當(dāng)前全世界的半導(dǎo)體嚴(yán)重缺貨,導(dǎo)致汽車廠商和電子廠商不得不推遲發(fā)布新產(chǎn)品?,F(xiàn)在并沒(méi)有快速擺脫困境的有效方法。全球最大的芯片代工廠臺(tái)灣積體電路制造股份公司(簡(jiǎn)稱“臺(tái)積電”)預(yù)計(jì),未來(lái)半導(dǎo)體的需求依舊旺盛。為了滿足激增的需求,臺(tái)積電計(jì)劃在未來(lái)三年投入高達(dá)1,000億美元以提高產(chǎn)能。
臺(tái)積電的一位發(fā)言人發(fā)電子郵件給《財(cái)富》雜志稱,公司的投資預(yù)計(jì)“未來(lái)幾年過(guò)后會(huì)出現(xiàn)增長(zhǎng)”。發(fā)言人還強(qiáng)調(diào),臺(tái)積電的1,000億美元投資計(jì)劃并不是針對(duì)目前的芯片短缺問(wèn)題?!拔覀円恢迸c客戶密切合作,而且我們不會(huì)為了投機(jī)擴(kuò)大產(chǎn)能。”
臺(tái)積電在今年1月的投資者會(huì)議上宣布,其1,000億美元投資將用于“增加產(chǎn)能,以支持生產(chǎn)和先進(jìn)半導(dǎo)體技術(shù)的研發(fā)”,但國(guó)際媒體最近才獲悉這一消息,并且依舊對(duì)這個(gè)龐大的數(shù)字感到震驚。
貝恩公司(Bain & Co.)的合伙人拉維·韋賈伊拉貢卡文說(shuō):“在芯片行業(yè)投資1,000億美元或許不像其他行業(yè)那樣令人瞠目結(jié)舌。”
事實(shí)上,芯片開(kāi)發(fā)成本高昂,要花光1,000億美元并不難。
臺(tái)積電計(jì)劃在菲尼克斯新建的半導(dǎo)體工廠預(yù)算為120億美元,定于明年開(kāi)始施工。上周,競(jìng)爭(zhēng)對(duì)手英特爾也宣布投資200億美元,同樣在亞利桑那州新建兩家芯片工廠。
韋賈伊拉貢卡文稱:“很大一部分成本來(lái)自生產(chǎn)設(shè)備,每新發(fā)明一代芯片,其生產(chǎn)設(shè)備的成本就越高?!卑雽?dǎo)體通常按照芯片上每一個(gè)晶體管之間間隔的納米(nm)距離進(jìn)行分類。間隙越窄,芯片的功能越強(qiáng)大,生產(chǎn)難度自然越大。
今年1月,臺(tái)積電的首席財(cái)務(wù)官黃仁昭(Wendell Huang)表示,公司僅今年就計(jì)劃投入280億美元,比2020年的資本支出增加了近100億美元。黃仁昭稱,今年約80%的資本支出將用于公司三款最先進(jìn)的芯片制程,分別是7納米、5納米和3納米芯片。
目前只有臺(tái)積電和三星在生產(chǎn)5納米芯片,預(yù)計(jì)英特爾在2022年可以生產(chǎn)復(fù)雜程度相對(duì)較低的7納米芯片。由于業(yè)內(nèi)有能力生產(chǎn)高端芯片的廠商少之又少,因此臺(tái)積電需要滿足對(duì)高端芯片的絕大部分需求,所以其他廠商可能不會(huì)效仿臺(tái)積電投資1,000億美元。
市場(chǎng)研究機(jī)構(gòu)Counterpoint Research的分析師布拉迪·王說(shuō):“低端芯片制造商發(fā)現(xiàn)需求激增,是因?yàn)榭蛻粼谛鹿谝咔槠陂g紛紛囤貨,增加了訂單。”他認(rèn)為,疫情結(jié)束之后,芯片需求會(huì)恢復(fù)到正常水平。但高端芯片的需求飆升卻是得益于技術(shù)進(jìn)步的長(zhǎng)期趨勢(shì)。
布拉迪·王表示,電動(dòng)汽車和自動(dòng)駕駛汽車、云計(jì)算數(shù)據(jù)中心、5G手機(jī)等都增加了對(duì)更尖端的計(jì)算能力的需求。有媒體爆料,臺(tái)積電已經(jīng)獲得了蘋(píng)果和英特爾的大宗訂單,這些訂單導(dǎo)致臺(tái)積電目前的先進(jìn)制程產(chǎn)能不足,因此必須擴(kuò)大產(chǎn)能。
目前尚無(wú)法確定臺(tái)積電未來(lái)三年的1,000億美元預(yù)算是否會(huì)成為未來(lái)投資的新基準(zhǔn)。韋賈伊拉貢卡文認(rèn)為,臺(tái)積電未來(lái)的投入可能取決于三星、英特爾、聯(lián)華電子和格芯(GlobalFoundries)等主要競(jìng)爭(zhēng)對(duì)手的投入力度。但如果競(jìng)爭(zhēng)對(duì)手增加了全球產(chǎn)能,臺(tái)積電也不需要承擔(dān)起如此大的供應(yīng)量。
韋賈伊拉貢卡文說(shuō):“有競(jìng)爭(zhēng)是好事。你會(huì)希望保持領(lǐng)先優(yōu)勢(shì)。”(財(cái)富中文網(wǎng))
翻譯:劉進(jìn)龍
審校:汪皓
當(dāng)前全世界的半導(dǎo)體嚴(yán)重缺貨,導(dǎo)致汽車廠商和電子廠商不得不推遲發(fā)布新產(chǎn)品?,F(xiàn)在并沒(méi)有快速擺脫困境的有效方法。全球最大的芯片代工廠臺(tái)灣積體電路制造股份公司(簡(jiǎn)稱“臺(tái)積電”)預(yù)計(jì),未來(lái)半導(dǎo)體的需求依舊旺盛。為了滿足激增的需求,臺(tái)積電計(jì)劃在未來(lái)三年投入高達(dá)1,000億美元以提高產(chǎn)能。
臺(tái)積電的一位發(fā)言人發(fā)電子郵件給《財(cái)富》雜志稱,公司的投資預(yù)計(jì)“未來(lái)幾年過(guò)后會(huì)出現(xiàn)增長(zhǎng)”。發(fā)言人還強(qiáng)調(diào),臺(tái)積電的1,000億美元投資計(jì)劃并不是針對(duì)目前的芯片短缺問(wèn)題。“我們一直與客戶密切合作,而且我們不會(huì)為了投機(jī)擴(kuò)大產(chǎn)能。”
臺(tái)積電在今年1月的投資者會(huì)議上宣布,其1,000億美元投資將用于“增加產(chǎn)能,以支持生產(chǎn)和先進(jìn)半導(dǎo)體技術(shù)的研發(fā)”,但國(guó)際媒體最近才獲悉這一消息,并且依舊對(duì)這個(gè)龐大的數(shù)字感到震驚。
