Re: [新聞] 最大客戶恐轉單…台積電地位要被動搖了
這種新聞習慣上都會去找一下來源
應該是下面這篇 英文好的可以直接看
後面補上copilot翻譯的中文版本
來源:https://reurl.cc/gYM48R
原文如下
標題:
Apple and Intel 14A: A lifeline for Intel or just a tactical threat against
TSMC?
蘋果與英特爾 14A:是英特爾的救命繩,還是對台積電的戰術威脅?
日期:27. July 2025 06:00 Samir Bashir
Apple flirting with Intel is like the head of the class looking over the math
homework of the nerdy person sitting next to him out of pity – polite,
perhaps genuinely curious, but rarely long-term. Nevertheless, according to
Jeff Pu (GF Securities), what is leaking out is not without explosive force,
report our colleague Muhammad Zuhair from WccfTech: Apple is said to be
evaluating early versions of Intel’s 14A process (1.4 nm). And NVIDIA is
also reportedly putting out feelers. Sounds like silicone romance, but could
be pure power politics at the level of global semiconductor dependencies.
蘋果公司與英特爾之間的接觸,猶如班上成績優異的學生出於善意而翻閱鄰座書呆子的數
學作業——表面上禮貌,或許懷有真誠的好奇心,但極少具有長期性合作的意圖。然而,
根據 GF Securities 的分析師 Jeff Pu 所透露的訊息,我們的同事 Muhammad Zuhair(
WccfTech 報導)指出:此事的外洩內容可能具有重大影響力。據悉,蘋果目前正在評估
英特爾 14A 製程(1.4 奈米)之初期版本;此外,輝達(NVIDIA)亦傳出正在釋出合作
意向。這一系列動作表面上看似半導體領域的策略合作,但更可能代表一場圍繞全球半導
體供應鏈依賴所展開的權力競局。
Intel 14A: A process between hope and ultimatum
The story is quickly told, but deeply ramified: Intel has its back to the
wall. The foundry business is in deficit, the industry’s trust is as full of
holes as Swiss cheese – just think of Intel’s failed 10 nm odyssey and the
eternal promise “next year we’ll be back in front”. So now the big hit:
14A – a process node based on the PowerVia backbone of the 18A, spiced up
with 2nd Gen RibbonFET and PowerDirect. Sounds futuristic, yes, but above all
it is a desperate appeal: “Buy from us, otherwise we’ll drop out of the
node rally!” An ultimatum that means something like: If Apple and NVIDIA don
’t get on board, that’s it for Intel’s ambition to become a semiconductor
superpower again.
Intel 14A:介於希望與最後通牒之間的製程節點
這個故事雖可簡短敘述,卻牽動深遠:英特爾目前處於艱困處境。其晶圓代工業務長期虧
損,產業對其信任如同瑞士乳酪般千瘡百孔——回顧英特爾在 10 奈米製程上的挫敗歷程
,以及多年來未曾實現的「明年我們將重返領先」承諾,即可略見一斑。
如今,英特爾寄予厚望的突破點是 14A 製程節點。該技術建立在 18A 製程的 PowerVia
架構之上,並結合第二代 RibbonFET 及 PowerDirect 技術。儘管展現未來性,最核心的
仍是一項迫切呼籲:「請選擇我們,否則我們將退出先進製程競爭。」
這項訊息的實質意涵即是:若蘋果與輝達未選擇採用該製程節點,英特爾重新奪回半導體
領導地位的企圖恐將功敗垂成。
Apple: between diversification and blackmail
Why would Apple, TSMC’s premium customer, suddenly come knocking on Intel’s
door? Quite simply: pressure. TSMC is currently holding the chip world
hostage. If you want high-end, you have to pay. And those who don’t pay
simply don’t get any capacity or are put off until later slots. An untenable
situation for Apple. Especially in view of the fact that TSMC will not bring
its new production facilities (A14, not to be confused with Apple’s SoC of
the same name) into series production until 2028. A dual sourcing model with
Intel would therefore be more of a threat than a technical necessity. Apple
is showing TSMC the yellow card: “If you become too expensive or too slow, we
’ll go to Intel.” Of course, only if Intel delivers at all by then. And
this is precisely where the problem lies: PDK samples and 5-nines yield are
worlds apart.
