Nvidia pitches Vera CPU to Chinese clients, targeting $200 billion market

Nvidia is taking orders from Chinese customers for its new Vera server CPU, with shipments potentially starting in August, Reuters reported, marking a fresh attempt to rebuild business in China after Chief Executive Jensen Huang said last October that the company’s market share there had fallen to effectively zero. The push comes as U.S. export controls have blocked Nvidia’s most powerful AI accelerators from reaching China, steering the company toward a product category that currently faces fewer restrictions. Vera is Nvidia’s first standalone central processor and is designed for computing tasks tied to AI agents, including database queries and code compilation. Nvidia says the Arm-based chip is 1.8 times faster than comparable x86 processors on workloads linked to autonomous AI systems. At least one major Chinese cloud provider plans to order more than 300 servers with two Vera chips each for initial testing, though large-scale adoption remains uncertain because of software compatibility challenges, the difficulty of migrating workloads built around domestic chips, and pricing that SemiAnalysis estimates at more than $20,000 per processor and about $10 million for a 256-chip rack depending on memory configuration. The effort also highlights the shifting economics of AI infrastructure as the industry moves from training large models toward inference, a phase that increases the importance of CPUs and custom silicon alongside GPUs. Nvidia is expecting $20 billion in Vera revenue by the end of its current fiscal year, which closes in January 2027, while geopolitical risk remains elevated because Washington could tighten controls if the chips are judged to support advanced AI development in China and Beijing continues to promote domestic semiconductor alternatives.