NVIDIA Rising in the AI and Data Center Markets, Despite U.S. Export Bans
The latest earnings call shows continued growth and robust demand internationally
Last Wednesday was the earnings call for NVIDIA. Nvidia beats on earnings and revenue as data center sales jump 73% (CNBC)
CNBC reports on this “booming data center business.”
NVIDIA expects about $45 billion in sales in the current quarter, but it said its guidance would have been about $8 billion higher except for lost sales from a recent export restriction on its China-bound H20 chips.
Reports indicate the possibility of restricted chips entering China through Singapore, NVIDIA’s second largest market, with $2 billion is sales. This might mandate the addition of export-compliant trackers in the H20 chips.
On the same day as the earnings call, Bloomberg Technology interviewed Jensen Huang, whose reaction to the chip export ban to China was: “China will carry on with AI, with or without NVIDIA. But, the demand remains strong for NVIDIA’s tech.
The Hopper GPU microarchitecture was used to produce the H20 chip, a modified version of the H100, to produce a fundamental design with the lowest level of performance in an attempt not to trigger the export ban, without success so far. As of 2025, the H20 had become the most prominent chip in China.
When asked if Huang would produce a chip compliant with the export ban, he said “H20 is as far down as we can take Hopper.” He added data center chips don’t have to be small. His competitor for China, Huawei has tech comparable to an H200. He also noted Huawei’s CloudMatrix scales up something larger than Grace Blackwell (see below).
If China can’t rely on U.S. stacks, the must use Huawei, he added. China has the second largest AI market, the largest population of AI researchers, so he didn’t want to have to abandon this market. He mentioned lobbying efforts with the U.S. government to relax the ban for NVIDIA’s tech.
Huang presented an optimistic view of the big picture demand, which should bolster the prospects of other tech companies such as AMD, ARM, Broadcom, Intel, Micron, Super Micro, Dell, Qualcomm, Marvell, et al.
The interviewer noted NVIDIA’s CFO Colette Kress, and her mention of the customers separate from the hyperscalers. These include healthcare, industrial, robotics, consulting firms, and other sectors. Apart from the hyperscalers, Huang spoke of Elon Inc, and xAI, with projects such as Optimus, Omniverse, Grok, which he considered “all world class.” He said human robots deployed into the world would be “the next multi-trillion dollar industry.”
When asked about his views of the current federal administration’s vision, he supported the use of tariffs to promote on-shore manufacturing. Huang talked about his partnership with TSMC in an Arizona buildout.
He also encouraged the government’s rescinding of the Biden-era AI Diffusion Rule, so the world can build on U.S. stacks.
Analysts were compelled to ask what made up for the China export shortfall. Did Blackwell do $3 billion above expectations, filling the shortfall?
Huang mentioned that “Blackwell is just a homerun. NVLink is a homerun architecture.” (NVIDIA touts Grace Blackwell (GB200) NV72 as delivering 30x faster LLM inference.)
The interview ended with Huang talking about his upcoming European trip. He sees AI as being part of the national infrastructure for every country. Huang will deliver the GTC Paris keynote at VivaTech, Europe’s biggest startup & tech event, on June 11th.
I decided to pick up this recently released book: “The Thinking Machine: Jensen Huang, Nvidia, and the World's Most Coveted Microchip,” by Stephen Witt.
The blurb on Amazon.com begins: “In June of 2024, thirty-one years after its founding in a Denny’s restaurant, Nvidia became the most valuable corporation on Earth.”
This biography of Jensen Huang, is an immigrant’s story of someone who excelled scholastically, then in his working life took on risky projects which could have ruined his career, but he was always able to deliver. My complete review will be forthcoming.