在經(jīng)過了好長(zhǎng)一段時(shí)間后，中國(guó)晶圓代工業(yè)者中芯國(guó)際(SMIC)終于有一些好消息可以透露；該公司希望產(chǎn)業(yè)界能擺脫對(duì)中芯過往的負(fù)面印象：包括沉重的虧損、與臺(tái)灣同業(yè)的訴訟糾紛，以及公司內(nèi)部管理高層的多次異動(dòng)等等。
如今的中芯年度營(yíng)收創(chuàng)下17億美元的最高紀(jì)錄(2012年)，正收益(positive income)則達(dá)到1,590萬美元(創(chuàng)七年新高紀(jì)錄)，還有個(gè)面帶微笑、風(fēng)度翩翩的CEO邱慈云(Tzu-Yin Chiu)。
在 日前接受EETimes美國(guó)版編輯獨(dú)家專訪過程中，自2011年8月執(zhí)掌中芯以來首度面對(duì)西方媒體的邱慈云謹(jǐn)慎用詞，分享了該公司的短、中、長(zhǎng)期發(fā)展方 向，暢談中芯在次40nm節(jié)點(diǎn)制程的策略，以及他個(gè)人對(duì)FDSOI (fully depleted silicon on insulator)技術(shù)的看法，也提及與IBM的合作關(guān)系，還有中芯為擴(kuò)充北京廠產(chǎn)能的募資。而從邱慈云聚焦于“產(chǎn)能利用率、差異化與先進(jìn)制程”的公司策略，也能看出中芯確實(shí)已經(jīng)改頭換面。
不談與全球晶圓代工龍頭臺(tái)積電(TSMC) 將如何競(jìng)爭(zhēng)，低調(diào)的邱慈云著墨更多的是如何執(zhí)行公司策略；在他的腦海中，對(duì)中芯來說最迫切的議題是向世界展示該公司能“持續(xù)在晶圓代工市場(chǎng)成長(zhǎng)茁壯”。從 中芯最近這幾季的財(cái)報(bào)結(jié)果看來，該公司的新策略執(zhí)行成效良好，問題是這樣的好表現(xiàn)能否繼續(xù)下去，中芯能維持目前的逐季成長(zhǎng)情勢(shì)多久？
中芯國(guó)際業(yè)績(jī)呈現(xiàn)逐季成長(zhǎng)趨勢(shì)
pGVesmc
本文授權(quán)編譯自EE Times，版權(quán)所有，謝絕轉(zhuǎn)載
本文下一頁(yè)：膽子夠大才能碰晶圓代工生意
相關(guān)閱讀：
• 武漢新芯化身XMC，與中芯國(guó)際再無瓜葛
• 中芯國(guó)際首席執(zhí)行官邱慈云博士出任燦芯半導(dǎo)體董事長(zhǎng)
• 發(fā)展遲滯，中國(guó)芯片制造、封裝業(yè)急待轉(zhuǎn)型升級(jí)pGVesmc
{pagination}
業(yè)界咸認(rèn)，得膽子夠大才能碰晶圓代工生意；半導(dǎo)體市場(chǎng)總是面臨周期性的生產(chǎn)短缺與過剩，偶爾還得因應(yīng)劇烈的終端市場(chǎng)需求波動(dòng)，同時(shí)所有人都得為了下一代制程節(jié)點(diǎn)技術(shù)拼命。
態(tài)度謹(jǐn)慎的邱慈云在受訪過程中很少偏離中芯國(guó)際議題，但他自豪地引述中國(guó)工信部(Ministry of Industry and Information Technology's)的軟件與IC推廣中心(CSIP) 2012年11月統(tǒng)計(jì)數(shù)據(jù)指出，有75%的中國(guó)無晶圓廠IC業(yè)者將中芯列為最優(yōu)先代工伙伴，而在一年前，該比例只有59%：“我們的客戶都認(rèn)同我們的服務(wù)品質(zhì)，特別是交貨速度?！?
