Inside the Huang Effect: What Really Happened in Taipei?

Last week, a confidential gathering between NVIDIA’s Jensen Huang and 12 elite executives from TSMC, Foxconn, Quanta, and top U.S. VCs in Taipei’s Grand Hyatt turned into a watershed moment—when sensitive remarks allegedly made behind closed doors (no phones, no recording) surfaced in major outlets within hours.

The core claims? Huang reportedly stated that China will dominate the next 5-10 years of AI development, pointing to over 1 million Chinese AI workers vs. 20,000 in Silicon Valley. He allegedly warned that U.S. export sanctions are counterproductive, accelerating Chinese self-sufficiency rather than slowing it. Most striking: he flagged Huawei’s Ascend 910C as only 8-12% slower than NVIDIA’s H100, with production hitting 200,000 units monthly—and predicted China would have more total AI computing power than the rest of the world combined by 2027.

The real question isn’t whether these words were spoken—it’s why they were leaked.

Read between the lines: A strategic “leak” accomplishes multiple objectives simultaneously. It signals to Washington that continued restrictions may backfire (pressuring policy). It subtly repositions NVIDIA as a rational actor analyzing market realities, not a Trump loyalist. And it creates a public record of prescient warnings, protecting the company’s credibility regardless of how geopolitics unfold.

Or Huang got outplayed—and someone wanted the world to know NVIDIA’s internal calculus on the race.

Either way, the leak reveals what Silicon Valley won’t say publicly: the AI competition with China isn’t slowing down, and Washington’s toolbox of leverage might be smaller than advertised. The next move belongs to policymakers—and they’re running out of time to respond.