跳到主要內容

AI 的下一個瓶頸會是「一束光」嗎?磷化銦(InP)缺料風暴如何重塑光通訊產業?

一句話結論: AI 算力的傳輸正從電互連走向光互連,而光互連的核心材料磷化銦(InP)產能高度集中、擴產週期長,供應鏈警告「InP 缺料將比記憶體更缺」,這輪缺料風暴正在重塑整個光通訊產業的版圖。
Sivers 攜手 SemiNex 啟動 340 萬美元磷化銦光源計畫,瞄準共同封裝光學(CPO)

一條供應鏈警告,讓「一束光」成為熱搜

8 月中旬,台灣熱搜榜突然出現「磷化銦缺料」話題。工商時報以「磷化銦 InP 缺料風暴襲來,將比記憶體更缺?」為題,遠見雜誌則追問「磷化銦是下個稀土?光通訊概念股漲翻後的隱憂」。對多數人來說,磷化銦聽起來像某種冷門化學品,但對 AI 資料中心而言,它是 800G/1.6T 光模組與共同封裝光學(CPO)的關鍵材料——沒有它,GPU 之間的光纖通道就無法點亮。

磷化銦是什麼?為何 AI 突然離不開它?

磷化銦(Indium Phosphide, InP)是 III-V 族化合物半導體,具備極高的電子遷移率與直接能隙特性,是製造雷射二極體(EML、CW 雷射)與高速光偵測器的核心材料。傳統資料中心用銅線傳電訊號,但在 800G、1.6T 甚至 3.2T 的頻寬需求下,銅線的功耗與距離限制到了極限,業界轉向「光互連」:訊號轉成光、走光纖、再轉回電。

這意味著每一條 AI 伺服器互連通道都需要雷射光源。光模組從可插拔(pluggable)走向共同封裝光學(CPO)——把光引擎直接封進交換器晶片旁——對 InP 雷射的數量與功率需求只增不減。簡言之:AI 算力堆得越快,需要的「光」就越多,而光來自 InP。

兩則新聞揭開缺料真相

Sivers 宣布與 SemiNex 的 3.4M 美元開發計畫:高功率外部雷射、DFB 陣列與 SOA 增益段

第一則:瑞典光晶片公司 Sivers Semiconductors 與美國 SemiNex 宣布一項 340 萬美元的聯合開發計畫,目標是「AI 資料中心互連與共同封裝光學所需的下一代 InP 光源」——包括高功率外部雷射、供波長多工用的 DFB 雷射陣列,以及 SOA 光放大器增益段。客戶送樣與初期量產鎖定 2027 下半年,正好對齊 CPO 的規模化時程。第二則:光通訊大廠 Coherent 在法說會直言「磷化銦產能仍是我們的首要限制條件」(indium phosphide capacity continues to be our primary constraint),並透露其 6 月季度 InP 雷射產出年增約 80%,正在把產線從 3 吋晶圓轉向 6 吋——後者產出約為前者的四倍、晶圓成本約一半,良率還更高。

Coherent 法說會:InP 是首要限制條件,雷射產出年增 80%,6 吋平台產出約為 3 吋四倍

另一家光模組廠 Applied Optoelectronics(AAOI)則指出,收發器市場存在 20% 到 40% 的需求失衡,瓶頸就在雷射光源。三家公司的訊息指向同一個結論:AI 光學瓶頸正從「模組組裝」向上游移動到「光源與 InP 材料」本身。

缺料的連鎖反應:從晶圓到伺服器交期

Sivers 產能模型回顧:產業共識是雷射為瓶頸,光學元件廠進入賣方市場

InP 缺料之所以棘手,在於供給端的結構性限制:全球 InP 磊晶與晶圓產能高度集中在少數廠商(如住友電工、AXT 子公司等),新產能從投資到量產動輒 12 到 18 個月,擴產週期遠長於矽晶圓。AI 伺服器交期因此從「等 GPU」變成「等 GPU、也等光模組、更等雷射」。對終端業者來說,缺料直接拉長 AI 叢集的建置時程;對投資人來說,則要學會分辨誰是「真受惠」——掌握 InP 磊晶、雷射封裝或 CPO 光源能力的廠商,與只掛上「光通訊」標籤蹭題材的股票,接下來走勢會明顯分化。

短期內,InP 缺料沒有速效解。但供應鏈的反應(6 吋轉換、跨廠合作、提前鎖產能)已經讓產業方向變得清晰:AI 的下一步,卡在「一束光」上,而這束光的關鍵,就是磷化銦。

常見問題 (FAQ)

Q1:磷化銦(InP)是什麼?為什麼 AI 需要它?