Apple:介於供應多元化與策略施壓之間
蘋果公司作為台積電的核心高端客戶,為何突然轉向英特爾表達興趣?原因其實非常明確
:市場壓力。台積電目前在高階晶片製程領域擁有主導地位,對整個晶片市場形成高度依
賴。欲取得先進製程的產能,客戶須支付高額代價;未願付出者則往往被排入後續時程或
直接排除於外。這樣的情況對蘋果而言已難以承受。
尤其需要注意的是,台積電預計要到 2028 年才會將其最新的 A14 製程技術(注意:與
蘋果自家命名的 A14 SoC 並無關聯)導入量產階段。在此背景下,與英特爾啟動雙重採
購模式,更像是蘋果的一項策略性威脅,而非出於技術需求。其意圖明確:向台積電示警
——「若你變得太昂貴或進度太慢,我們將轉向英特爾。」
但此舉能否付諸實行,仍取決於英特爾是否具備交付能力。事實上,僅有設計開發套件(
PDK)並不代表已達量產良率標準。在業界要求達成「五個九」(99.999%)良率的高度門
檻下,設計階段與實際生產間仍存在巨大的技術落差。
NVIDIA: The AI monolith is looking for new foundries
The fact that NVIDIA is also being touted as an interested party is less
surprising. The entire AI industry is currently steaming like an overheated
reactor, and TSMC’s output is not enough. A second supplier would be worth
its weight in gold for NVIDIA – especially in the mid-range GPU segment. But
here too, everything is still on paper. No tape-outs, no test chips, no
concrete deals. Rumor has it that Intel has already poached high-ranking TSMC
engineers from Arizona – a tactical move, but no proof of production
readiness. And manufacturing is not a LinkedIn profile, but hard
infrastructure, long-term contracts, machines, lithography expertise and,
last but not least, trust.
NVIDIA:AI 巨頭積極尋找新晶圓代工夥伴
NVIDIA 被認為是可能的合作方之一,其實並不令人意外。眼下整個 AI 產業如同過熱的
反應爐般迅速膨脹,而台積電的產能顯然無法充分應對市場需求。對 NVIDIA 而言,特別
是在中階 GPU 市場領域,尋求第二供應來源將極具戰略價值。
然而,目前所有計畫仍停留在紙上談兵階段——尚無設計定案(tape-out)、無測試晶片
,也未達成任何具體合作協議。有市場傳聞指出,英特爾已在亞利桑那州挖角多位台積電
高階工程師,作為策略佈局的一環,但此舉並不能視為具備量產能力的明確證據。
值得強調的是,晶片製造不僅是人才履歷的堆疊,更是建構於實體基礎設施之上的長期投
資:包括生產設備、微影技術專業、穩定的契約架構,以及最重要的——產業信任。
A strategic mix: semiconductors as a diplomatic currency
What at first glance seems like a minor industry announcement is actually
part of a global game. Intel – once the guardian of Moore’s Law – is now a
supplicant. Apple – the tech oligarch of the post-PC era – is using its
market position to play suppliers off against each other. And TSMC – the
quasi-monopolist – threatens to become the structural weak point of the
Western tech world, especially in view of Taiwan’s geopolitical fragility.
If Apple gets involved with Intel, it will not be out of charity, but out of
strategic calculation: spreading risk, pushing down prices, securing
production time. If Intel is able to deliver – which as things stand today
is more of a wish than a reality – an alternative axis of power will emerge
in the chip war.
