字号·· | 护眼
连线

美国AI芯片战正以最愚蠢的方式升温America’s AI Chip War Is Heating Up in the Dumbest Possible Ways

点「原文对照」整页切到原文,或双击某段只看那段的原文。

本周,特朗普政府让中国公司获取全球部分最先进 AI 芯片的难度大幅增加。多年来,阿里巴巴和字节跳动等科技公司可以从海外数据中心租用算力,从而使用强大的英伟达芯片,而这些芯片通常被禁止实体进口到中国。这被称为远程访问。

This week, the Trump administration made it significantly harder for Chinese companies to access some of the world’s most advanced AI chips. For years, tech firms like Alibaba and ByteDance could rent computing power from data centers overseas, allowing them to use powerful Nvidia chips they were generally prohibited from physically importing into China. This is known as remote access.

但根据周一宣布的新计划,想要向中国提供大量算力的海外数据中心将不得不申请特殊许可证,并应对其他繁琐的新规。白宫还对向包括新加坡和马来西亚在内的五个国家出口先进芯片实施了更严格的规定,美国官员怀疑这些国家可能发生了大量走私活动。

But under new plans announced on Monday, overseas data centers that want to provide large amounts of computing power to China will have to get special licenses and deal with other annoying new kinds of red tape. The White House also imposed stricter rules on shipments of advanced chips to five countries, including Singapore and Malaysia, where US officials suspect significant levels of smuggling may have taken place.

这些并非新的政策构想,类似提案在华盛顿酝酿已久。但周一宣布的时机无法被忽视。是的,这关乎黄仁勋的性生活。

These are not new policy ideas, and similar proposals have been kicking around in Washington for a while. But the timing of Monday’s announcement is impossible to ignore. Yes, this is about Jensen Huang’s sex life.

正如你可能已经听说的,英伟达联合创始人黄仁勋上周在承认与中国女商人 Alice Huang 有染后辞去 CEO 一职,后者被指与中国情报机构有联系。Alice 曾任职于新加坡数据中心公司 Megaspeed,该公司提供远程访问服务,并被怀疑从事芯片走私。

As you’ve probably heard by now, the Nvidia cofounder resigned as CEO last week after admitting he had an affair with Alice Huang, a Chinese businesswoman accused of having ties to Chinese intelligence services. Alice previously worked at the Singapore-based data center company Megaspeed, which offered remote access services and is suspected of engaging in chip smuggling.

这场丑闻可能让白宫感到极为尴尬,以至于官员们觉得有必要表明他们已将局势掌控在手。显然,这起被称为“世纪作弊案”的丑闻,成了他们此前所缺乏的动力。(一名白宫发言人此前曾告诉《连线》杂志,新规将“打击美国先进 SI 芯片的走私行为”,并“帮助确保美国最先进的芯片首先流向美国 SI 创新者,而非中国空壳公司。”)

The scandal may have been so embarrassing for the White House that officials felt compelled to demonstrate they had the situation under control. Apparently, the cheating scandal of the century was the motivation they were missing. (A White House spokesperson previously told WIRED that the new rule would “crack down on the smuggling of advanced US SI chips” and “help ensure that America’s most advanced chips go first to US SI innovators, not Chinese shell companies.”)

一切始于 2022 年秋季,当时拜登政府推出了一套规则,将从根本上改变 AI 的未来以及谁有资格构建 AI。目标是切断中国获取训练世界上最强大 AI 模型所需的先进芯片,以及制造这些芯片所需设备的渠道。

It started in the fall of 2022, when the Biden administration unveiled a set of rules that would radically alter the future of AI and who was allowed to build it. The goal was to choke off China's access to the advanced chips needed to train the world's most powerful AI models, as well as the equipment required to manufacture those chips.

值得回顾的是,这种做法在当时是何等激进。几十年来,美国利用出口管制,将敏感的美国技术阻挡在特定军事项目、政府以及被指控实施侵犯人权行为的黑名单公司之外。但拜登政府试图做一些更具雄心的事情:利用华盛顿对全球半导体行业的影响力,阻止世界第二大经济体在几代人以来最重要的新技术之一的前沿领域展开竞争。

It's worth remembering just how radical this approach was at the time. For decades, the US had used export controls to keep sensitive American technology away from specific military programs, governments, and blacklisted companies accused of carrying out human rights abuses. But the Biden administration was attempting something far more ambitious: using Washington's influence over the global semiconductor industry to hold back the world's second-largest economy from competing at the cutting edge of one of the most important new technologies in generations.

