Wired reports that President Trump called Jensen Huang in the middle of an Nvidia all-hands meeting, and that tells you plenty about how much of Nvidia's China strategy now depends on the CEO's access to one man.
Huang stopped what he was doing and took the call. According to Wired, Trump rang the Nvidia CEO on Thursday morning while Huang was on stage at the company's Santa Clara headquarters for an all-hands meeting, with staff in the room and thousands more watching by livestream. Employees could hear Trump's voice through Huang's microphone, though not clearly enough to catch the words. Less than a minute later, Huang was back with his own staff.
That's unusual access. It's also no small thing. Nvidia isn't a normal supplier trying to win a tax break or a factory ribbon-cutting. Its chips sit at the center of the fight over whether China gets access to the hardware that powers frontier AI.
The call fits a pattern #
This wasn't Huang's first direct line into Trump's schedule. Axios reported last October that Huang said all of Trump's calls to him came at 10:30 at night, the president's time. Fortune and the Associated Press reported in June that Huang described Trump calling him in the middle of the night to talk about jobs, reindustrializing the United States, national security and winning. You don't need much imagination to see why that matters when export licenses can move billions of dollars in or out of Nvidia's reach.
The sharpest example came in May. Semafor reported that Huang had been left off Trump's China delegation because Nvidia's chip sales to China were politically awkward, then Trump called after seeing coverage of the snub and told Huang he wanted him on the trip. Huang flew to Alaska and met Air Force One. Nvidia said Huang was attending the summit at Trump's invitation to support the administration's goals.
Google, Amazon and Meta Are Out-Growing Nvidia With Their Own AI Chips Custom AI chips from Google, Amazon, Microsoft and Meta are growing shipments 44.6% this year, nearly triple Nvidia's GPU growth rate, according to TrendForce. Nvidia's revenue and margins are still climbing, but its own biggest customers are quietly building the competition. - custom AI chips outpacing Nvidia GPU growth rates - why tech giants building proprietary AI accelerators instead
That is not how a normal policy process feels from the outside.
Nvidia has spent the last year trying to keep a China business alive under rules that keep changing. In its annual filing, the company said the U.S. government told it in April 2025 that exports of its H20 chips to China would need a license. Nvidia then took a $4.5 billion charge tied to H20 inventory and purchase obligations. The rules kept moving. By August 2025, the government had granted licenses allowing some H20 shipments to China-based customers, but Nvidia said those licenses produced only about $60 million in revenue. In February 2026, the company received a license for small amounts of H200 chips to specific China-based customers, with no revenue from that program as of the filing.
So when Huang gets the president on his phone, investors should pay attention. The difference between a license, a stalled license and a reopened market is not theoretical. It's revenue.
Washington still has a process, at least on paper #
Trump has been publicly warm toward Huang. The American Presidency Project's transcript of a White House event in April 2025 shows Trump calling him one of the most brilliant people you could meet, and Huang thanked Trump for policies that helped speed Nvidia's U.S. manufacturing plans. The Washington Post later reported that Huang praised Trump as America's unique advantage in AI and said he was proud to contribute to Trump's White House ballroom project. CBS News separately reported that Nvidia was among donors tied to the ballroom effort.
Frankly, this is the part that should make you uneasy even if you like the company. Transformer News reported in March that a White House official described Huang as an example of how informal lobbying works in Trump's second term, where access to the president can leave normal lobbying channels and internal policy fiefdoms looking like an afterthought. That doesn't prove any rule was broken. It does show why the arrangement draws attention.
Compare it with Apple. Tim Cook has spent years working Washington through dinners, public praise, domestic investment pledges and patient relationship-building. Elon Musk has his own route into Trumpworld. But Nvidia's issue is narrower and more combustible: can it sell specific AI chips into China, in what volume, and under what conditions? That question belongs to national security agencies, Commerce Department export controls and the White House. Huang's access makes those lines look blurred.
If you own Nvidia, the read-through is uncomfortable no matter how bullish you are on AI demand. Wired's all-hands story arrived just after Nvidia reported quarterly revenue of $96 billion, up 106% from a year earlier, and Trump congratulated Huang publicly later the same day. The company is huge enough to shape the AI market, but one of its most sensitive growth questions still turns on Washington's mood. Nvidia is now bankrolling the company that buys its chips and the numbers are getting hard to ignore
Nvidia is in talks to guarantee up to $250 billion for OpenAI's data center buildout in Ohio, with separate discussions covering $350 billion more in chip financing. The deal reignites circular-financing concerns as Nvidia's total AI deal exposure crossed $750 billion last week. - nvidia financing openai infrastructure deals - gpu supplier funding data center buyers
That's the risk. A policy path built around a CEO's personal access can work beautifully until it doesn't. Wired's reporting doesn't say what Trump wanted during the all-hands call. It doesn't need to. The important fact is that the call came through while Huang was standing in front of his own employees, and everybody in the room could hear who was on the other end.
Also read: OpenAI Now Lets You Link Multiple Gmail Accounts to ChatGPT • OpenAI Will Cut Off Cursor's Access to Its Models on November 12 • OpenAI's Price Cut On GPT-5.6 Sol Sent OpenRouter Usage Up Nearly 14 Times