交貨速度對(duì)那些特別講究產(chǎn)品交貨期(turn-around time)的客戶來說特別重要，邱慈云表示：“我們需要提供這類客戶完整的IP組合與產(chǎn)品特征描述，而且必須一次到位?！彼赋?，中芯在這些日子以來，某 些產(chǎn)品的交貨期可縮短到4個(gè)月，一般其它產(chǎn)品則是6~7個(gè)月。
而中芯國(guó)際今日的成功優(yōu)勢(shì)之一，是中國(guó)本地迅速崛起的無晶圓廠IC業(yè)者；如展訊(Spreadtrum)就是中芯最大、也最重要的客戶之一。
展訊目前的代工伙伴包括臺(tái)積電、中芯與聯(lián)電(UMC)，該公司CEO李力游透露，一旦展訊在今年底進(jìn)入28nm制程，將會(huì)選擇臺(tái)積電為主要代工廠：”但當(dāng)然，中芯也會(huì)進(jìn)入28nm節(jié)點(diǎn)，我們也將與他們合作，如果不是明年，就是在2015年?！?
雖然中芯最近的產(chǎn)能利用率稍有下滑，但邱慈云表示對(duì)目前情況感到滿意；根據(jù)中芯提供的統(tǒng)計(jì)資料，該公司 2012年第四季的產(chǎn)能利用率為91%，但在 2012年第二季曾達(dá)到過95%。接下來幾年，中芯最重要的策略則是差異化；“我們要為成熟制程技術(shù)增添附加能力?！鼻翊仍浦赋?，中芯已定義出九個(gè)目標(biāo)市場(chǎng)領(lǐng)域執(zhí)行差異化策略，基于競(jìng)爭(zhēng)性，他不愿透露完整的列表，只表示包括智能卡、CMOS影像傳感器與電源管理IC。
本文授權(quán)編譯自EE Times，版權(quán)所有，謝絕轉(zhuǎn)載
本文下一頁(yè)：先進(jìn)制程的挑戰(zhàn)
相關(guān)閱讀：
• 武漢新芯化身XMC，與中芯國(guó)際再無瓜葛
• 中芯國(guó)際首席執(zhí)行官邱慈云博士出任燦芯半導(dǎo)體董事長(zhǎng)
• 發(fā)展遲滯，中國(guó)芯片制造、封裝業(yè)急待轉(zhuǎn)型升級(jí)pGVesmc
{pagination}
先進(jìn)制程的挑戰(zhàn)
或許對(duì)中芯來說，最艱難的挑戰(zhàn)──也是一家晶圓廠最重要的長(zhǎng)期生存關(guān)鍵──就是先進(jìn)制程技術(shù)。為了趕上臺(tái)積電、GlobalFoundries等競(jìng)爭(zhēng)對(duì)手，中芯必須在先進(jìn)制程技術(shù)上大舉投資，但該公司也需要謹(jǐn)慎行事，密切觀察客戶的需求與對(duì)手的動(dòng)態(tài)。
在 28nm節(jié)點(diǎn)技術(shù)部分，是由中芯的上海研發(fā)團(tuán)隊(duì)進(jìn)行研發(fā)，未來量產(chǎn)則會(huì)在北京廠；邱慈云表示：“我們將開始接受在2013年底開始投片(tapeout) 的客戶訂單?！彼赋觯行镜?8nm會(huì)同時(shí)采用High-K/金屬閘極(HKMG)，以及poly/SiON兩種技術(shù)；而20nm制程也會(huì)同時(shí)采用平面 晶體管(planar)與FinFET兩種技術(shù)：“我們不能(因?yàn)橹蛔鲆环N技術(shù)而)流失客戶?！?
