Nvidia Can't Live Without China
Fresh & Hot curated AI happenings in one snack. Never miss a byte 🍔
This snack byte will take approx 4 minutes to consume.
When Jensen Huang stepped up to Nvidia's earnings call last week, he delivered what amounted to a diplomatic ultimatum wrapped in corporate speak.
"China's AI moves on with or without U.S. chips," the CEO told analysts, his message clearly aimed at the Trump administration's trade warriors.
But behind that defiant tone lay a sobering reality:
Nvidia just took a $5.5 billion write-down on inventory for its H20 chip—what analysts called the biggest write-down in the history of the chip industry.
That's not just a financial hit. It's a public admission that the world's most valuable semiconductor company has been checkmated by geopolitics.
Since 2022, Washington has been steadily tightening the screws on AI chip exports to China, starting with Nvidia's most advanced H100 processors. The company responded like any savvy tech giant would: they designed around the restrictions.
Enter the H20—a chip specifically engineered for the Chinese market to comply with U.S. export controls while still delivering decent AI performance. It was a clever workaround that allowed Nvidia to keep selling to Chinese tech giants like Tencent and ByteDance.
For a while, it worked beautifully.
Then came April 2025.
The US government informed Nvidia on Monday that the H20 would require a license to export to China "for the indefinite future." The new rules addressed Washington's concerns that the chips could be diverted to Chinese supercomputers—effectively killing Nvidia's last legal pathway into the world's second-largest economy.
The financial carnage was immediate. Beyond the massive write-down, export restrictions have already cost Nvidia $2.5 billion in lost sales last quarter, with another $8 billion hit expected this quarter.
That's $10.5 billion in lost revenue for a single year—more than most companies make in total.
China is Quarter of Everything
Here's why Nvidia can't simply shrug off China and focus on booming demand elsewhere: the math doesn't work.
China represents roughly 25% of the global AI chip market—a $50 billion opportunity that Nvidia estimates could grow exponentially as Chinese companies race to build their own AI capabilities.
Wall Street analysts are betting on Nvidia hitting $200 billion in revenue this year and $300 billion by 2028. Those projections assumed China would eventually come back into play.
As UBS analyst Tim Arcuri put it: "China is a quarter of the market. It's a big number." He added that Nvidia would have a "dominant hold" on that market if it were able to compete there. That "if" is doing a lot of heavy lifting.
The company's $3.3 trillion valuation—triple the market cap of its nearest chip rival—is built on the premise of continued explosive growth. Voluntarily sitting out a quarter of your addressable market makes those growth targets look increasingly fantastical.
While Nvidia has been playing regulatory whack-a-mole with Washington, Chinese competitors haven't been sitting idle.
Huawei Technologies is preparing to test its newest and most powerful AI processor, hoping to replace some higher-end products of U.S. chip giant Nvidia, with the company readying its new AI chips for mass shipment as China seeks Nvidia alternatives.
The competitive dynamics are shifting rapidly.
At Computex 2025, Huang told global media that Nvidia's market share in China has sharply declined, which will drive Chinese customers to support local chip companies such as Huawei. It's a remarkable admission from a CEO who built his empire on technological dominance.
Analysts pointed to Huawei as a clear leader in China's race to find a Nvidia competitor, with the U.S.-blacklisted company working on its own "Ascend 910" GPU series, including the latest Ascend 910C and the upcoming Ascend 910D processor.
The irony is palpable:
U.S. export controls designed to slow China's AI development are accelerating the very domestic competition that threatens Nvidia's long-term dominance.
Morgan Stanley estimates China can now supply 34% of its AI chip needs domestically, jumping to 82% by 2027. Every quarter Nvidia stays locked out, that self-sufficiency ratio grows.
The Geopolitical Gordian Knot
The path back to China runs straight through Washington, but the political headwinds are only getting stronger. Recent developments suggest the restrictions are tightening rather than loosening:
The Commerce Department has issued new warnings against the use of U.S. AI chips for Chinese models and is targeting "diversion tactics" to prevent smuggling
Washington specifically singled out Huawei's Ascend chips as a national security concern
President Trump has accused China of violating recent trade deals, threatening a rebound in tensions
Nvidia CEO Jensen Huang even discussed concerns about Huawei Technologies' growing artificial intelligence capabilities with U.S. lawmakers—a sign that the company is now actively lobbying Washington about the competitive threat created by export restrictions.
