IT LOOKS LIKE the perfect time to be a chipmaker。 The market for semiconductors continues to grow rapidly。 By the end of the decade it will exceed $1trn globally， up from $500bn this year， forecasts VLSI Research， a firm of analysts。 Demand keeps outstripping supply； the chip shortage is now expected to last well into 2023， paralysing factories of everything that needs processors—which in this day and age is basically everything。 Western governments have earmarked billions to build chipmaking capacity within their borders in order to become less dependent on Asian suppliers。 America alone is planning to spend $52bn over the next five years。

現在似乎是投資晶片製造業的最佳時機。半導體市場在持續快速增長。英國半導體行業研究機構預測，到本世紀末，全球半導體行業市值將從今年的5千億美元增長至10萬億美元以上。晶片短缺、供不應求的局面預計會持續到2023年，這使得所有需要處理器的工廠陷入癱瘓——現如今，幾乎所有商品的製造都離不開處理器。西方各國政府撥出數十億美元在各自境內擴增晶片產能，減輕對亞洲供應商的依賴。僅美國就計劃將於未來五年內在這上面投入520億美元。

In this context the initial public offering （IPO） of GlobalFoundries， a contract manufacturer which makes chips for other firms， seems a safe bet。 The firm， which unveiled its prospectus on October 4th and is expected to list soon， is the world’s fourth-biggest chip foundry by revenues。 The typical characteristics of an IPO—a lowish offering price and a small proportion of shares available to public investors， both of which have yet to be decided— should ensure a healthy “pop” in the share price in the early days of trading。 But GloFo， as semiconductor aficionados endearingly call the firm， is also an example of how tough the chip business has become， not-withstanding the favourable climate。

在這種行情下，晶片製造的承包商格羅方德的首次公開募股似乎勝券在握。這家營收排行全球第四的晶片代工廠在10月4日公佈招股書，預計很快上市。一起IPO的典型特徵——發行價較低並向公眾投資者提供小部分股份（兩者目前都尚未確定）——應確保股價在交易初期實現一次穩健的“大漲”。但是GloFo——半導體愛好者對其的愛稱——也是一個例子，說明了雖然市場前景樂觀，但晶片企業依然舉步維艱。

GloFo is a product of consolidation， caused by the industry’s unforgiving economics that demand ever-tinier silicon furrows and hence ever-costlier fabrication plants （or “fabs”）。 The most advanced of these now cost more than $20bn apiece。 After a spin-off in 2009 from AMD， which designs processors for personal computers and servers in data centres， GloFo later acquired Chartered Semiconductor， another foundry， and the chip-manufacturing business of IBM， a purveyor of assorted information-technology wares。

GloFo是行業合併的產物，原因是該行業嚴格的行規對矽溝的精細度要求越來越高，導致加工裝置也越來越貴。其中最先進的裝置價值超過20億美元。2009年，GloFo從美國超威——為個人電腦和資料中心伺服器設計處理器的半導體公司——分拆出來後，又收購了另一家代工廠，新加坡特許半導體制造公司（Chartered Semiconductor），並承包了美國IBM——資訊科技產品供整合應商——的晶片製造業務。

With billions from Mubadala Investment Company， a sovereign-wealth fund from the United Arab Emirates， which currently owns all of GloFo， the firm tried to keep up with rivals in the race to forge cutting-edge electronic circuitry。 In 2018 it gave up and started catering to the lower end of the market。 These are semiconductors which go into products such as cars and machine tools， and therefore do not need the highest-performing processors， rather than data centres or smartphones。 This niche is still a $54bn market， according to Gartner， another market-researcher。

阿聯酋主權財富基金穆巴達拉投資公司（Mubadala Investment Company）目前擁有GloFo的全部股權，而憑藉其投資的數十億美元，GloFo曾試圖在鍛造尖端電子電路的競爭中跟上對手的步伐。不過，它在2018年就放棄了，開始迎合市場上的低端需求。這些半導體用於汽車和機床等產品而非資料中心或智慧手機，因此不需要最高效能的處理器。據市場研究機構高納德（Gartner）稱，這一利基市場價值仍高達540億美元。

Today GloFo operates a handful of fabs across the world， employs around 15，000 people and has a market share of 7% in the chip-manufacturing business。 Most of its customers， which include AMD， Broadcom， another American chip designer， and NXP， a Dutch one， are “single sourced”。 That means their chips cannot be made by other foundries， such as Samsung of South Korea and in particular Taiwan Semiconductor Manufacturing Company （TSMC）， the world’s mightiest chip manufacturer， which controls more than half the market。

如今，GloFo在全球範圍內擁有多家的工廠，僱用員工約15000名，在晶片製造業務的市場份額為7%。它的大多數客戶，包括超威（AMD）、美國晶片設計公司博通（Broadcom）和荷蘭的恩智浦半導體公司（NXP），與它都是 “單一供應源關係”。這意味著他們的晶片不能由其他代工廠生產，如韓國的三星，尤其是臺灣半導體制造公司臺積電（TSMC）。臺各電是全球最大的晶片製造商，控制著一半以上的市場。

The difference in size goes a long way to explaining why TSMC is hugely profitable whereas GloFo struggles to generate cash。 In the first six months of this year the Taiwanese giant boasted sales of $26bn and profits of $9。8bn。 Although GloFo’s revenues rose to $3bn in the same period， up by nearly 13% on a year ago， and its accounting losses have been narrowing， it still lost $300m between January and June。

規模上的差異充分說明臺積電能賺得盆滿缽滿，而GloFo卻捉襟見肘。今年上半年，臺積電的銷售額達到260億美元，利潤達到98億美元。儘管同期GloFo的收入上升到30億美元，較一年前增長了近13%，會計損失也在持續縮減，在上半年仍虧損了3億美元。

Investing in GloFo will therefore be a wager that the company can ride the current tailwinds in its industry and start making serious money。 But it may also be a bet that another firm snaps up GloFo for itself。 In July it emerged that Intel， the world’s largest chipmaker by revenues， was in takeover talks with the firm。 These did not go anywhere because the parties could not agree on a price。 Once GloFo is listed it should be clearer how much it is worth。 Negotiations could restart。 Then again， with GloFo’s numbers now public， Intel may have a hard time convincing its shareholders that it needs to pay the $25bn that GloFo is expected to fetch。 ■

因此，投資GloFo將是一個賭注，賭該公司能夠藉著當前行業風口開始賺大錢。不過，這一賭注也可能是賭另一家企業併購GloFo。7月，世界第一晶片製造商英特爾與GloFo進行收購談判。由於雙方在價格上無法達成一致，談判無疾而終。一旦GloFo上市，其價值便會不言而喻。談判也可以重啟。不過，GloFo的市值現在屬於公開狀態， 預計購需要支付250億美元，估計英特爾很難說服股東掏腰包。■

編譯：孔珍珍

編輯：翻吧君