经济学人：扶不起的芯片梦

作者：经济学人        来源：一天一篇外刊

How low mighty Intel has fallen. Half a century ago the American chipmaker was a byword for the cutting edge; it went on to dominate the market for personal-computer chips and in 2000 briefly became the world’s second-most-valuable company. Yet these days Intel, with a market capitalisation of $100bn, is not even the 15th-most-valuable chip firm, and supplies practically none of the advanced chips used for artificial intelligence (AI). Once an icon of America’s technological and commercial prowess, it has lately been a target for subsidies and protection. As we published this, President Donald Trump was even mulling quasi-nationalisation.

英特尔辉煌不再。半个世纪前，英特尔曾是尖端科技的代名词，主导了个人电脑芯片市场，并在 2000 年短暂地成为全球市值第二高的公司。如今，英特尔的市值仅为 1000 亿美元，甚至未能跻身全球最有价值的前15芯片企业，而且几乎不供应任何人工智能（AI）所需的先进芯片。作为美国科技与商业实力的象征，英特尔如今成了美国补贴与保护的对象。在我们发稿之际，美国总统特朗普甚至还在考虑对英特尔进行某种形式的“准国有化”。

More than ever, semiconductors hold the key to the 21st century. They are increasingly critical for defence; in the ai race between America and China, they could spell the difference between victory and defeat. Even free-traders acknowledge their strategic importance, and worry about the world’s reliance for cutting-edge chips on tsmc and its home of Taiwan, Yet chips also pose a fiendish test for proponents of industrial policy. Their manufacture is a marvel of specialisation, complexity and globalisation. Under those conditions, intervening in markets is prone to fail—as Intel so vividly illustrates.

进入 21 世纪，半导体的重要性空前凸显。它们在国防中的重要性日益显现；在中美人工智能竞赛中，半导体甚至可能决定胜负。即便是自由贸易的拥护者也承认其战略意义，并担忧全球在尖端芯片上对台积电的依赖。与此同时，半导体也为产业政策的支持者带来了棘手的考验。芯片制造本身就是专业化、复杂性和全球化的奇迹。在这样的条件下，政府干预市场往往难以成功——而英特尔正是最鲜明的例证。

To see how much can go wrong, consider its woes. Hubris caused the firm to miss both the smartphone and the ai waves, losing out to firms such as Arm, Nvidia and tsmc. Joe Biden’s CHIPS Act, which aimed to spur domestic chipmaking, promised Intel $8bn in grants and up to $12bn in loans. But the company is floundering. A fab in Ohio meant to open this year is now expected to begin operations in the early 2030s. Intel is heavily indebted and generates barely enough cash to keep itself afloat.

要了解问题可能有多严重，不妨看看英特尔的困境。傲慢使这家公司错过了智能手机和人工智能的浪潮，市场份额拱手让给了 Arm、英伟达和台积电等企业。乔·拜登推动的《芯片法案》旨在刺激本土芯片制造，承诺向英特尔提供 80 亿美元的补助，以及最高 120 亿美元的贷款。但这家公司依然举步维艰。原本计划今年在俄亥俄州投产的一座晶圆厂，如今预计要到 2030 年代初才能投入运营。英特尔债务沉重，现金流几乎仅够维持生存。

The sums needed to rescue it keep growing. By one estimate Intel will need to invest more than $50bn in the next few years if it is to succeed at making leading-edge chips. Even if the government were to sink that much into the firm, it would have no guarantee of success. The company is said to be struggling with its latest manufacturing process. Its sales are falling and its plight risks becoming even more desperate.

能让英特尔走出困境所需的资金还在不断增加。根据一项估算，如果要在尖端芯片制造上取得成功，未来几年英特尔需要投入超过 500 亿美元。即便美国政府为这家公司砸下如此巨资，也无法保证一定成功。据称，英特尔在最新制程工艺上仍举步维艰。其销售额正在下滑，处境则有可能变得更加岌岌可危。

The Biden administration failed with Intel, but Mr Trump could make things worse. He has threatened tariffs on chip imports, and may try to browbeat firms such as Nvidia into using Intel to make semiconductors for them. These measures might buy Intel time but they would be self-defeating for America. Chipmaking is not an end in itself but a critical input America’s tech sector requires to be world-beating. Forcing firms to settle for anything less than the best would blunt their edge.

拜登在英特尔问题上已告失败，特朗普可能会让情况更糟。他曾威胁要对芯片进口加征关税，并试图施压像英伟达这样的公司，强迫它们使用英特尔来代工生产芯片。这些措施或许能为英特尔赢得一些时间，但对美国而言却是自毁长城。芯片制造并非最终目的，而是美国科技产业保持全球领先所必需的关键投入。若强迫企业接受不够先进的方案，只会削弱它们的竞争优势。

作者：经济学人        来源：一天一篇外刊