The strategic contradiction is obvious: restrictions designed to maintain U.S. technological leadership are forcing the world's most advanced AI chip company to cede market share to Chinese competitors.
As Huang argued in his earnings call: "Export controls should strengthen U.S. platforms, not drive half of the world's AI talent to rivals."
Perhaps most troubling for Nvidia is the innovation acceleration happening in China.
Chinese tech giants like Tencent, Alibaba, and ByteDance—companies that were major Nvidia customers—are now being forced to work with domestic alternatives. This isn't just about market share; it's about ecosystem development.
"In the end, the platform that wins the AI developers wins AI," Huang noted.
Every Chinese developer forced to optimize for Huawei's Ascend chips instead of Nvidia's CUDA platform is a developer lost, potentially forever. Network effects in software ecosystems are notoriously difficult to reverse.
The Chinese government has provided massive backing for this transition. Venture funds backed by Beijing invested $184 billion into AI startups between 2000 and 2023, according to Morgan Stanley.
That's not investment in Nvidia's ecosystem—it's investment in replacing it.
Wall Street analysts like Morgan Stanley's Joe Moore remain optimistic: "We remain convinced that there will be at least some recovery of the China opportunity."
But that confidence looks increasingly misplaced given recent political developments.
UBS's Arcuri argues that modifications allowing more powerful chips could still comply with Washington's aims, since restrictions on advanced chipmaking equipment limit what China can actually produce domestically. "If those equipment restrictions don't change, there is this inherent ceiling to what they can do," he said.
But this analysis misses the broader strategic picture.
Even if China can't immediately replicate Nvidia's most advanced chips, they're rapidly building alternative ecosystems that don't depend on U.S. technology.
The brutal irony of Nvidia's China situation is that the company is still years ahead of any rival. Its ecosystem of chips, systems, and software tools remains the gold standard for AI development globally.
Chinese companies would prefer to use Nvidia's technology to stay competitive internationally—if they could.
But geopolitics has other plans.
The company that achieved the impossible—building a $3.3 trillion valuation by riding the AI wave—now finds itself a pawn in a larger game between superpowers. The most advanced chip company in human history is being forced to watch one of its biggest markets develop without them.
As export restrictions tighten and Chinese alternatives mature, Nvidia faces an uncomfortable truth: you can have the best technology in the world, but if geopolitics won't let you sell it, innovation becomes irrelevant.
The road back to China runs through Washington.
But at this point, even Nvidia's $3.3 trillion market cap might not be enough to move the political needle.
About the author: Rupesh Bhambwani is a technology enthusiast specializing in the broad technology industry dynamics and international technology policy.
When not obsessing over nanometer-scale transistors, energy requirements of AI models, real-world impacts of the AI revolution and staring at the stars, he can be found trying to explain to his relatives why their smartphones are actually miracles of modern engineering, usually to limited success.
The problem for the US is that China is doing the right things while the US is doing the wrong things.
The US government is trying to “slow” China’s accumulation of technological know-how, but the actions it takes slow its own technological progress. Take the policies of the last few weeks:
A Chinese visa “crackdown” is coming that will send thousands, if not tens of thousands, of talented engineers back to China (or deter them from moving to the US in the first place).
Compare this to China’s Thousand Talents program, which specifically aims to bring in more foreign talent to China.
The White House is also pushing massive funding cuts for science, R&D, and subsidies for emerging technology across NASA, the NIH, among others.
China, on the other hand, is pumping more money into emerging technologies, like solar, batteries, EVs, AI, and robotics.
Further, US export restrictions on key American chips help China catch up by creating more demand for domestic alternatives. Learning curve effects are rapidly taking root in China as a consequence: https://www.lianeon.org/p/the-experience-curve?utm_source=publication-search
US policy complements China’s by accelerating China’s technological ascent.
The US should be focused on itself, drawing in more of the best talent from the world. Investing more in emerging technologies and selling those technologies abroad on a greater scale to accelerate its own learning curves.