中芯國(guó)際執(zhí)行長(zhǎng)邱慈云
pGVesmc
在被問到中芯對(duì)FDSOI制程的看法時(shí)，邱慈云則表示：“以純物理學(xué)的觀點(diǎn)，我認(rèn)為FDSOI擁有很大的潛力，可提供具競(jìng)爭(zhēng)力的組件架構(gòu)?！钡仓赋觯拖裆钪械暮芏嗍挛铮骸耙苍S這并不是一個(gè)技術(shù)議題，而更偏向是一個(gè)市場(chǎng)行銷議題，端看FDSOI被多少客戶接受，以及該技術(shù)是否夠支撐起做為下一代主流制程技術(shù)。”邱慈云強(qiáng)調(diào)，除非有來自客戶對(duì)FDSOI的明確需求，中芯不會(huì)貿(mào)然承諾提供該技術(shù)。
邱慈云并坦承，中芯已經(jīng)從IBM取得一些重要的技術(shù)授權(quán)，但是在28nmHKMG 制程方面：“并不是以IBM的授權(quán)技術(shù)為基礎(chǔ)，我們是與IBM共同研發(fā)，使用一些我們自有的技術(shù)?！?
本文授權(quán)編譯自EE Times，版權(quán)所有，謝絕轉(zhuǎn)載
本文下一頁(yè)：持續(xù)募資擴(kuò)產(chǎn)
相關(guān)閱讀：
• 武漢新芯化身XMC，與中芯國(guó)際再無瓜葛
• 中芯國(guó)際首席執(zhí)行官邱慈云博士出任燦芯半導(dǎo)體董事長(zhǎng)
• 發(fā)展遲滯，中國(guó)芯片制造、封裝業(yè)急待轉(zhuǎn)型升級(jí)pGVesmc
{pagination}
持續(xù)募資擴(kuò)產(chǎn)
在投資先進(jìn)制程技術(shù)之外，晶圓代工業(yè)者也持續(xù)面臨著為擴(kuò)充產(chǎn)能而繼續(xù)投入資本支出的壓力。中芯在2000年成立，獲得中國(guó)政府不少資金支持；當(dāng)時(shí)的中芯開創(chuàng) 了一個(gè)全新的商業(yè)模式，也就是與中國(guó)各地方政府洽談晶圓廠興建事宜，再交由該公司管理。但這種商業(yè)模式雖讓中芯得以節(jié)省成本，卻無法維持長(zhǎng)久。
剛將識(shí)別標(biāo)志改成“XMC”的武漢新芯(Wuhan Xinxin Semiconductor Manufacturing)，一開始也是由中芯負(fù)責(zé)營(yíng)運(yùn)，但所有權(quán)是武漢地方政府；XMC的管理高層最近接受訪問時(shí)表示，該公司已經(jīng)是獨(dú)立的晶圓代工業(yè)新秀，并非中芯的姊妹公司(可參考相關(guān)閱讀)。
那么目前中芯的北京廠募資情況如何？邱慈云表示，該公司北京廠已經(jīng)從投資界找到增資金援：“中芯國(guó)際需要獲利，重振投資人的信心?！彼赋?，中芯打算透過北京市政府所擁有的投資基金，讓北京廠成為一座合資晶圓廠，但相關(guān)事宜還在討論階段。
中芯是否也將取得中國(guó)中央政府的資金支持？對(duì)此邱慈云表示：“目前沒有，我們只與地方政府層級(jí)接觸?！钡矎?qiáng)調(diào)，中國(guó)對(duì)半導(dǎo)體產(chǎn)業(yè)的態(tài)度十分友善，也對(duì)全球業(yè)者的投資抱持開放態(tài)度，包括Hynix、Samsung都曾在中國(guó)興建大型晶圓廠。
本文授權(quán)編譯自EE Times，版權(quán)所有，謝絕轉(zhuǎn)載
編譯：Judith Cheng
參考英文原文：Exclusive: China chip CEO details SMIC's foundry plans，by Junko Yoshida
相關(guān)閱讀：
• 武漢新芯化身XMC，與中芯國(guó)際再無瓜葛
• 中芯國(guó)際首席執(zhí)行官邱慈云博士出任燦芯半導(dǎo)體董事長(zhǎng)
• 發(fā)展遲滯，中國(guó)芯片制造、封裝業(yè)急待轉(zhuǎn)型升級(jí)pGVesmc
{pagination}
Exclusive: China chip CEO details SMIC's foundry plans
Junko Yoshida
SMIC CEO Tzu-Yin Chiu hopes to put to bed the industry’s image of the Chinese foundry as hip-deep in red ink and fighting TSMC in court.
