$8 billion. That’s the kind of revenue hit Nvidia just said “hello darkness” to, thanks to the U.S. government’s increasingly aggressive chip export restrictions.
As Nvidia closed the books on Q1 of its fiscal 2026 (ending April 28), it pulled back the curtain on how deeply the Trump-era licensing requirements have impacted it. The company reported a staggering $4.5 billion write-off due to its inability to ship its H20 AI chips to China. To make things even more dramatic, another $2.5 billion in potential sales had to be left on the table in Q1 alone.
And buckle up—because Q2 isn’t looking any friendlier. Nvidia now expects the same licensing hurdles to cost it another $8 billion, eroding what would have otherwise been a $45 billion quarter. That’s nearly a fifth of projected revenue. Gone.
So what’s the hold-up? In short: the H20 AI chip. Nvidia had developed the chip to work within export restrictions, but the U.S. doubled down. Now, even the H20 is too advanced to get clearance, leaving the Chinese data center market (and $50 billion in AI spending) out of reach for Nvidia.
CEO Jensen Huang didn’t sugarcoat it. On the company’s earnings call, he called China “one of the world’s largest AI markets” and “a springboard to global success.” But with the H20 officially banned, Huang admitted their Hopper-based data center business in China is effectively dead in the water.
While Nvidia can’t nerf the Hopper architecture enough to meet restrictions, Huang did give a rare golf clap to the Biden administration for scrapping a proposed rule that would’ve made things even worse.
In the meantime? Nvidia is searching for new ways to play the game in China. But right now, it’s benched—and watching $8 billion slip through its fingers.