戰略混合:半導體作為外交籌碼
乍看之下只是一則產業消息,實際上卻是全球布局的一環。英特爾——曾經是摩爾定律的
守護者,如今成了市場的請求者;蘋果——後 PC 時代的科技寡頭——正運用其市場地位
,在供應商之間進行策略性的博弈;而台積電——近乎壟斷的晶圓代工巨擘——在台灣地
緣政治脆弱性的影響下,有可能成為西方科技體系的結構性弱點。
若蘋果選擇與英特爾合作,動機絕非基於善意,而是基於精密的策略考量:分散供應風險
、壓低生產成本,並確保晶片製造時程。如果英特爾能夠如期達成交付能力——以目前情
況而言仍屬願景而非現實——則在半導體競爭格局中將可能形成一條新的權力軸線。
Conclusion: beacon of hope or smokescreen?
For Intel, an Apple or NVIDIA deal with 14A would be the saving straw. For
Apple, it is an option for a better negotiating position. A warning shot for
TSMC. But nothing is fixed yet. No wafer has been delivered, no SoC is
running. Only PDKs, roadmaps and analyst speculation. You could also say that
the ball is on the pitch – but the pitch is still under construction. And
whether Intel plays along or is left on the bench again will only be decided
when Apple says: “Tape-out, we’re in.” Until then, it’s all just
diplomatic haggling – with silicon as the pawn sacrifice.
結論:希望燈塔,抑或煙霧彈?
對英特爾而言,與蘋果或輝達在 14A 製程上的合作,無疑可能成為扭轉局勢的關鍵契機
;而對蘋果而言,這是一項提升自身談判籌碼的策略選項,同時也是對台積電釋出的預警
信號。
然而,目前尚無任何確切成果:既未交付晶圓,也無系統晶片進入實際運作階段。現階段
僅有設計開發套件(PDK)、技術藍圖以及市場分析推測。形象地說,雖然「比賽用球已
放上場」,但球場本身仍在施工中。至於英特爾是否能夠正式參與競賽,抑或再度被排除
在外,將取決於蘋果是否做出關鍵表態:「Tape-out,我們加入。」
在此之前,這場角力仍只是充滿策略意涵的外交談判——以半導體作為談判桌上的犧牲籌
碼。
Source: WccfTech, 9to5mac
心得:
看了原文,就是一個市場的消息而已,
如同原文的結論所說,未交付晶圓也沒有實際運作,都是市場推測
至於為何變成台灣新聞之後都會很驚悚就不得而知了XD
難不成又是傳說中的"嚇甩磨"嗎~
各位半導體鍵盤分析師怎麼看
--
※ 發信站: 批踢踢實業坊(ptt.cc), 來自: 163.17.149.199 (臺灣)
※ 文章網址: https://www.ptt.cc/bbs/Stock/M.1753600625.A.AC8.html
推
07/27 15:23,
3小時前
, 1F
07/27 15:23, 1F
推
07/27 15:26,
3小時前
, 2F
07/27 15:26, 2F
→
07/27 15:26,
3小時前
, 3F
07/27 15:26, 3F
推
07/27 15:28,
3小時前
, 4F
07/27 15:28, 4F
→
07/27 15:28,
3小時前
, 5F
07/27 15:28, 5F
→
07/27 15:28,
3小時前
, 6F
07/27 15:28, 6F
推
07/27 15:31,
3小時前
, 7F
07/27 15:31, 7F
推
07/27 15:34,
3小時前
, 8F
07/27 15:34, 8F
推
07/27 15:35,
3小時前
, 9F
07/27 15:35, 9F
推
07/27 15:35,
3小時前
, 10F
07/27 15:35, 10F
→
07/27 15:35,
3小時前
, 11F
07/27 15:35, 11F
推
07/27 15:39,
3小時前
, 12F
07/27 15:39, 12F
推
07/27 15:40,
3小時前
, 13F
07/27 15:40, 13F