SHANGHAI – Semiconductor Manufacturing International Corp. (SMIC), for the first time in a long time, has a good story to tell.
The company hopes to put to bed the industry’s collective memory of SMIC as a huge Chinese foundry hip-deep in red ink, pursued by its Taiwanese rival in litigation, and paralyzed by internal management turmoil.
In contrast, the new SMIC comes with record-high revenue—$1.7 billion for the full year of 2012—positive income of $15.9 million (the company’s highest in seven years), and a smiling, personable CEO, Tzu-Yin Chiu, who took the company’s helm in August, 2011.
In the first one-on-one interview given to the Western media since he became SMIC’s CEO, Chiu carefully chose his words as he laid out his company’s short, medium and long-term strategies Thursday (March 21). In an exclusive conversation with EE Times at SMIC headquarters here, he touched on everything from the company’s sub 40-nm strategy to his view on fully depleted silicon on insulator (FDSOI) technology. He also covered SMIC’s relationship with IBM and financing for its Beijing fab expansion.
The most telling sign of the new SMIC is found in Chiu’s focus on “utilization, differentiation and advanced technology”– with priority in that order, not the reverse.
SMIC's CEO Tzu-Yin Chiu
Instead of speaking with bravado of how he plans to compete with the world’s biggest foundry—Taiwan Semiconductor Manufacturing Co. (TSMC)—the low-key CEO spoke more about execution. In Chiu’s mind, the pressing issue for SMIC is to demonstrate to the world that SMIC can “continue to grow and thrive in a foundry environment,” as he put it.
If the last several quarters are any indications, the new plan is going well. But the obvious question is whether the good news keeps coming. How long can SMIC sustain its current quarter-after-quarter growth?
As the chip industry knows, the foundry business is not for the faint-hearted. The semiconductor market regularly revolves in cycles of shortage and overcapacity, with sometimes dramatic fluctuations in end-market demands, while everyone sweats the future of next-node process technology.
Cautiously, Chiu’s interview rarely strayed from SMIC’s message. But he proudly cited a survey published by the Ministry of Industry and Information Technology's China Software and Integrated Circuit Promotion Center (CSIP) in November 2012, which noted that 75 percent of Chinese fabless companies chose SMIC as their preferred foundry partner. That number in the same survey a year before was only 59 percent.
“Our customers are recognizing the quality of our service, and especially speed,” said Chiu.
Speed is critical for customers whose survival depends on product turn-around time. For that, “we need to offer them a reasonably complete IP portfolio and characterization,” he said. “And we must get it right the first time.” For some products, SMIC’s turn-around time these days is as short as four months. For other products, it’s six to seven months.
One SMIC advantage today is the big success of some emerging China fabless companies. As Spreadtrum’s CEO Leo Li noted and SMIC’s Chiu confirmed, Spreadtrum, for example, is one of the “the biggest” and “most important customers” of SMIC. But, once Spreadtrum, which currently gets its chips made at TSMC, SMIC and UMC, moves to a 28-nm node for the company’s SoCs in the fourth quarter of this year, Li said that Spreadtrum is going to TSMC. “But of course, SMIC is also moving to 28-nm. We’ll work with them, if not next year, then in 2015,” said Li.
Fab utilization
Although SMIC saw “a slight drop” its fab utilization ratio, Chiu said that he’s “happy about the loading situation.” According to company data, SMIC’s fourth-quarter utilization rate in 2012 was 91 percent, after reaching 95 percent in the second quarter.
Over the next few years, the most important for SMIC will be the company’s differentiation strategy. “We bring our mature technology to add additional capabilities,” said Chiu.
SMIC’s team reportedly identified “nine market segments,” according to Chiu, where the company thinks it can win through differentiation. He declined to share the entire list, citing competitive reasons. But it includes smart cards, CMOS image sensors and power management ICs.
