desk net short Net short -931 lots (inferred from branch profile, mostly hedging — not necessarily bearish, not case-confirmed).
Branch inventory reconstruction
Accumulating holders Branches that accumulated shares during the window and still hold positive inventory; they are the absorbing side.
Distributing sellers Branches that built inventory earlier and have been selling it down; the remaining number estimates what has not cleared.
Net-short desks Trading-desk branches with negative inventory or net selling, often hedge, offsetting, or short-term flow. It is not automatically a bearish view.
Accumulating
台灣摩根士丹利美商高盛永豐金-匯立法銀巴黎永豐金證券凱基-台北
Distributing
元大證券國泰證券新加坡商瑞銀中國信託統一-城中群益金鼎-古亭
Desks
own scale
富邦證券香港上海匯豐群益金鼎台新
Price Accumulating & Distributing share one scale (comparable); Desks is independent. Zero is the baseline
Branch details Unit: lots

Accumulating

  • 台灣摩根士丹利 +4,492
    sold 0%· still adding
  • 美商高盛 +3,182
    sold 0%· still adding
  • 永豐金-匯立 +2,153
    sold 0%· still adding
  • 法銀巴黎 +809
    sold 2%· still adding
  • 永豐金證券 +606
    sold 9%· still adding
  • 凱基-台北 +502
    sold 12%· still adding

Distributing

  • 元大證券 1,034 left
    sold 46%· ~22 sessions to clear
  • 國泰證券 533 left
    sold 58%· ~8 sessions to clear
  • 新加坡商瑞銀 222 left
    sold 73%· paused recently
  • 中國信託 70 left
    sold 83%· ~17 sessions to clear
  • 統一-城中 10 left
    sold 76%· paused recently
  • 群益金鼎-古亭 23 left
    sold 38%· paused recently

Desks

  • 富邦證券 -1,720
    unclassified
  • 香港上海匯豐 -986
    hedge desk · will cover
  • 群益金鼎 -926
    unclassified
  • 台新 -728
    unclassified
Still selling 1,655 lots ~30 sessions to bottom at most
Recent buy/sell force +3,261 absorbed / -1,209 sold buyers slightly ahead
Hedge-desk net short -888 lots potential cover-buying, not selling pressure
TDCC big holders · absolute-holding cross-check
>1000-lot holder share Retail (<400 lots) share Price (right axis)

Mixed direction: big-holder and retail changes are not aligned.

Describes chip flow, not investment advice. Branch data discloses only high-volume branches with ~1-day lag; TDCC is weekly. Inventory is a relative reconstruction from the window start, not absolute holdings.

Historical evidence

Valuation regime and subsequent returns

Each day's PE is converted to its own-history percentile (as of that day), then we look at adjusted-close returns over the next 240 trading days by valuation regime.

PE percentile bandSample days240-day avg returnMedianWin rateAvg excess (vs index)
0-20 84+71.4% +77.4% 100% +55.2%
20-40 148+32.3% +13.2% 65.5% +16.9%
40-60 328+67.4% +75.4% 81.4% +37.2%
60-80 958+39.1% +41.7% 67.3% +19.7%
80-100 You are here1525+30.2% +26.5% 81.4% +18.4%

Percentiles use an expanding own-history window (no look-ahead); forward return is 240 trading days on adjusted close. Sample days overlap; distribution is indicative only.

Current PE ≈ 32.32, at the 97.2th percentile of its own history (80-100 band).

Valuation × momentum historical grid

Every day in its own history is bucketed by valuation tercile (own-history PE) and 60-day momentum sign into six cells; we show adjusted-close returns over the next 120 trading days per cell.

