Key Takeaways
- Bank of America estimates Nvidia’s Vera CPU platform could deliver approximately $20 billion in revenue during H2 fiscal year 2027
- Initial Vera CPU shipments may reach 4–5 million units across the first two quarters post-launch, priced at $4,000–$5,000 each
- The rise of agentic AI and reinforcement-learning applications is creating increased demand for CPU-based infrastructure
- U.S. government official stated that only a “trivial” quantity of H200 processors has been exported to China despite approved licenses
- Chinese market revenue for Nvidia plummeted 53% to $4.55 billion in the latest quarter, contrasting with companywide revenue growth of 85% to $81.6 billion
Nvidia (NVDA) stock declined approximately 0.37% during Wednesday’s premarket session to $211.01, as investors processed divergent signals from the semiconductor giant despite strength in broader technology indices.
Two distinct storylines are commanding market attention this week — one spotlighting Nvidia’s emerging revenue frontier, the other highlighting geographic challenges.
According to Bank of America’s recent semiconductor analysis authored by analyst Vivek Arya, Nvidia’s most significant upcoming growth catalyst may not originate from its graphics processing unit segment. Instead, the opportunity lies in central processing units.
As artificial intelligence applications evolve from training massive language models toward deploying autonomous agents throughout enterprises, computational requirements are transforming. Graphics processors powered the initial AI boom. Central processors, particularly Nvidia’s Vera architecture, could fuel the subsequent expansion phase.
Bank of America’s research suggests Nvidia is positioning itself to capture roughly $20 billion in Vera CPU revenue throughout the latter half of fiscal 2027. Approximately 50% of that figure is anticipated from processors supporting existing GPU installations, while the remainder would stem from dedicated CPU infrastructure designed specifically for agentic AI and reinforcement-learning applications.
The volume forecasts are noteworthy. Nvidia may deliver 4 to 5 million Vera processors during only the initial two quarters following commercial availability — a stark contrast to the 2.5 million Grace CPUs shipped cumulatively through the present. Pricing is projected to range between $4,000 and $5,000 per unit.
To provide perspective, Nvidia’s Data Center compute segment generated $60.4 billion during Q1, representing approximately 74% of the company’s record-breaking $81.6 billion quarterly performance. Graphics processing remains the dominant revenue contributor. However, BofA’s analysis indicates the CPU market represents a substantial emerging opportunity.
China: Limited Access Under Strict Oversight
Regarding the Chinese market, Jeffrey Kessler, undersecretary of commerce for industry and security, informed Bloomberg on Tuesday that merely a “trivial” volume of Nvidia H200 processors has reached Chinese customers operating under U.S.-authorized licenses.
Kessler explained that applicants must satisfy stringent national security criteria, including implementing protections to ensure chips don’t reach Chinese military applications. He refrained from disclosing specific purchaser identities or exact shipment quantities.
The H200 became eligible for Chinese export following President Trump’s December authorization. The Commerce Department subsequently established formal regulations requiring verified purchasers and national security compliance.
Among confirmed authorized buyers, Reuters reported that ZTE Kangxun Telecom and Maginfra have obtained approval to acquire advanced processors from both Nvidia and AMD.
Nevertheless, market access remains constrained. Nvidia has communicated to investors that it anticipates minimal near-term AI processor revenue from China, partially because Beijing is aggressively encouraging domestic buyers to source chips from local manufacturers.
What the Numbers Say
Nvidia’s Chinese market revenue decreased 53% year-over-year to $4.55 billion in Q1, declining from $9.66 billion during the comparable prior-year period. Overall company revenue nonetheless surged 85% to achieve a record $81.6 billion.
NVDA has advanced roughly 9% year-to-date but has experienced negative returns over the trailing month.
Nvidia maintains its expectation for negligible additional near-term AI processor revenue from China, despite the H200 export authorization now being implemented.





