Peace on the Table in Shenzhen, Chips Underneath: Jensen Huang's Fall and the Next Round of the AI Blockade
On November 19, Xi, Trump and Putin hold their first trilateral summit in Shenzhen while a trade truce hangs over rare earths and chips. A week earlier, Congress's campaign felled Jensen Huang. A deep ledger of the AI blockade: America's tower top, China's tower body, and the truce, legislation and hedging that come next.
November 19, Shenzhen. China's Foreign Ministry confirmed on November 16 that on the final day of the APEC Economic Leaders' Meeting, Xi Jinping and Donald Trump will hold a bilateral meeting at midday, and at 2:30 p.m. Xi, Trump and Vladimir Putin will sit down at Kylin Villa for the first China-US-Russia heads-of-state meeting in history. The White House says the agenda is nuclear arms reduction and Iran; Foreign Ministry spokeswoman Mao Ning says it is about "world peace and the strategic stability of major-power relations." But everyone knows what sits under the table: the trade and technology war between China and the United States, freshly extended into another truce. And one week before the talks, that technology war claimed its biggest casualty yet — Nvidia founder Jensen Huang.
Expectations for Shenzhen are deliberately low. Analysts broadly agree the symbolism outweighs any substance, and US officials have already said no agreement is expected. The real US-China agenda is damage control: the trade truce was extended on November 9 to January 10, 2027; Washington accuses Beijing of delivering only two-thirds of its promised rare-earth exports, which Beijing denies; and the Xi-Trump bilateral is about implementing the truce and rare-earth supply. US officials have also said plainly that Nvidia and the Huang affair will not be on the trilateral table — even though it is the newest and bloodiest episode of the technology war.
First, how Huang fell. On November 8 he sent an all-staff letter admitting an extramarital affair and apologizing to his family, employees and shareholders; the board opened an independent investigation the same day. On November 11 the board accepted his resignation as chairman, CEO and director, effective immediately, with CFO Colette Kress taking over as interim CEO. Far more than private morality was stacked against him: the chairs of three House committees had written on November 4 demanding he testify on November 18; Senator Kevin Cramer said publicly that Huang should resign "in good conscience"; the FBI is investigating, shareholder lawsuits are coming, and Nvidia has lost roughly $600 billion in market value since the board announced its probe into his relationship with a businesswoman. That businesswoman has been identified by US intelligence, according to media reports, as a Chinese spy — a claim China's Foreign Ministry has dismissed as groundless. The investigation reportedly began back in October 2025, yet it detonated only after the November 3 midterms, in which the Democrats won big and are set to retake Congress in January. Trump's repeated public defenses could not save him.
Why is the campaign against Huang, in essence, a campaign against China? Start with identity. Huang was born in Tainan, Taiwan, and is the most prominent ethnic-Chinese face in American technology. Across the Greater China commentariat, many have noted that it is precisely that face which makes a "spy" allegation easier to stick — former media figure Hu Xijin wrote that the affair is a wake-up call for Chinese-American elites. Then consider his position: Huang was the most senior US chief executive openly opposed to chip blockades on China, arguing repeatedly that export controls would only force China to build its own stack and cost American firms the Chinese market. When H200 exports to China were approved in January 2026 — on condition that 25% of sales go to the US government — he was the industry's hardest-pushing advocate, and congressional China hawks had long resented him for it. Finally, watch what followed: before he had even left, Senators Elizabeth Warren and Jim Banks were using the moment to push the AI Oversight Act, the MATCH Act and the Chip Security Act into the annual defense authorization bill, tightening chip exports to China further. Bloomberg wrote within days that China was the saga's "biggest loser" — an inadvertent admission that this was never just a sex scandal. What was removed was not merely an unfaithful CEO, but the last heavyweight voice in Washington willing to say out loud that America should sell chips to China — and the chill now runs through every Chinese-American engineer watching.
Now the ledger of the blockade itself. America's advantage is the top of the tower: the most advanced process nodes and equipment alliance, the strongest accelerators and the CUDA ecosystem, the frontier models and the deepest capital. But the backlash is compounding. According to IDC, Nvidia's shipments in China fell 44% year on year in the first half of 2026, its share of China's data-center AI chip market dropped to 34%, and Huawei climbed to 28% — a rout has become a dogfight. The approved H200 brings in less than 1% of Nvidia's data-center revenue: not because China will not buy, but because trust and ecosystems have already shifted. Every turn of the screw hands another slice of the Chinese market to domestic substitutes, intact, and forces the rival to rebuild an entire supply chain of its own.
China's advantage is the body of the tower: the world's largest application market and use cases, a vast engineering workforce, fast-ramping domestic chips led by Huawei's Ascend line, open-source models spreading capability across society, and upstream leverage such as rare earths. Its weaknesses are equally clear: a real gap in advanced nodes and high-bandwidth memory, a capped total supply of high-end training compute, yields and capacity still climbing, and a CUDA ecosystem whose stickiness cannot be replaced overnight. One new signal is talent: Nvidia employs nearly 4,000 people in China, mostly in R&D, with annual attrition of just 0.9%. Since Huang's fall, his China teams are unsettled and domestic firms are openly recruiting them. A blockade fought long enough starts sending the talent home.
Three lines will shape what comes next. First, managed truce as the norm: Shenzhen sets the tone of fighting without breaking, leaders' meetings guard the floor, and chips move through case-by-case approvals in the H200 style — pay the levy, pass the review, ship a portion. Second, the blockade hardens from executive orders into law: once the Democrats take Congress in January, chip-security legislation will make controls stricter and far harder to unwind, and "doves" like Huang will have even less room in Washington. Third, China's hedging accelerates on all fronts at once: domestic substitution, rare-earth counter-leverage, open-source ecosystems and third-country markets. The real contest is not any single ban but the ecosystem map three to five years out — America holds the top of the tower, China takes the body, and whoever keeps the world's developers wins the second half. Inside Kylin Villa on November 19, the leaders will talk about nuclear weapons and world peace. What decides the next decade is the chip supply chain outside the room — cut, and endlessly reconnected.
November 17, 2026
Share this article
What do you think of this story?
Anonymous feedback, up to 5 reactions per article.
Reader comments
Comment without signing in, or sign in to display your member name.
Discuss the technology and the story. Hate, politics and war, threats, abuse, personal accusations, sexual content, ads, links and personal information are not allowed.
No comments yet. Be the first to share.