InP 是 III-V 族化合物半導體,能高效發出與接收光訊號,是光通訊雷射(EML、CW 雷射)與高速光偵測器的核心材料。AI 資料中心頻寬需求從 800G 往 1.6T/3.2T 推升,銅線傳輸到極限後改用光互連,每一條光通道都需要 InP 雷射光源。

Q2:為什麼說 InP 缺料「比記憶體更缺」?

記憶體(HBM、DRAM)產能分散在多家大廠、擴產相對快;InP 磊晶與晶圓產能高度集中,擴產週期長達 12 到 18 個月,加上 AI 光互連需求爆發式成長,供需缺口短期內難以填補。

Q3:CPO(共同封裝光學)與傳統光模組差在哪?

傳統光模組是可插拔的獨立元件,插在交換器面板上;CPO 把光引擎直接封裝進交換器晶片附近,省去面板連接,降低功耗與延遲,但對雷射光源的數量、功率與可靠性要求更高。

Q4:Sivers 與 SemiNex 的 340 萬美元計畫重要嗎?

金額不大,但象徵意義重要:它顯示「高功率外部雷射、DFB 陣列、SOA 增益段」這類 CPO 專用 InP 光源已成為產業卡位重點,且量產時程(2027 下半年)對齊 CPO 規模化,是供應鏈提前佈局的訊號。

Q5:投資人如何分辨「真受惠」的光通訊股?

看產業鏈位置:掌握 InP 磊晶/晶圓、雷射設計與封裝、或 CPO 光源能力的廠商是真受惠;僅掛光通訊標籤、沒有對應產品與訂單的公司屬於蹭題材。另外要追蹤法說會提到的「瓶頸」詞彙——瓶頸在哪裡,利潤就在哪裡。

留言

這個網誌中的熱門文章

Intel 14A Defect Density Is Its Best Since 22nm — Is Intel Back in the Leading-Edge Race?

One-sentence takeaway: Intel's 14A process is cutting defect density faster than any node since 22nm, and customers have moved from watching to asking about capacity — if risk production stays on track for H2 2027, it's the strongest signal yet that Intel is back in the leading-edge game. "We have not seen this performance since 22nm." When Intel CFO David Zinsner dropped that line at the Deutsche Bank 2026 technology conference, the semiconductor world took notice. 14A — Intel's first 1.4nm-class node — is backing up the company's comeback story with data, not slogans. What is 14A, and why it matters 14A is Intel's most advanced planned process node, a "1.4nm-class" technology targeting high-volume manufacturing in 2028. It packs three headline technologies: second-generation RibbonFET gate-all-around transistors, PowerDirect backside power delivery, and High-NA EUV lithography. In short, it's the most technically complex node Intel ...

Google's Antitrust Remedies Enter Deep Water: Breakup, AI Mode, and the Browser

Bottom line: The U.S. DOJ's remedies phase against Google is redefining the commercial rules of "search" — from Chrome's fate to AI distribution and the ad business, every step could reshape global tech. Google's search monopoly case has been called "the most important antitrust case of the internet era." In August 2024, a federal judge ruled Google violated antitrust law; now the remedies phase is in deep water. The DOJ's proposals include breaking up the ad business, divesting Chrome, and ending default search agreements — each step ripples through the entire tech industry. Timeline: from monopoly ruling to remedies In August 2024, the D.C. federal court ruled that Google violated the Sherman Act by paying billions annually to make Apple, Samsung, and others set Google as the default search engine. The remedies trial runs through 2026, with DOJ options including: Breaking up the ad business: Google's ad tech stack is accused of stifl...

Why Is NVIDIA Spending Billions to Buy Up America's "Dark Fiber"?

One-line conclusion: NVIDIA is reportedly spending $5–10 billion to acquire long-haul "dark fiber" networks across the United States, signaling that the AI infrastructure race is shifting from raw compute power to the networks that connect it. NVIDIA is reportedly acquiring long-haul "dark fiber" networks across the United States, with total capacity estimated at 7.6 Pbps and a price tag between $5 billion and $10 billion. The news sent optical communications stocks surging globally: Taiwan's optical module makers jumped on July 22, and three more hit the daily limit on July 23. Many now read this as the moment the AI arms race moved from "who has more GPUs" to "who owns the network." What Is Dark Fiber, and Why Buy Instead of Lease? Dark fiber refers to fiber-optic cable that has already been laid but has no transmission equipment installed and carries no optical signal . The fiber cores sit "dark" and dormant, waiting to...