The company can't keep redesigning forever.
In the quiet arithmetic of global power, the United States moved again to slow China's ascent in artificial intelligence — this time by closing the gaps that had allowed modified chips to slip through earlier restrictions. Nvidia and AMD, the twin pillars of the world's AI infrastructure, felt the weight of that decision in their share prices, even as demand for their technology surged everywhere else on earth. The episode illuminates a deepening contradiction at the heart of American strategy: the same companies Washington depends on to lead the AI age are the ones it must restrain in the name of national security.
- Nvidia fell 3.2% and AMD dropped roughly 3% after reports that Washington may close the regulatory loopholes both companies had quietly relied upon to keep selling chips to China.
- The Commerce Department is weighing a July move to extend export controls to lower-powered AI chips — the very workaround products Nvidia engineered after the first round of restrictions hit it for $400 million in a single quarter.
- Beijing's AI ecosystem shuddered in response, with major hardware suppliers Inspur and Unisplendour each tumbling 10% as Chinese markets absorbed the implications.
- Washington is also considering barring Chinese AI firms from leasing cloud computing capacity through Amazon and Microsoft, threatening to cut off a second critical pathway to advanced AI development.
- Analysts warn the restrictions may slow but not stop China — Chinese firms are already sourcing chips through third-party countries and accelerating domestic semiconductor development as alternative routes multiply.
Markets reacted sharply on Wednesday when reports emerged that the Biden administration was preparing to tighten semiconductor export controls targeting China. Nvidia, whose chips are the essential engine of modern AI training, dropped as much as 3.2% in after-hours trading, while Advanced Micro Devices fell roughly 3%. Together, the two companies command more than four-fifths of the global market for data center accelerator chips — the processors that make systems like ChatGPT possible.
The immediate concern was that Washington intended to close loopholes both companies had used to sell modified, lower-powered chip versions to China without triggering licensing requirements. For Nvidia, which draws about one-fifth of its annual revenue from China, the stakes were concrete. The company had already engineered workaround chips to stay beneath earlier regulatory thresholds — a strategy that had softened the blow of restrictions first announced in August 2022, which had still cost the company $400 million in a single quarter.
The geopolitical logic behind the move was familiar: the Biden administration has made constraining China's technological rise a pillar of its national security posture, with AI seen as a decisive frontier in military and scientific competition. Beyond chips, officials were also weighing restrictions on Chinese firms' access to cloud computing services offered by Amazon and Microsoft — platforms now indispensable for training large AI models. In Chinese markets, the news triggered immediate losses in AI-related hardware stocks, with Inspur and Unisplendour each falling 10%.
Yet the long-term picture remained uncertain. Analysts noted that Chinese firms were already pursuing alternatives — sourcing chips through third-party nations, investing in domestic production, and engineering their own technical workarounds. Bloomberg Intelligence analyst Robert Lea cautioned that while new restrictions might delay Chinese AI progress, an increasingly inventive response from Chinese firms would likely limit the damage over time. The deeper problem for American policymakers was structural: the global semiconductor supply chain is so deeply interwoven that closing one door tends to open others. Nvidia declined to comment, but the tension between Washington's security ambitions and the commercial realities of its largest chipmakers showed no sign of easing.
The stock market reacted sharply on Wednesday to news that the Biden administration was preparing to tighten restrictions on the sale of advanced semiconductors to China. Nvidia, the dominant maker of chips essential to training artificial intelligence systems, dropped as much as 3.2% in after-hours trading. Advanced Micro Devices fell roughly 3%. The two companies control more than four-fifths of the global market for data center accelerator chips—the specialized processors that power systems like ChatGPT.
The concern was straightforward: Washington was considering closing loopholes that had allowed both companies to continue selling modified, less-powerful versions of their chips to China without triggering export licensing requirements. According to reporting from the Wall Street Journal, the Commerce Department could move as soon as July to expand restrictions to include these lower-capability semiconductors. For Nvidia, which derives about one-fifth of its annual revenue from China, the implications were immediate and material. The company had already designed workaround chips specifically to stay beneath the regulatory thresholds—a strategy that had helped it partially offset the impact of earlier restrictions announced in August 2022. Those initial curbs had cost Nvidia $400 million in a single quarter.
