富士通与松下合作，还有6点意见没能统一

富士通半导体(Fujitsu Semiconductor)、松下(Panasonic)连手与日本开发银行(Development Bank of Japan Inc. ，DBJ)在近日签订了一项最终协议，将成立一家新的无晶圆厂系统芯片(system LSI)设计公司；此举意味着日本电子产业已经发生了翻天覆地的变化。 上述合作案代表了一个时代的终结。数十年来，日本电子业者一直沉迷于拥有自己的晶圆厂，日本芯片供货商也将开发系统芯片视为未来发展的关键，以摆脱仅此一招 的传统内存业务、朝产品多元化方向迈进；但现在，半导体制造业务已非神圣不可侵犯，富士通与松下在最近这几年都陆续将旗下的晶圆厂脱手，甚至将藉由成立合资新公司摆脱系统芯片业务。 根据富士通与松下所公布的消息，即将成立的无晶圆厂新公司首席执行官为西口泰夫(Yasuo Nishiguchi)，持股比例分别为富士通40%、松下20%、日本开发银行40%；新公司员工数约2,800人(其中700人来自松下的系统芯片事 业部门)，预计在2014财务年度第四季开始营运。 但官方新闻稿似乎还是遗漏了许多重要消息，留下来的疑问包括新公司的晶圆代工伙伴是否包括英特尔(Intel)在内、特别是其14纳米制程？此外富士通与松下也未透露新公司将会开发那些汽车应用相关IC？还有Unifier平 台──松下针对数字消费类电子应用所开发的ARM核心SoC──是否会成为新公司的系统芯片基础架构？ 以下是六个EETimes美编辑在日本向富士通与松下发言人所提出的问题，以及他们的响应： 问题一：新公司是否会履行松下与英特尔之间的14纳米制程代工协议？ 对此一位富士通发言人表示，他对相关细节不清楚；松下发言人则表示：“将松下系统级芯片(System LSI，SLSI)部门与英特尔之间所达成的协议转移至新公司，是合乎逻辑的思考。” 在 稍早之前，松下SLSI事业部门总监冈本(Yoshifumi Okamoto)在一份声明中指出：“我们将透过合作、利用英特尔的14纳米三闸晶体管(Tri-Gate)制程技术，为下一代的SoC带来提升的性能与 功耗优势。”当时冈本并未提到松下与富士通将成立的新公司，而事实上松下在其中的持股比例也只有20%。 本文授权编译自EE Times，版权所有，谢绝转载 第2页：新成立的公司将开发哪些芯片？ 第3页：两家会各自保留哪些芯片业务？新公司如何获利？ {pagination} 问题二：松下的Unifier将成为新系统芯片设计公司的产品之一吗？ Unifier是松下的工程团队所开发的一个公用平台，能提供电视、DVD播放器、手机等各种不同的数字消费性产品应用，大幅提升SoC软件的重复使用率；该平台包含系统芯片硬件(CPU核心、视频编译码器)与软件(操作系统、软件)。 而针对新公司产品是否包括Unifier平台的问题，松下发言人谨慎表示：“还有待观察，一切细节需要在新公司内部进行讨论。” 问题三：新成立的公司将开发哪些类别的系统芯片？ 在 今年稍早松下、富士通的一份联合声明中，两家公司透露了一系列将锁定的技术领域包括：1. 高性能解决方案(包括支持云端基础架构如超高速网络的性能服务器以及核心技术)；2. 虚拟与影像解决方案(新一代数位电视以及影像识别应用)；3. 无线解决方案(支持无所不在网络之行动与超低功耗无线连结解决方案)。 不过身为母公司之一的松下现正试图摆脱传统的消费性电子业务，如数字电视与手机，因此新公司究竟将会开发那些数字消费性系统芯片，还未真正明朗化。 问题四：新公司是否会开发车用SoC？ 针 对此一问题，富士通发言人肯定的表示，汽车仪表板与显示器使用的系统芯片会是新公司的开发项目之一；他指出：“车用ASSP是富士通半导体的强项。”而松下则并不打算将汽车相关组件业务转移至新公司，该公司发言人表示，将保留旗下的汽车与工业系统(Automotive and Industrial Systems，AIS)事业，芯片设计团队也不会转往新公司。 松下汽车相关组件包括影像传感器、车用LED，以及功率半导体、高性能大功率组件等，都将保留在母公司；而松下系统芯片事业群的大部分都会转移至新公司，该部门主要是开发消费性产品应用SoC，其中四成业务为电视、25%为蓝光产品、手机与调谐器各占10%。 本文授权编译自EE Times，版权所有，谢绝转载 第3页：两家会各自保留哪些芯片业务？新公司如何获利？ {pagination} 问题五：新公司将如何获利？ 在 官方新闻稿中，新公司特别强调拥有“世界级管理资源”，并将专注于获利，扩展业务至全球市场，争取在数年之内让公司股票公开上市(IPO)。新公司的首席执行官西口泰夫并非来自富士通或松下，他先前曾担任京瓷(Kyocera)首席执行官，后来则在一家私募股权公司──颇负盛名但也受争议的凯雷集团 (Carlyle Group)──担任主管，负责搜寻日本当地的收购目标。 根据富士通半导体与松下相关业务目前的营收估计，合资新公司年营收可达到1,500亿日圆；新公司不会有自己的晶圆厂，将以无晶圆厂模式经营，但这并不能保证其获利能力，也不能确定其客户群能超越两家母公司的现有客户。 在全球众多芯片供货商都积极争取汽车电子领域不断成长的商机的此刻，这家新公司的汽车芯片也不一定能在强敌环伺中保证获利；松下就不打算将汽车芯片业务转移至新公司。而据了解，在两年前成立合资新公司的计划首度浮现时，富士通原本想邀瑞萨(Renesas)加入新公司，但瑞萨却决定走自己的路、维持在全球车用半导体市场的领导地位。 问题六：富士通半导体与松下将会各自留下哪些芯片业务？ 富士通半导体将保留的是内存业务，主要是FRAM与FCRAM；该公司表示，FRAM拥有非挥发特性，且高速写入性能优于SRAM。富士通自1999年就推出FRAM，目前有两条产品线，分别是串行接口(I2C、SPI)以及并列接口产品。 富 士通的FCRAM (Fast Cycle RAM，快速周期随机存取内存)则以低功耗特性为要求，拥有低功耗DDR SDRAM接口或PSRAM (pseudo SRAM)接口两种产品。因为FCRAM具备低功耗与高速数据传输功能，富士通一直认为该技术是数字消费性电子产品的理想内存，如数字电视、数字摄影机 等需要大量数字数据高速处理的装置，此外也适合手机等移动设备应用。 除了将在新合资公司开发的系统级芯片，还有已经出售给Spansion的微控制器(MCU)与模拟组件业务，富士通半导体表示该公司未来只会专注在内存业务。而松下将保留的半导体组件业务，则是锁定汽车与工业应用领域。 本文授权编译自EE Times，版权所有，谢绝转载 编译：Judith Cheng 参考英文原文：Fujitsu-Panasonic Co.: 6 Things Left Unsaid，by Junko Yoshida {pagination} New Fujitsu-Panasonic Co.: 6 Things Left Unsaid Junko Yoshida Will Intel 14-nm process be a part of the picture? TOKYO — Fujitsu Semiconductor and Panasonic, together with Development Bank of Japan Inc. (DBJ), marked a sea change in the Japanese electronics industry last week by signing a definitive agreement to establish a new fabless-type company to design and develop system LSI. The move confirms the end of an era. For decades, Japanese electronics companies were obsessed with owning their own fabs. Japanese chip vendors had also regarded developing system LSIs as the key to their future, as they wanted to diversify products from their traditional one-trick memory business. The semiconductor production business is no longer a sacred cow either. Fujitsu and Panasonic have jettisoned most of their own fabs over the last few years. Further, both companies are willing to let go of their system LSI businesses, as they merge into