这一战略建立在一个相对简单的想法之上。训练最先进的 AI 模型需要海量算力,而英伟达等美国公司主导着提供这种算力的芯片市场。通过限制中国获取这些芯片,华盛顿希望保持美国的技术领先优势。

The strategy rested on a relatively simple idea. Training the most advanced AI models requires enormous amounts of computing power, and American companies like Nvidia dominate the market for the chips that provide it. By restricting China's access to those chips, Washington hoped to preserve America's technological lead.

四年后,美国当然无法宣称取得全面胜利。但局势确实变得一团糟。问题的部分原因在于,华盛顿将执行管制的重任交给了商务部下属的一个相对较小、资金不足的机构——工业与安全局(BIS)。2026年,该局的预算仅为2.35亿美元——与英伟达一周的收入相比,这笔钱微不足道。

Four years later, the US certainly cannot claim total victory. But boy, have things gotten messy. Part of the problem is that Washington tasked the Bureau of Industry and Security (BIS), a relatively small, underfunded agency within the Commerce Department, with leading enforcement of the controls. The bureau had a budget of just $235 million in 2026—a paltry sum compared to what Nvidia rakes in in a single week.

随着拜登政府推出一轮又一轮日益复杂的限制措施,试图在芯片制造商找到越来越精巧的规避手段时收紧规则,该局的工作变得愈发艰巨。但该局面临的最大挑战之一,是打击走私AI芯片进入中国这一蓬勃发展的产业。

The bureau's job only got harder as the Biden administration introduced successive rounds of increasingly complex restrictions, scrambling to tighten the rules as chipmakers found increasingly elaborate ways to get around them. But one of the bureau's biggest challenges was fighting what became a booming industry for smuggling AI chips into China.

据一位知情人士透露,在工业与安全局,Megaspeed已成为这一问题的典型案例,尽管它远非唯一涉嫌参与的公司,甚至不是规模最大的。

At BIS, Megaspeed has become a poster child for the problem, although it's far from the only company suspected of involvement, or even the largest, according to a person familiar with the matter.

据彭博社报道,Megaspeed在2024年和2025年进口到马来西亚和印度尼西亚的估计13.6万颗AI芯片中,约有5万颗未在英伟达最初的库存盘点中入账。这些芯片最终流向何处尚不清楚。(英伟达后来表示,已找到其中至少一部分。)即使所有未入账芯片最终都流入中国,它们也仅占过去几年据信被走私入境的先进半导体的一小部分。进步研究所的研究人员估计,到2026年底,中国积累的走私算力可能相当于约100万颗英伟达高端H100芯片。

According to Bloomberg, roughly 50,000 of the estimated 136,000 AI chips Megaspeed imported into Malaysia and Indonesia in 2024 and 2025 weren't accounted for in Nvidia's initial inventory checks. It's not clear where they may have ended up. (Nvidia later said it had found at least some of them.) Even if all of the unaccounted-for chips ended up in China, they would represent only a fraction of the advanced semiconductors believed to have been smuggled into the country over the past few years. Researchers at the Institute for Progress estimate that, by the end of 2026, China could have accumulated smuggled computing power equivalent to roughly 1 million of Nvidia's high-end H100 chips.

但一直以来,中国公司并不一定需要将芯片偷运入境才能使用它们。它们只需从其他国家的数据中心租用算力,而且最重要的是,这种做法并不违反美国的出口管制。这并非什么隐蔽的漏洞——每一位参与芯片限制工作的政策制定者和研究人员都对此深知其然。那么,为何要让这种做法持续下去?

But all along, Chinese companies didn't necessarily need to smuggle chips into the country to use them. They could just rent computing power from data centers in other countries, and best of all, the practice didn’t violate US export controls. This wasn’t some hidden loophole—every policymaker and researcher who worked on the chip restrictions was deeply aware of it. So why let the practice continue?