→
07/27 15:44,
2小時前
, 14F
07/27 15:44, 14F
→
07/27 15:44,
2小時前
, 15F
07/27 15:44, 15F
→
07/27 15:45,
2小時前
, 16F
07/27 15:45, 16F
推
07/27 15:47,
2小時前
, 17F
07/27 15:47, 17F

推
07/27 15:49,
2小時前
, 18F
07/27 15:49, 18F
→
07/27 15:54,
2小時前
, 19F
07/27 15:54, 19F
推
07/27 15:55,
2小時前
, 20F
07/27 15:55, 20F
→
07/27 15:55,
2小時前
, 21F
07/27 15:55, 21F
→
07/27 15:55,
2小時前
, 22F
07/27 15:55, 22F
推
07/27 15:57,
2小時前
, 23F
07/27 15:57, 23F
推
07/27 15:59,
2小時前
, 24F
07/27 15:59, 24F
推
07/27 16:03,
2小時前
, 25F
07/27 16:03, 25F
推
07/27 16:09,
2小時前
, 26F
07/27 16:09, 26F
→
07/27 16:09,
2小時前
, 27F
07/27 16:09, 27F
→
07/27 16:09,
2小時前
, 28F
07/27 16:09, 28F
→
07/27 16:09,
2小時前
, 29F
07/27 16:09, 29F
→
07/27 16:09,
2小時前
, 30F
07/27 16:09, 30F
推
07/27 16:10,
2小時前
, 31F
07/27 16:10, 31F
→
07/27 16:10,
2小時前
, 32F
07/27 16:10, 32F
推
07/27 16:14,
2小時前
, 33F
07/27 16:14, 33F
推
07/27 16:22,
2小時前
, 34F
07/27 16:22, 34F
推
07/27 16:26,
2小時前
, 35F
07/27 16:26, 35F
→
07/27 16:26,
2小時前
, 36F
07/27 16:26, 36F
噓
07/27 16:29,
2小時前
, 37F
07/27 16:29, 37F
→
07/27 16:29,
2小時前
, 38F
07/27 16:29, 38F
推
07/27 16:36,
2小時前
, 39F
07/27 16:36, 39F
推
07/27 16:40,
2小時前
, 40F
07/27 16:40, 40F
→
07/27 16:43,
1小時前
, 41F
07/27 16:43, 41F
→
07/27 16:45,
1小時前
, 42F
07/27 16:45, 42F
→
07/27 16:57,
1小時前
, 43F
07/27 16:57, 43F
→
07/27 16:59,
1小時前
, 44F
07/27 16:59, 44F
→
07/27 17:05,
1小時前
, 45F
07/27 17:05, 45F
推
07/27 17:09,
1小時前
, 46F
07/27 17:09, 46F
推
07/27 17:29,
1小時前
, 47F
07/27 17:29, 47F
→
07/27 17:31,
1小時前
, 48F
07/27 17:31, 48F
→
07/27 17:31,
1小時前
, 49F
07/27 17:31, 49F
推
07/27 17:58,
44分鐘前
, 50F
07/27 17:58, 50F
推
07/27 17:59,
43分鐘前
, 51F
07/27 17:59, 51F
→
07/27 17:59,
43分鐘前
, 52F
07/27 17:59, 52F
推
07/27 18:04,
38分鐘前
, 53F
07/27 18:04, 53F
→
07/27 18:10,
32分鐘前
, 54F
07/27 18:10, 54F
→
07/27 18:20,
22分鐘前
, 55F
07/27 18:20, 55F
→
07/27 18:21,
21分鐘前
, 56F
07/27 18:21, 56F
推
07/27 18:26,
16分鐘前
, 57F
07/27 18:26, 57F
討論串 (同標題文章)
完整討論串 (本文為第 2 之 3 篇):
265
464
Stock 近期熱門文章
265
464
35
109
PTT職涯區 即時熱門文章