one as a new company. The announcement last week revealed a few details of the new fabless chip company -- including its CEO (Yasuo Nishiguchi), its ratio of voting rights (Fujitsu 40%, Panasonic 20%, DBJ 40%) and the number of employees (approximately 2,800 -- 700 of which are from Panasonic's system LSI business). The new company is scheduled to start operating in the fourth quarter of fiscal 2014. What's left unsaid More revealing, however, was what the companies did not say in their joint press release. Left unanswered in the announcement is whether the new company's foundries would include Intel, especially at its 14-nm process. There is no indication of how many automotive-related ICs the new company will develop, or whether Unifier -- Panasonic's ARM-based SoC platform for digital consumer products -- will become a major part of the new company's system LSI architecture. We asked those questions to spokespeople at Fujitsu and Panasonic in Japan, and here's what we learned. Question No. 1: Will the new company honor the manufacturing agreement Panasonic made with Intel for the use of Intel's 14nm process? A Fujitsu Semiconductor spokesman said he's unclear on the issue. A Panasonic spokesman, in contrast, told EE Times, "It is logical to think that the deal reached between Panasonic's System LSI (SLSI) business division and Intel will be carried over to the new company." Yoshifumi Okamoto, director of Panasonic's SLSI Business Division, said in an early announcement: "We will deliver highly improved performance and power advantages with next-generation SoCs by leveraging Intel's 14nm Tri-Gate process technology through our collaboration." At that time, Okamoto made no mention of the emerging Fujitsu-Panasonic system LSI company, in which Panasonic, it turns out, has only 20% voting rights. Question No. 2: Is Panasonic's Unifier part of the new system LSI company's platform offerings? Rather than developing different software and hardware for TVs, DVD recorders, mobile phones, and other digital consumer products independently, Panasonic engineers designed a common platform called Unifier, so that they can dramatically increase the reuse of software on their SoCs. The Unifier platform consists of system LSI (including CPU cores and video codecs) and software (incorporating operating system and middleware). Whether the new Fujitsu-Panasonic system LSI company will include the Unifier platform "remains to be seen," the Panasonic spokesman cautiously noted. "Details need to be discussed within the new company." Question No. 3: What sort of "system LSIs" will the new company develop? The new company will develop system LSIs focused on "visual imaging" and "networking" solutions, according to Fujitsu and Panasonic. More specifically, in a joint press release earlier this year, the two companies listed "areas of focus" that include: 1) High-performance solutions (high-performance servers and core technologies that support cloud infrastructure such as ultra-high-speed networks), 2) visual and imaging solutions (next-generation DTV, applications for image recognition, etc.), and 3) wireless solutions (mobile and extremely low-power wireless connectivity solutions that support ubiquitous networks). However, at a time when Panasonic -- the parent company -- is trying to get out of the traditional consumer electronics business, such as DTV and mobile, it is unclear how much of digital consumer SoCs will becom a part of the new company's agenda. Question No. 4: Will the new company also develop automotive SoCs? The Fujitsu spokesman made it clear that system LSIs for automotive dashboards and displays are definitely a part of the plan. "Fujitsu Semiconductor has a strong ASSP business around automotive," he added. In contrast, Panasonic will not assign its