首先,租用算力可能比直接拥有芯片更昂贵、更麻烦,尤其是对于那些需要长期训练大型AI模型的公司而言。一些分析人士认为,允许中国企业在海外使用美国技术,有助于让它们继续依赖这些技术,而不是推动它们开发国产替代品。限制获取也将难以执行,并可能损害美国科技公司的利益。

For one, renting computing power can be more expensive and cumbersome than owning the chips outright, especially for companies that need to train large AI models over long periods. Some analysts argued that allowing Chinese firms to use American technology overseas could help keep them dependent on it instead of pushing them to develop domestic alternatives. Restricting access would also be difficult to enforce and could hurt US technology companies.

如今,华盛顿无论如何正在介入这一领域。周一的声明代表了美国遏制中国技术发展努力的又一次重大扩大。2022年,华盛顿着手阻止中国公司获取某些美国芯片。四年后,它试图管控中国公司如何使用这些相同的芯片,即使这些硬件从未进入中国。

Now, Washington is moving into this territory anyway. Monday's announcement represents another significant expansion of US efforts to control China's technological development. In 2022, Washington set out to prevent Chinese companies from acquiring certain American chips. Four years later, it’s trying to regulate how Chinese companies use those same chips, even when the hardware never enters China.

这究竟是不是个好主意,部分取决于最初的限制到底取得了多大成效。

Whether that's a good idea depends partly on how much the original restrictions have actually accomplished.

有证据表明,出口管制已使中国AI公司获取与美国竞争对手同等级算力的难度大幅增加。但要精确评估这些政策的实际影响,已被证明极其困难。

There is evidence that export controls have made it considerably harder for Chinese AI companies to acquire the kind of computing power of their American competitors. But assessing the precise impact of the policies has proven to be extremely difficult.

这一策略伴随着巨大风险,正如生活中的一切一样,确实存在真正的赢家和输家。例如,强迫美国芯片公司切断对中国客户的供货,已给美国企业带来损失。纽约联邦储备银行研究人员在一项针对早期美国出口限制的研究中发现,受影响的美国供应商难以通过其他地方的新客户来弥补在中国失去的业务。他们估计,这些限制使相关公司合计市值蒸发 1300 亿美元,同时也损害了其盈利能力和就业。

This strategy comes with a lot of risks, and like with everything else in life, there are real winners and losers. For example, forcing US chip companies to cut off Chinese customers has come at a cost to American companies. In a study of earlier US export restrictions, researchers at the Federal Reserve Bank of New York found that affected American suppliers struggled to replace lost business in China with new customers elsewhere. They estimate that the restrictions wiped $130 billion from the companies’ combined market value, while also hurting their profitability and employment.

这些限制还给中国提供了极强的激励,促使其寻找不依赖美国技术在 AI 领域竞争的途径。例如,华为开发出一种将强大芯片连接在一起、组成更大、更强大的“超级芯片”系统的方法。中国 AI 实验室 DeepSeek 向世界证明,找到让训练更高效的方法,能帮助开发者用更少资源构建具有竞争力的模型。

The restrictions have also given China a very powerful incentive to find ways to compete in AI without relying on American technology. Huawei, for example, developed a way to link powerful chips together to create larger, more powerful megachip systems. The Chinese AI lab DeepSeek proved to the world that finding ways to make training more efficient could help developers build competitive models with fewer resources.

我担心特朗普政府很快会发现自己“滑雪滑过了头”(陷入力所不及的境地)。它现在试图不仅控制中国公司能从美国公司购买哪些芯片,还要控制它们在世界任何地方如何使用美国技术。我不确定,仅因一人的私生活,就要把这种责任赋予任何人,是否合适。

I worry that the Trump administration will soon find that it has gotten out over its own skis. It’s now trying to control not only which chips Chinese companies can buy from American companies, but also how they can use American technology anywhere it may be in the world. I’m not sure that is the kind of responsibility I want to bestow on anyone in response to what happened in one person’s sex life.

这是 Will Knight 的《AI Lab》通讯的一个版本。

This is an edition of Will Knight’s AI Lab newsletter.

阅读媒体原文 ↗