One of the hottest areas Chiu sees is the Chinese government push for smart cards. This covers SIM cards, social security and health-care, transportation and bank cards. Potential volume aside, cards need to be designed differently–specific to each application, said Chiu. This allows SMIC to work closely with local card vendors and provide differentiation. SMIC’s specialty includes “l(fā)ow power dissipation” especially for smart cards in contactless applications, Chiu said.
Chiu also sees SMIC as a key CMOS image-sensor foundry. SMIC has identified a burgeoning demand for cameras in smartphones, feature phones, home security and surveillance.
SMIC reportedly scored a breakthrough in the development of backside-illuminated (BSI) CMOS image-sensor technology in the last quarter. Chiu noted, during the most recent quarterly financial results call: “The first fab chip demonstrated good image quality. The complete BSI process technology is targeted to contribute revenue in 2014.”
BSI delivers higher resolution to mobile phone cameras and high performance video cameras. This, in turn, allows SMIC customers to address the higher-end market. The differences SMIC can offer in this application include “l(fā)ow leakage capability,” according to Chiu.
Sub 40-nm processes
Perhaps the toughest challenge facing SMIC--and most essential to the foundry’s long-term survival--is the issue of advanced technology.
To keep pace with rivals like TSMC and GlobalFoundries, SMIC must make substantial investments in advanced technology. Yet, the company needs to tread carefully, watching closely both customer demands and the competition’s next steps.
As for 28-nm node, SMIC’s Shanghai facility has been working on its R&D, while high-volume production will take place in SMIC’s Beijing fab. “We will start accepting customers’ tapeout by the end of 2013,” Chiu noted.
He noted that SMIC’s 28-nm offering will be both high-K, metal gate (HKMG) and poly/SiON. Similarly, SMIC’s 20-nm chips will be both planar and FinFET, Chiu said. “We can’t lose our customers” by not doing one or the other, he explained.
When asked about SMIC’s interest in FDSOI, Chiu said, “From a pure physics’ point of view, I think FDSOI has very high potential. I think it gives a competitive device structure.”
But he added, as with many things in life, “Perhaps, this is not the issue of technology but more a matter marketing. It all depends on how many customers will accept FDSOI, and whether FDSOI can build enough of a following,” he said. “Unless there is a clear indication from our customers that FDSOI is what they want,” SMIC is not likely to commit to it.
Chiu acknowledged that SMIC has licensed significant technologies from IBM. But as for the 28-nm HKMG development, Chiu noted, “This is not based on a licensing program with IBM. We are jointly developing it with IBM, using some of our own technology.”
Financing strategy
Beyond investments in advanced technology, foundries are constantly under pressure to keep up capital expenditures for capacity expansions.
SMIC was founded in 2000, with generous financial backing from the Chinese government. SMIC then pioneered a new business model under which the foundry talks municipalities into building a capital intensive fab while SMIC agrees to manage it.
The model has helped lower SMIC’s capital costs--most of the time but not always. Wuhan Xinxin Semiconductor Manufacturing, now called XMC, was orignially operated by SMIC but owned by a local government, in this case, Wuhan.
Now that the Wuhan foundry is called XMC, the new management team at XMC earlier this week declared that the Wuhan-based fab is “an independent foundry startup--not SMIC’s sister company.” The fab’s ownership is still in the hands of the city of Wuhan. SMIC never bought it back, and it’s now being run by Simon Yang, SMIC’s former COO who resigned the company after Chiu became SMIC’s CEO.
How has SMIC’s Beijing fab been financed?
SMIC, by going to the investment community for its Beijing fab, is raising funds internally. “SMIC needs to execute profitability and regain confidence from investors,” said Chiu.
SMIC's rising revenue by quarter
source: SMIC
SMIC is planning to make it a joint-venture fab, “by leveraging investment funds owned by the Beijing municipal government,” according to Chiu.
The investments for SMIC’s Beijing expansion, however, are still under discussion.
Any funding or help from the Chinese central government? “No, so far, we’ve only gotten to the local level,” said Chiu. But Chiu hastened to point out that China has been very friendly, and open to the global semiconductor industry. Companies such as Hynix, Samsung and Intel have been all benefited, building mega-fabs in China, he added.
責(zé)編：Quentin
