In the last few years, the U.S. government has imposed some of the toughest tech export restrictions in modern history, targeting China’s access to advanced artificial intelligence (AI) chips. These policies were meant to slow down China’s AI development and keep powerful GPUs like those made by Nvidia and AMD out of Chinese hands. But now, the very companies these restrictions were supposed to protect are saying: the plan failed — and it’s hurting American business more than it’s helping national security.
The Beginning of U.S. Chip Controls
It started in October 2022. The U.S. Commerce Department’s Bureau of Industry and Security introduced sweeping rules to block the sale of high-end AI chips and chipmaking tools to China. The first targets were Nvidia’s A100 and H100 GPUs — chips considered the gold standard for training AI models.
Washington’s reasoning was simple: stop China from using American chips in military applications. Officials called it a move against “military-civil fusion.” The controls also affected machines from companies like ASML, used to make the most advanced semiconductors.
But companies found workarounds. Nvidia introduced slower versions like the A800 and H800, which met the legal limits but still gave Chinese firms powerful computing tools. In October 2023, the U.S. updated its rules to block even those chips. A license was now required for almost any AI-capable chip sold to China — effectively shutting the door.
Then, in January 2025, the Biden administration proposed an “AI Diffusion Rule,” dividing countries into tiers. China was in Tier 3, the most restricted. But even before it took effect, President Trump announced plans to reverse it. In May 2025, Nvidia CEO Jensen Huang openly praised Trump’s plan, calling the current system flawed.
Nvidia: The Biggest Casualty
Of all the American tech companies, Nvidia has been hit the hardest. Before the restrictions, Nvidia held about 90–95% of China’s high-end GPU market. AI giants like Alibaba, Tencent, and ByteDance were major customers, using these chips to train massive AI models like chatbots and recommendation engines.
But after the restrictions, Nvidia’s market share in China dropped to 50%. In April 2025, Nvidia announced it would take a $5.5 billion loss due to licensing issues with its new H20 chip — specifically designed for China. On an earnings call, Huang admitted the company had to forfeit $15 billion in potential H20 sales.
That’s not all. In a single quarter, Nvidia lost $2.5 billion in revenue due to blocked China sales. These are no small numbers. China used to represent 13–15% of Nvidia’s global revenue. A big portion of the company’s future growth was tied to China’s booming AI sector.
To adapt, Nvidia tried engineering chips like the H20 and now the “Blackwell” series with lower specs to meet U.S. rules. But regulators kept tightening the definitions, and each workaround got shut down. Huang called this back-and-forth “innovation under duress” — expensive, inefficient, and damaging.
AMD and Intel Join the Pain
AMD is facing similar problems. It sells AI chips like the Instinct series, and China is a huge part of its business — roughly 24% of its total revenue. In May 2025, AMD warned that new U.S. rules would cut $1.5 billion from its annual sales.
On a call with investors, AMD said that without the restrictions, their Q2 2025 revenue could’ve been $8.1 billion. Instead, they had to lower expectations to $7.4 billion.
Even Intel, which isn’t known for AI chips, is now caught in the dragnet. In April 2025, it told Chinese customers that its Gaudi AI accelerators would also require licenses under the new export criteria. Any chip with DRAM bandwidth over 1400 GB/s or I/O above 1100 GB/s is now blocked. Intel didn’t comment publicly, but behind the scenes, it admitted that some of its chips are no longer accessible to China.
National Economic Impact
This isn’t just about a few companies. A Federal Reserve study found that U.S. chip firms affected by the 2022 export rules experienced sharp declines in revenue, profitability, and access to credit. Many simply couldn’t find alternative buyers to replace China.
According to the U.S.–China Business Council, semiconductor exports to China fell by $2.9 billion in 2022. In Oregon, where Intel has a major presence, chip exports dropped by $1.8 billion. Nationally, semiconductor shipments to China fell from $14.2 billion in 2021 to $11.2 billion in 2022 — despite chips still being America’s second-largest export to China.
Industry analysts estimate the total China AI market is worth $50 billion a year, much of which is now beyond reach for U.S. companies. Nvidia and AMD can’t access this growth, while competitors from South Korea or Chinese firms like Huawei may step in.
So… Did the Restrictions Work?
This is the big question. Did the sanctions actually slow China down, or just hurt U.S. businesses?
On the surface, China’s chip imports dropped. In 2023, China imported 479.5 billion chip units worth $349.4 billion — a 15.4% decrease in value from 2022. Imports of lower-end chips fell even more, around 24%, as Chinese firms started buying from domestic suppliers instead.
But a deeper look tells a different story. A Federal Reserve Bank study notes that after the initial plunge in 2022, China’s chip imports bounced back. In fact, total chip imports in 2024 are now tracking the same upward trend from before the sanctions.
Why? Because most of the U.S. controls focus on advanced chips, like AI accelerators. But China still imports huge volumes of older chips — the kind used in cars, phones, and appliances. These “legacy chips” make up the bulk of imports and are still widely available from places like Taiwan and South Korea.
China is also building its own capacity. By 2024, it had the world’s largest chip fabrication output by volume, especially in older 28nm and 45nm nodes. While these aren’t cutting-edge, they’re good enough for many industrial uses — and they reduce reliance on U.S. suppliers.
Conclusion: A Costly Strategy
When the Biden administration introduced these chip controls, the goal was to keep military-grade AI away from China. But the execution has been messy.
- Nvidia lost $15 billion in sales, with over $5.5 billion in write-downs.
- AMD expects $1.5 billion in losses.
- Intel’s chips are now partly blocked.
- U.S. chip exports to China dropped by $2.9 billion in 2022.
- Domestic job and innovation risks are mounting.
Even worse, China’s chip access hasn’t truly collapsed. Instead, it adapted — importing more legacy chips and developing its own alternatives. The sanctions may have slowed Beijing’s ambitions, but they didn’t stop them.
Now, with the potential reversal under a Trump presidency, the U.S. tech industry is at a crossroads. Is it better to restrict access and risk domestic losses, or to embrace global competition and trust U.S. innovation to stay ahead?
It’s a debate that goes far beyond Nvidia’s bottom line — it’s about the future of the global tech order.