Valuation
Momentum
Cheap (own-history bottom 1/3)
Mid
Expensive (own-history top 1/3)
60-day momentum up
+3.2%
Win rate 35.1% Excess -2.1%
57d
+10.9%
Win rate 55% Excess -0.8%
238d
Now
+16.1%
Win rate 76% Excess +8.3%
1706d
60-day momentum down
+3.8%
Win rate 21.5% Excess -2.1%
144d
+24.5%
Win rate 79.6% Excess +12.1%
431d
+25.8%
Win rate 80.2% Excess +18.9%
587d

Currently in “Expensive (own-history top 1/3) × 60-day momentum up”: across 1706 historical days, 120-day forward avg gained 16.1%, win rate 76%, beat the market by 8.3%.

Days = trading days in this bucket; returns use adjusted prices, ex-costs. 60-day momentum: direction of this stock's price over the past 60 trading days.

Signals not active (historical stats for reference)

These are not occurring now; historical samples shown for comparison.

252-day price high (breakout)

Pending

Adjusted close makes a new 252-day high; repeats within 20 days counted once.

43 times historically; 60-day forward avg +10.53%, win rate 48.8%, beating the market by +6.3 pts on average.

Monthly revenue YoY ≥ 20% (from filing deadline)

Pending

Monthly revenue YoY ≥ 20%; event date is the first trading day after the filing deadline (no look-ahead).

103 times historically; 60-day forward avg +8.17%, win rate 57%, beating the market by +6.05 pts on average.

Foreign net buying for 5 straight days

Pending

Foreign net flow (ex-dealer) positive for 5 straight days; event on day 5; repeats within 20 days counted once.

63 times historically; 60-day forward avg +13.83%, win rate 70.5%, beating the market by +8.66 pts on average.

What does 嘉澤 do? Company profile

電子零組件業;連接器、連接線、端子與相關精密零組件 台灣證券交易所上市

Overview

3533 對應證券為嘉澤端子工業股份有限公司普通股,市場別為上市,非 ETF、ETN、權證或特別股。公司成立於 1986-08-23,上市日期為 2007-12-10,總部位於基隆市安樂區武訓街15號。董事長為朱德祥,總經理為何德佑。公開資料顯示實收資本額約新台幣 1,125,346,910 元,已發行普通股約 112,534,691 股。113 年年報揭露截至 2025-03-31 集團員工合計 14,144 人,其中主管 123 人、一般人員 5,325 人、作業人員 8,696 人。主要法人董事股東包含嘉明投資股份有限公司代表人朱德祥,年報列示持股 9,797,037 股、約 8.71%;金嶺投資股份有限公司代表人何德佑,年報列示持股 10,956,273 股、約 9.74%。公司及子公司生產基地與市場布局涵蓋台灣、中國大陸、越南等地;越南廠近年被公司視為擴充伺服器與高階產品產能的重要據點。113 年合併營收為新台幣 300.89 億元,年增 22.90%,稅後淨利約 92.77 億元,EPS 82.77 元;2025 年稅後純益據媒體引述公司公告為 78.65 億元,EPS 70.17 元,擬配發現金股利 35 元;2026 年第 1 季營收約 93.28 億元,EPS 約 21.37 元。

Business

嘉澤是專業連接器與連接線製造商,核心產品為 CPU Socket、DDR 記憶體連接器、PCIe 高速連接器、I/O、M.2、Type-C、線纜、端子、精密模具與其他電子零組件,應用於伺服器、AI 伺服器、高速運算、桌上型電腦、筆記型電腦、汽車電子、工業控制、醫療、網通與消費性電子。113 年年報按產品別揭露,連接器含連接線營收 293.29 億元、占 97.48%,其他電子零組件 7.60 億元、占 2.52%。2025 年第 3 季產品營收結構據法說與媒體報導為伺服器 40.3%(CPU Socket、DDR、PCIe)、桌機 24.64%、筆電 11.03%、策略客戶 9.83%、汽車 8.09%、工業醫療 2.44%、嘉基相關產品 3.63%;2026 年第 1 季法說資料顯示伺服器占比進一步升至約 51.02%,桌機約 17.24%、筆電約 8.99%、策略客戶約 6.83%、車用約 6.92%、工業約 3.6%、嘉基約 5.41%。商業模式為依國際 CPU、GPU、伺服器、PC、車用與工控客戶平台規格共同開發、認證後量產,收入受平台轉換週期、針腳數增加、高速傳輸規格升級與客戶出貨量影響。公司年報揭露主要原料為銅材、塑膠粒與鋼材,主要銷售以外銷為主,113 年外銷比重約 88.57%。公司年報未具名揭露前大客戶,113 年前三大客戶各占營收約 7.19%、7.03%、6.41%,未有單一客戶超過 10%,但年報文字提及公司產品已打入 NVIDIA、Amazon、Tesla 等國際大廠供應鏈;市場與法說報導亦顯示 Intel 與 AMD 伺服器平台升級是 CPU Socket 與高速連接器成長的重要驅動。

