Nvidia’s GPU market share in China? According to Jensen Huang: roughly zero. US export controls have effectively locked the chip giant out of the world’s second-largest AI market. But now Nvidia has found a new way in — and it doesn’t go through GPUs.
On June 12, it became known that Chinese customers can already place orders for the Vera processor. Shipments could begin as early as August.
What Is Vera?
Vera is Nvidia’s first standalone CPU — an 88-core ARM-based processor designed specifically for agentic AI workloads. It’s part of the Vera Rubin platform, which pairs the CPU with Rubin GPUs. But as a standalone CPU, it doesn’t fall under the same export restrictions as Nvidia’s high-end GPUs.
That’s the crucial point: selling CPUs in China is far less fraught than selling GPUs. And Nvidia is now exploiting exactly that gap.
First Customers, Big Ambitions
A major Chinese cloud provider is already planning to purchase over 300 servers. Initial deployment will be in data centers outside China — for testing. Nvidia expects $20 billion in revenue from Vera chip sales by the end of its fiscal year in January 2027.
This puts Nvidia in direct competition with Intel and AMD, who have long dominated the CPU market with x86 architecture. With ARM, Nvidia is taking a different path — and with China sales, one that’s politically less charged than the GPU route.
Why It Matters
The AI infrastructure story often gets reduced to GPUs. But agentic systems — AI agents that act, plan, and coordinate autonomously — also need massive CPU power. Vera is Nvidia’s answer to exactly that demand.
And for China, it’s a signal: Nvidia isn’t giving up the market. It’s just looking for a different entrance.
Sources: