英特尔（最近一次）向移动回归

    对于奥特里尼来说，努力发展无线业务不仅仅是为了使收入多样化，也不仅仅是为了要让公司进入一个正在成长的市场（当然这也是其中的一部分原因），而是为了救赎。2005年成为CEO后，奥特里尼一直固执地坚持英特尔可以基于公司历史上著名的x86架构开发自己的芯片，而不是像大多数竞争对手一样使用英国ARM公司授权的芯片技术。但直到现在，他的赌注还没有见到成效。英特尔曾宣称它的Moorestown芯片将在2011年初植入智能手机，但现在这种芯片主要只是用于机器人技术和上网本。

    讽刺的是，英特尔以前也曾为早期的智能手机和Palm Treo等PDA设备生产ARM处理器。但2006年，奥特里尼却把整条移动产品生产线（称做Xscale）以6亿美元的价格卖给了迈威科技（Marvell Technology Group），从事实上切断了英特尔进军移动设备领域的通道。英特尔在移动领域的失策之举还不止这一件。2010年，奥特里尼决定与诺基亚（Nokia）合作，开发基于Linux的MeeGo操作系统。不过一年后，诺基亚就抛弃了这个项目，转而选择了微软（Microsoft）的Windows Phone。而英特尔也改弦更张，在九月份选择了一个新的合作伙伴——谷歌（Google）。英特尔的软件开发者们付出了大量努力，以确保谷歌的安卓（Android）系统可以在英特尔芯片上完美运行。

    与世界上增长最快的手机操作系统合作，这的确是奥特里尼一个明智之举（奥特里尼从2004年起就是谷歌董事会的成员），但英特尔的欠帐显然还有很多。高宏乐通集团（Cowen Group）分析师拉杰•希思指出：“英特尔让人感觉他们承诺的太多，做到的太少。”

    与此同时，行业内的其他公司也在不断进步。目前，市场上90%的智能手机使用的是依托于ARM架构的微处理器。其中高通公司（Qualcomm）独自揽下了51%的市场份额。

    奥特里尼坚称，MeeGo系统的失败只让英特尔耽误了两个月的时间而已，而在芯片生产技术方面，英特尔要领先竞争对手大概三年。

    英特尔的9个半导体工厂加在一起，每秒钟可生产100亿个晶体管。英特尔最新、最大的加工工厂名叫D1X，座落于俄勒冈州的希斯波洛，位于波特兰以西约20英里处。D1X也是俄勒冈州有史以来最大的建设项目之一。这座先进的工厂将于明年开工，届时它也将成为全球第一家生产14纳米芯片的工厂。该厂投资将达到50亿美元以上。

    根据摩尔定理，一个芯片的晶体管数量大约每24个月就会翻一番。这说明芯片生产是一项耗资巨大的业务，需要英特尔不断投资新的生产技术，建设新的生产设施。但是除了PC领域之外，英特尔的生产技术优势能帮上它多少忙，现在还不清楚。比如，说起先进程度，那些生产ARM处理器的厂商所使用的生产技术并不如英特尔Medfield芯片的生产技术，但这并不妨碍他们从英特尔的碗里抢饭吃。投资银行Piper Jaffray的分析师古斯•理查德指出：“英特尔试图把过去在PC领域给它带来过成功的那套工具原封不动地搬到移动业务上，但是这并不起作用。”

    For Otellini, getting it right in wireless isn't just about diversifying revenue or driving the company into growing markets (though that's part of it). Otellini's mobile gambit is about redemption. Since becoming CEO in 2005, Otellini has stubbornly insisted that Intel could develop its own chips based on x86, the company's historic microprocessor design standard, instead of licensing technology from the British company ARM Holdings, as most of his competitors have. So far his bet hasn't paid off. The company proclaimed that its Moorestown system-on-a-chip would be in smartphones in early 2011, but instead it is mostly used in robotics and netbooks.

    Ironically, Intel used to manufacture ARM processors for early smartphones and PDAs like the Palm Treo. But in 2006, Otellini sold its entire mobile product line, called XScale, to Marvell Technology Group for $600 million, essentially cutting off its access to mobile devices. And that wasn't Intel's only questionable move in mobile. In 2010, Otellini decided to partner with Nokia to develop MeeGo, a Linux-based operating system. A year later Nokia (NOK) ditched the project and opted to adopt Microsoft's (MSFT) Windows Phone instead. Intel changed course and in September picked a new partner: Google (GOOG). Intel software developers worked to make sure that Android could run on the company's chips without a glitch.

    Pairing up with the world's fastest-growing operating system is a smart move for Otellini (who has been on Google's board since 2004). But there's little question Intel has a lot of catching up to do. "The perception is that they've overpromised and underdelivered," says Raj Seth, an analyst with Cowen Group.

    Meanwhile the rest of the industry has moved ahead. Processors designed using ARM Holdings' technology now power 90% of smartphones. Qualcomm (QCOM) leads the pack with 51% market share (see chart below).

    Otellini insists the MeeGo debacle put Intel behind schedule by just two months. And when it comes to chip-manufacturing technology, he contends that Intel is about three years ahead of the competition.

    Intel's nine semiconductor factories churn out a collective 10 billion transistors per second. Its latest and greatest fabrication plant, called D1X, is in Hillsboro, Ore., about 20 miles west of Portland. D1X is one of the biggest construction projects in Oregon's history. When the state-of-the-art facility opens next year, it will be the first 14-nanometer factory in the world. And it will have cost Intel upwards of $5 billion.

    Keeping up with Moore's law, which states that the number of transistors incorporated in a chip will approximately double every 24 months, is an expensive business that requires Intel to constantly invest in new manufacturing technologies and build new plants. But beyond the PC, it's not clear how much that manufacturing lead is helping. The biggest facilities making ARM-designed processors today are based on a less advanced process than the one Intel uses to make its Medfield chip, for example, yet that hasn't stopped the competition from eating Intel's lunch. "Intel has tried to bring the same tools that have made it successful in the past to mobile, and it's not working," says Piper Jaffray analyst Gus Richard.