Key facts
- Nvidia is pitching its new Vera CPU to Chinese clients, with potential availability in August.
- The Vera chip is designed for agentic AI, performing tasks autonomously.
- Nvidia expects Vera chip sales to reach $20 billion by the end of its fiscal year.
- U.S. export controls have impacted Nvidia's AI chip shipments to China.
- A major Chinese cloud company plans to order over 300 servers equipped with two Vera CPUs each for testing.
Nvidia has begun actively marketing its new "Vera" central processing unit (CPU) to clients in China, signaling a strategic pivot to revive its market presence in the country. The company has informed potential customers that the Vera chip, designed for autonomous AI agents, could be available as early as August and that they can commence placing orders. This initiative comes as Nvidia faces significant challenges in China due to U.S. export controls on advanced AI chips, which have stalled shipments of its H200 GPU.
Nvidia's CEO, Jensen Huang, previously stated that the company's market share in China had effectively dropped to zero. The introduction of Vera, built on Arm technology, positions Nvidia in direct competition with established CPU giants Intel and AMD, both of which are also racing to supply server CPUs for AI data centers. Nvidia claims Vera runs up to 1.8 times faster than comparable processors from its rivals and anticipates it will become a multibillion-dollar business, projecting $20 billion in revenue from Vera sales by the end of its fiscal year in January.
Some Chinese clients have expressed interest, with one major cloud company planning to order over 300 servers, each containing two Vera CPUs, for initial testing. The success of large-scale adoption, however, may depend on software ecosystem compatibility and the challenges of migrating workloads from domestic AI chips. Pricing for a single Vera processor is expected to be over $20,000 before bulk discounts, with a full rack of 256 chips costing around $10 million. The global AI race is shifting towards inference computing, where CPUs like Vera face increased competition from GPUs and custom chips, contributing to a current CPU shortage.