貝恩公司(Bain & Co.)的合伙人拉維·韋賈伊拉貢卡文說(shuō):“在芯片行業(yè)投資1,000億美元或許不像其他行業(yè)那樣令人瞠目結(jié)舌。”
事實(shí)上,芯片開(kāi)發(fā)成本高昂,要花光1,000億美元并不難。
臺(tái)積電計(jì)劃在菲尼克斯新建的半導(dǎo)體工廠預(yù)算為120億美元,定于明年開(kāi)始施工。上周,競(jìng)爭(zhēng)對(duì)手英特爾也宣布投資200億美元,同樣在亞利桑那州新建兩家芯片工廠。
韋賈伊拉貢卡文稱:“很大一部分成本來(lái)自生產(chǎn)設(shè)備,每新發(fā)明一代芯片,其生產(chǎn)設(shè)備的成本就越高?!卑雽?dǎo)體通常按照芯片上每一個(gè)晶體管之間間隔的納米(nm)距離進(jìn)行分類。間隙越窄,芯片的功能越強(qiáng)大,生產(chǎn)難度自然越大。
今年1月,臺(tái)積電的首席財(cái)務(wù)官黃仁昭(Wendell Huang)表示,公司僅今年就計(jì)劃投入280億美元,比2020年的資本支出增加了近100億美元。黃仁昭稱,今年約80%的資本支出將用于公司三款最先進(jìn)的芯片制程,分別是7納米、5納米和3納米芯片。
目前只有臺(tái)積電和三星在生產(chǎn)5納米芯片,預(yù)計(jì)英特爾在2022年可以生產(chǎn)復(fù)雜程度相對(duì)較低的7納米芯片。由于業(yè)內(nèi)有能力生產(chǎn)高端芯片的廠商少之又少,因此臺(tái)積電需要滿足對(duì)高端芯片的絕大部分需求,所以其他廠商可能不會(huì)效仿臺(tái)積電投資1,000億美元。
市場(chǎng)研究機(jī)構(gòu)Counterpoint Research的分析師布拉迪·王說(shuō):“低端芯片制造商發(fā)現(xiàn)需求激增,是因?yàn)榭蛻粼谛鹿谝咔槠陂g紛紛囤貨,增加了訂單?!彼J(rèn)為,疫情結(jié)束之后,芯片需求會(huì)恢復(fù)到正常水平。但高端芯片的需求飆升卻是得益于技術(shù)進(jìn)步的長(zhǎng)期趨勢(shì)。
布拉迪·王表示,電動(dòng)汽車和自動(dòng)駕駛汽車、云計(jì)算數(shù)據(jù)中心、5G手機(jī)等都增加了對(duì)更尖端的計(jì)算能力的需求。有媒體爆料,臺(tái)積電已經(jīng)獲得了蘋(píng)果和英特爾的大宗訂單,這些訂單導(dǎo)致臺(tái)積電目前的先進(jìn)制程產(chǎn)能不足,因此必須擴(kuò)大產(chǎn)能。
目前尚無(wú)法確定臺(tái)積電未來(lái)三年的1,000億美元預(yù)算是否會(huì)成為未來(lái)投資的新基準(zhǔn)。韋賈伊拉貢卡文認(rèn)為,臺(tái)積電未來(lái)的投入可能取決于三星、英特爾、聯(lián)華電子和格芯(GlobalFoundries)等主要競(jìng)爭(zhēng)對(duì)手的投入力度。但如果競(jìng)爭(zhēng)對(duì)手增加了全球產(chǎn)能,臺(tái)積電也不需要承擔(dān)起如此大的供應(yīng)量。
韋賈伊拉貢卡文說(shuō):“有競(jìng)爭(zhēng)是好事。你會(huì)希望保持領(lǐng)先優(yōu)勢(shì)。”(財(cái)富中文網(wǎng))
翻譯:劉進(jìn)龍
審校:汪皓
The world is currently facing a severe semiconductor shortage that has forced automakers and electronics manufacturers to delay the launch of new product lines. There’s no quick way out of the crunch. The world’s largest contract chip manufacturer, Taiwan Semiconductor Manufacturing Co. (TSMC), is expecting demand to remain high—so high, in fact, that it’s planning to invest a staggering $100 billion over the next three years to supply it.
The investment anticipates “growth that will follow in the next few years,” a TSMC spokesperson said in an email to Fortune, adding that the proposed $100 billion is not a response to the current chip shortage. “We always work closely with our customers, and we do not build speculative capacity.”
TSMC announced its $100 billion investment to “increase capacity to support the manufacturing and R&D of advanced semiconductor technologies” during an investors conference in January, but international media only recently became aware of the news and is still wrapping its head around the enormous sum.
“In the chip business maybe $100 billion is a little less eye-popping than in other industries, but it’s still a big number,” says Ravi Vijayaraghavan, a partner at consultancy Bain & Co.
Truth is, semiconductor development is an expensive business, and $100 billion won’t be at all hard to spend.
TSMC has budgeted $12 billion for the development of a new semiconductor factory in Phoenix, with construction due to start next year. Last week, competitor Intel likewise committed $20 billion to building two new plants, also in Arizona.
“A big chunk of the cost is in the manufacturing equipment, which gets more expensive with every new generation of chip,” Vijayaraghavan says. Semiconductors are typically categorized by the distance between each transistor on the chip, measured in nanometers (nm). The tighter the gap, the more powerful the chip, and the trickier it is to make.
In January, TSMC CFO Wendell Huang said that the chipmaker is committed to spending up to $28 billion this year alone—a roughly $10 billion increase over the group’s 2020 capital expenditure. Roughly 80% of this year’s capex, Huang said, will be spent on the company’s three most advanced chipmaking processes—namely 7nm, 5nm, and 3nm chipsets.
TSMC and Samsung are the only chip manufacturers currently producing 5nm chips, with Intel expecting to manufacture less-sophisticated 7nm chips in 2022. With so few players on the cutting edge of the industry, TSMC has been left to satisfy a significant proportion of advanced demand, which is why its $100 billion investment might not be replicated by other players.
“Manufacturers of less-sophisticated chips are seeing a surge in demand because customers were stockpiling during the pandemic, increasing orders,” says Brady Wang, an analyst at Counterpoint Research, suggesting demand volumes will return to normal post-pandemic. But the surging demand for cutting-edge chips is buoyed by a long-term trend of advancing tech.
Electric and autonomous vehicles, cloud computing data centers, and 5G phones are all driving up the need for more sophisticated computing power, Wang says. TSMC, specifically, has reportedly scored major contracts from Apple and Intel, too—essentially maxing out the chipmaker’s current advanced capacity and necessitating the need to add more.
Whether TSMC’s budget of $100 billion over three years will become a new baseline for the company’s future investment is unclear. According to Vijayaraghavan, TSMC’s future spending might depend upon how much its closest competitors—Samsung, Intel, UMC, and GlobalFoundries—spend next. Because if competitors improve global capacity, TSMC won’t be on the hook to provide so much supply.
“It’s good to have competition,” Vijayaraghavan says. “You just want to keep an edge.”