Most companies can’t shrug off an $8 billion loss. Then again, most companies aren’t Nvidia (NVDA).
On Wednesday’s first-quarter earnings call, Wall Street zeroed in on Nvidia’s obvious weak spot: China. Thanks to U.S. export restrictions, Nvidia’s custom-built H20 chips, designed to skirt earlier rules, have essentially been made worthless. Nvidia CFO Colette Kress confirmed the damage: “Had the export controls not occurred, we would have had orders of about $8 billion for H20” in the quarter.
That $8 billion is more than rival AMD’s (AMD) total quarterly revenue — and Nvidia seems ready to just shrug off the loss
The tech giant reported another monster quarter: Revenue soared 69% year-over-year to $44.1 billion, while net income grew even faster. The stock popped more than 4.5% on Thursday, and Nvidia, with a $3.4 trillion market cap, is the most valuable company in the world — just ahead of Microsoft (MSFT) and now well ahead of Apple (AAPL).
Nvidia’s data center business alone grew 73% year-over-year and brought in over $39 billion. And with its next-generation Blackwell chips ramping up, inference workloads scaling globally, and governments racing to build AI capacity using Nvidia infrastructure, China has become — if not an afterthought — then certainly just one node in a much larger empire.
Nvidia has bigger chips to fry.
“This is the start of a powerful new wave of growth,” CEO Jensen Huang said.
Nvidia’s global ambitions are heating up
Still, Nvidia isn’t abandoning China.
According to a recent report, the company is designing a modified Blackwell AI chip specifically for the Chinese market that will comply with U.S. export restrictions. Huang didn’t confirm this directly — he said Nvidia doesn’t “have anything at the moment, but we’re considering it.”
He did, however, address the broader implications. The question, he said, isn’t whether China will have AI — “it already does” — but whether one of the world’s largest AI markets will run on American platforms. Export controls, in his view, should “strengthen U.S. platforms, not drive half of the world’s AI talent to rivals.”
Kress made the financial stakes clear: Nvidia expects a “meaningful decrease” in China data center revenue in the second quarter. Losing access to what it believes will be a $50 billion AI accelerator market would, she said, have “a material adverse impact” on the business and benefit rivals in China and beyond.
As for the politics, Huang said that President Donald Trump “has a plan, and I respect it.”
But the bigger picture is that Nvidia isn’t betting on any single market. It’s betting on a global AI infrastructure boom that’s only just getting started — what Huang called “sovereign AI.”
The company’s strategy isn’t about selling to tech firms anymore. Entire countries could soon become Nvidia customers.
“Nations are investing in AI infrastructure like they once did for electricity and internet,” Huang said. “Sovereign AI is a new growth engine for Nvidia.”
“Every nation now sees AI as core to the next industrial revolution, a new industry that produces intelligence and essential infrastructure for every economy,” he added. “Countries are racing to build national AI platforms to elevate their digital capabilities.”
Nvidia’s recent moves show how seriously it’s playing this global game. At Computex, Huang announced Taiwan’s first AI factory, a collaboration with Foxconn (HNHPF) and the Taiwanese government. Shortly after, Huang was present as Sweden launched its inaugural national AI infrastructure.
But perhaps most notable is Nvidia’s opportunity in the Middle East.
Huang was there for Trump’s tour through the region, where the president announced a 500-megawatt AI infrastructure project in Saudi Arabia and a 5-gigawatt AI campus in the United Arab Emirates. Wedbush Securities analysts called it a “watershed moment” for Big Tech that could be a $1 trillion opportunity.
Nvidia is positioned to capitalize on that global wave of AI infrastructure investment. As Huang said, this includes everything from powering cutting-edge training models to overhauling $500 billion of enterprise IT with its AI-ready systems. Add in industrial AI — robotic factories, digital twins, and more — and the company’s footprint keeps expanding.
“Just about every country needs to build out AI infrastructure,” Huang said. “There are umpteenth AI factories being planned.”
Nvidia’s next big bet
Behind all of this is a bigger technological shift — from generative AI to what comes next.
“Reasoning AI really busted through,” Huang said, referring to AI agents that can make decisions, use tools, remember, and plan. “Reasoning AI agents require orders of magnitude more compute.”
He added that the demand for AI reasoning is up, and “we would like to serve all of it, and I think we’re on track to serve most of it.” That shift — from model training to full life-cycle AI infrastructure — is fueling demand. Huang said Nvidia’s Grace Blackwell NVLink72 is “the ideal engine today, the ideal computer thinking machine, if you will, for reasoning AI.” And agentic AI, which takes reasoning a step further by acting autonomously, is making waves, too.
“Enterprise AI is ready to take off,” Huang said. “Much more than generative AI, agentic AI is game-changing.”
CFO Kress said the transition from generative to agentic AI will “transform every industry, every company, and country.” She said Nvidia envisions AI agents as a “digital workforce capable of handling tasks ranging from customer service to complex decision-making processes.” Huang said that Nvidia is in the “beginning of the build-out” — and that “there should be many, many more announcements in the future.”
More AI means more chips. More chips need bigger clusters. Bigger clusters demand faster networking.
And Nvidia is scaling every layer of that stack.
Huang called Spectrum-X — the company’s AI-optimized Ethernet platform — “a home run,” noting that it added two major cloud providers in the past quarter. The platform can effectively turn $10 billion AI clusters into $14 billion ones through increased efficiency. And NVLink is also becoming foundational to the AI data center — it exceeded $1 billion in the April quarter. Now analysts see networking as a meaningful second leg of Nvidia’s AI infrastructure business.
The U.S. export controls don’t appear to be going anywhere, especially as tense — and tech-heavy — trade talks between the U.S. and China continue. But Nvidia seems prepared for what’s ahead. The company just lost billions in China, and that might have been the least interesting thing about its quarter.
As Nvidia continues to navigate a geopolitical minefield, it’s not that China doesn’t matter — it’s that everything else now matters more.
As Huang said: “We’re off to the races.”