The broader context was geopolitical. The Biden administration has made containing China's technological advancement a centerpiece of its economic and national security strategy. Officials worry that Beijing's progress in artificial intelligence could shift the balance in military capabilities and scientific innovation. The restrictions on chip sales were one lever among several. The administration was also weighing whether to limit Chinese AI companies' access to cloud computing services provided by companies like Amazon and Microsoft—platforms that have become essential infrastructure for training large language models.
In Chinese markets, the news triggered a sell-off in AI-related stocks. Inspur Electronic Information Industry and Unisplendour, both major hardware suppliers, dropped 10% as traders absorbed the implications of tighter American controls. Yet the picture remained complicated. Nvidia and AMD, despite the immediate stock pressure, continued to benefit from a global surge in AI investment. Companies from Microsoft to Baidu to OpenAI were racing to acquire their chips to build the next generation of AI services. The two chipmakers had become indispensable to the worldwide AI boom, even as Washington tried to prevent that technology from reaching China.
Analysts offered a cautious view of how effective the new restrictions might prove. Robert Lea, an analyst at Bloomberg Intelligence, noted that Chinese firms had already begun exploring alternatives—sourcing chips from third-party countries, developing their own semiconductors, or finding other technical workarounds. "While further restrictions could delay AI developments by Chinese firms, I don't see a major long-term impact as Chinese firms take an increasingly innovative approach to workarounds," he said. The fundamental challenge for American policymakers was that the global semiconductor supply chain was deeply interconnected. Cutting off one route often meant Chinese companies would find another.
Nvidia declined to comment on the Journal's reporting. The company had navigated the previous round of restrictions by engineering solutions that stayed within the letter of the law while preserving access to one of the world's largest markets. Whether it could do so again remained unclear. What was certain was that the tension between American national security concerns and the commercial interests of its largest chipmakers would only intensify as artificial intelligence became more central to economic competition between the two superpowers.
Citações Notáveis
While further restrictions could delay AI developments by Chinese firms, I don't see a major long-term impact as Chinese firms take an increasingly innovative approach to workarounds.— Robert Lea, Bloomberg Intelligence analyst
A Conversa do Hearth Outra perspectiva sobre a história
Why does it matter that Nvidia gets a fifth of its revenue from China? That's a significant chunk.
It's not just the percentage—it's the absolute scale. China is the world's second-largest economy and has enormous demand for AI infrastructure. Losing that market, or even a portion of it, directly hits Nvidia's growth trajectory at a moment when AI is driving the entire semiconductor industry forward.
So the company designed weaker chips to get around the rules. That seems clever, but also fragile.
Exactly. It's a cat-and-mouse game. Nvidia found a technical solution that satisfied the letter of the law—these modified chips didn't trigger the licensing requirement. But now the government is saying it will close that loophole. The company can't keep redesigning forever.
The article mentions cloud services restrictions too. How does that change things?
It broadens the attack surface. Even if Chinese companies can't buy the best chips directly, they could rent computing power from American cloud providers to train their models remotely. Restricting that cuts off another pathway. It's about layering controls.
But the analyst says Chinese firms will find workarounds anyway. So what's the point?
That's the real tension. The restrictions might slow China down, but they probably won't stop it. Chinese companies are already exploring alternatives—building their own chips, buying from other countries, developing different architectures. The question is whether delay itself is valuable to American interests.
Does this hurt American companies beyond just lost sales?
Potentially. If you're Microsoft or Amazon, you're caught between American national security policy and losing business to competitors in other countries who don't have the same restrictions. And if Chinese firms do develop workarounds, they might become less dependent on American technology long-term, which could hurt in the future.