Key facts
- Nvidia is actively pitching its new Vera central processing unit (CPU) to Chinese clients for AI data centers.
- The company suggests that orders can commence soon, with potential availability as early as August.
- The Vera chip is designed for agentic AI, systems that perform tasks autonomously.
- Nvidia claims the Vera chip offers up to 1.8 times faster performance than comparable rival processors.
- A single Vera processor is priced at over $20,000 before bulk discounts.
- Nvidia projects $20 billion in revenue from Vera chip sales by the end of its fiscal year.
Nvidia is initiating sales pitches for its new "Vera" central processing unit (CPU) to clients in China, signaling a strategic pivot to regain market share amid stringent U.S. export controls. The company has informed potential customers that orders can be placed soon, with availability anticipated as early as August. This move comes as Nvidia's shipments of its H200 AI chip to China have stalled, contributing to a significant decline in its market share within the country.
The Vera chip is Nvidia's first standalone CPU designed specifically for agentic AI, which enables autonomous task performance. Nvidia claims it offers a substantial performance advantage, running up to 1.8 times faster than comparable processors from competitors like Intel and AMD. The company aims for Vera to become a multibillion-dollar business, with leading cloud providers such as Alibaba and ByteDance previously announced as collaborators.
Initial client interest has been noted, with one major Chinese cloud company planning to order over 300 servers equipped with Vera CPUs for testing purposes. However, widespread adoption may face hurdles related to software ecosystems, compatibility, and the migration of workloads currently built around domestic AI chips. The pricing for a single Vera processor is expected to exceed $20,000 before bulk discounts, with a fully configured rack costing around $10 million.
Nvidia projects $20 billion in revenue from Vera chip sales by the end of its fiscal year. This development occurs as the global AI landscape shifts from model training to inference computing, where CPUs are increasingly competitive with GPUs. Both Intel and AMD have reported extended lead times for their server CPUs, indicating a tight market.