automotive-related device businesses to the new company. Panasonic's Automotive and Industrial Systems (AIS) Company will stay with Panasonic. Automotive chip designers working for AIS, won't be transferred to the new company, according to the spokesman. Automotive-related devices such as image sensors, LEDs for vehicles, power semiconductors and high-performance high-voltage devices will remain within Panasonic. A majority of Panasonic's System LSI group will be transferred to the new company. The group had been specialized on the development of SoCs for consumer products, with 40% of its business in TV, 25% in Blu-ray, 10% for mobile and 10% on tuner ICs. Question No. 5: What will make the new system LSI company profitable? In the latest press release, the new company painstakingly stressed that it will have "world-class management resources" and will consolidate "in a profitable manner." The new company will "expand its business globally" and "aim for an IPO within several years." Good news is that the incoming CEO Yasuo Nishiguchi does not come from either Fujitsu or Panasonic. Formerly CEO of Kyocera Corp., he served most recently as an operating executive at the private equity firm, the prestigious but controversial Carlyle Group. Nishiguchi's responsibility at the Carlyle Group was the focus on Japanese buyout opportunities. The new company forecasts annual sales to be about 150 billion yen, based on the combined results of Fujitsu Semiconductor and Panasonic's relevant businesses in the latest fiscal year. The new company won't own any fabs of its own. The new company's fabless model, however, by no means guarantees its profitability. Nor will it assure broader design wins beyond the existing customers held by Fujitsu and Panasonic. While many chip vendors throughout the world are committed to the growing electronics content inside vehicles, the new company's automotive chip business won't necessarily profit from any consolidations among Japanese chip companies. After all, Panasonic is pursuing automotive chip business independent of the new Fujitsu-Panasonic company. Reportedly, Fujitsu originally wanted to bring Renesas into the new company, when the idea of the JV was first floated more than two years ago. Renesas, however, has decided to go it alone, while the company plans to maintain its leadership position in the global automotive electronics market. Question No. 6: What chip business, if any, will be left at Fujitsu Semiconductor and Panasonic? For Fujitsu Semiconductor, what's left will be memory business -- mainly focused on FRAM and FCRAM. According to Fujitsu, FRAM (ferroelectric memory) has advantages of non-volatile memory like EEPROM, keeping write data without power supply for data retention, and high-speed write performance of SRAM. FRAM, which Fujitsu has been shipping since 1999, has 2 product families; serial interface (I2C, SPI) and parallel interface product. Fujitsu's FCRAM (Fast Cycle RAM) is a random access memory featuring low-power consumption. It has a low-power DDR SDRAM interface or PSRAM (pseudo SRAM) interface in the product family. Because the FCRAM features both low-power consumption and high-speed data transfer capability, Fujitsu has long maintained that it's ideal for digital consumer electronics such as digital television and camcorder, that require high-speed processing of large-scale digital data, and optimal for mobile applications such as cellular phones. Now that those system-level SoCs will be designed at the new Fujitsu-Panasonic company and the company's original MCU and analog businesses were sold to Spansion, Fujitsu Semiconductor's main focus will be on the standalone memory business, said the company. For Panasonic, the company's device business will be all about chips for automotive and industrial applications.