Outlook 2025–2028

2025:伺服器、AI 伺服器、桌機與筆電平台世代交替帶動高單價產品滲透,全年仍受惠 CPU Socket、DDR5、PCIe Gen5/Gen6、高速線纜與車用連接器需求;但 2025 年 EPS 70.17 元低於 2024 年 82.77 元,顯示匯率、產品組合、原物料與基期因素仍造成獲利波動。2026:公司法說與媒體報導指向伺服器成為主要成長引擎,2026 年第 1 季伺服器營收占比已突破五成;管理層預期伺服器下半年出貨量較上半年成長約 15%,新平台 ASP 因針腳數增加與 PCIe 6.0 導入可望提升,整體營收有雙位數成長機會。成長動能包含 AMD Venice、Intel Oak Stream/Birch Stream 等新平台、CPU Socket 針腳數增加、PCIe Gen6 連接器與線纜、SO-CAMM、GPU Socket 驗證、車用 BMS 連接器,以及子公司嘉基在 AOC/AEC 與 CPO 相關產品的布局。2027:若 NPO/CPO Socket、AI 伺服器高速互連、PCIe Gen6 及更高階平台量產順利,伺服器與高速傳輸產品可望延續成長;部分法說整理指出 NPO 相關產品預計 2027 年量產,伺服器出貨量仍可能維持成長。2028:較屬推估,若 AI 資料中心、CPO/NPO、PCIe 7.0、DDR6、GPU/CPU 高腳位 Socket 與高速線纜趨勢延續,嘉澤有機會維持高階連接器升級紅利;但 2028 展望能見度低,需視 Intel、AMD、NVIDIA、CSP 與 ODM 平台設計是否持續採用可插拔 Socket 與線纜架構。主要風險包括:原物料銅、金、塑膠粒與鋼材價格上漲;新產品初期良率與越南新廠學習曲線;新台幣升值壓縮外銷獲利;PC/NB 需求疲弱;國際大廠平台延遲;客戶降價壓力;連接器同業競爭與替代架構如 cableless、CPO/NPO 方案演進可能改變既有產品價值鏈。

Supply Chain

Upstream
  • 銅材供應商(公司年報未揭露名稱) 嘉澤主要原料之一為銅材,用於端子、導體與高速連接器;年報僅揭露前大供應商以代號表示,未公開具名,因此不填台股代號。
  • 塑膠粒供應商(公司年報未揭露名稱) 嘉澤主要原料之一為塑膠粒,用於連接器絕緣與外殼材料;年報未公開具名供應商。
  • 鋼材與金屬材料供應商(公司年報未揭露名稱) 嘉澤主要原料之一為鋼材及其他金屬材料,用於結構件、屏蔽件與模具相關製程;年報未公開具名供應商。
  • 精密模具、自動化設備與電鍍製程供應鏈 嘉澤產品產製流程包含產品設計、塑膠與沖壓模具開發、射出、端子沖壓、電鍍、裝配、測試與倉儲運輸;具名供應商未公開。
Downstream
  • NVIDIA 美國上市公司;年報文字提及產品已打入 NVIDIA 供應鏈,主要關聯為 AI 伺服器、高速互連、GPU/伺服器平台相關連接器需求。
  • Amazon 美國上市公司;年報文字提及產品已打入 Amazon 供應鏈,推估關聯為雲端服務供應商與資料中心伺服器需求。
  • Tesla 美國上市公司;年報文字提及產品已打入 Tesla 供應鏈,關聯為車用電子、電動車與高壓/高速連接器需求。
  • Intel 美國上市公司;嘉澤 CPU Socket 與伺服器連接器需求受 Intel 伺服器與 PC 平台世代轉換影響,法說報導提及 Birch Stream、Oak Stream 等平台。
  • AMD 美國上市公司;嘉澤 CPU Socket 與伺服器連接器需求受 AMD 伺服器平台升級影響,法說報導提及 Venice/SP7 等新平台。
  • 雲端服務供應商與伺服器 ODM/OEM 客戶(未完全揭露) 公司前大客戶於年報以 A、B、C、D 公司表示,未具名;法說與新聞指出 CSP 與伺服器客戶是 2025-2026 主要成長來源。
  • 嘉基科技 6715 台股上市櫃公司;為嘉澤持股子公司及集團高速線纜、AOC/AEC、CPO 相關布局成員,並認列於嘉澤合併產品結構中。
  • 技嘉科技 2376 台股上市公司;屬主機板、伺服器與 AI 伺服器品牌/系統廠,為嘉澤產品可能下游應用族群之一;是否為直接客戶未由嘉澤年報具名確認。

Competitors

  • TE Connectivity 美國/瑞士背景國際連接器大廠;在汽車、工業、高速互連與部分 Socket/連接器市場具競爭關係。
  • Amphenol 美國上市國際連接器大廠;在高速、資料中心、通訊、工業與車用連接器市場與嘉澤部分產品線重疊。
  • Molex 美國 Koch 集團旗下未上市連接器大廠;在高速互連、GPU Socket、伺服器與消費電子連接器領域為重要競爭者。
  • 鴻海精密 2317 台股上市公司;旗下連接器、機構與伺服器組裝相關業務龐大,在部分連接器與伺服器供應鏈具競合關係。
  • 貿聯-KY 3665 台股上市公司;高速線束、資料中心、車用與工業連接線/線束供應商,與嘉澤在高速線纜與高階互連領域部分重疊。
  • 信邦電子 3023 台股上市公司;連接線組、工業、醫療、車用與通訊應用供應商,與嘉澤在部分高階連接器與線纜應用市場競爭。
  • 凡甲科技 3526 台股上市櫃公司;連接器廠,產品涵蓋伺服器、NB、消費電子等,與嘉澤在電子連接器領域部分競爭。
  • 佳必琪國際 6197 台股上市公司;連接器、連接線與電子零組件廠,部分 PC、消費電子與高速連接產品線與嘉澤重疊。
  • 胡連精密 6279 台股上市公司;車用端子與連接器供應商,與嘉澤車用連接器布局存在部分競爭。
  • 康控-KY 4943 台股上市公司;聲學、連接器與電子零組件相關供應商,與嘉澤在部分精密電子零組件應用可能重疊。
Sources(12)

This company profile is compiled from public sources (annual reports, filings, news and analyst reports), not finlab backtest data; the 2025–2028 outlook contains forward-looking estimates and is for reference only.

Method and limitations

  • All returns use adjusted close (etl:adj_close), already adjusted for dividends and splits.
  • Excess return is measured against the TAIEX total-return index (dividends included).
  • Valuation regime and percentile stats use an expanding own-history window (no future data).
  • Revenue-signal event dates align to the first trading day after the filing deadline, avoiding look-ahead.
  • Event samples may overlap in time; stats describe historical distribution only and are less reliable when samples are few.

This page is historical statistics for educational purposes and is not investment advice. Investing involves risk; past performance does not indicate future results. Assess and bear the risk yourself before trading.