Nvidia Corp. CEO Jensen Huang dismissed reports of delays in the rollout of the company’s Vera Rubin AI hardware, affirming that production remains on track with 'giant' volumes. Speaking on Wednesday in Tokyo regarding the company’s role in Japan’s AI ambitions, Huang stated that high-end AI accelerator systems are on track for delivery. However, he did not specify a given timeframe for the deliveries.

Huang’s latest confirmation addresses recent market concerns. KeyBanc analyst John Vinh stated on Tuesday that Nvidia’s Vera Rubin rollout could be slightly delayed due to thermal heat lid issues in SK Hynix Inc.’s HBM4 memory qualification. Despite this, Vinh noted minimal risk to financial estimates, expecting Nvidia to offset the delay by shipping more B300 GPUs instead of R200 chips. KeyBanc raised its price target on the stock to $330 from $310, citing stronger-than-expected demand and a significant increase in its 2027 CoWoS supply forecast to 1.1 million interposers.

Earlier in June, Huang had stated that the Vera Rubin accelerator was in full production. The platform incorporates high-bandwidth memory (HBM) from Samsung Electronics Co Ltd., SK Hynix Inc., and Micron Technology Inc.

Reported Setbacks in Key Deliveries

Research firm SemiAnalysis reported earlier this month that Nvidia’s Kyber NVL144 faced 'major setbacks' and would be delayed by over 12 months, pushing it to 2028. The firm also stated that Nvidia’s NVL72x2 back-to-back rack architecture was 'cancelled,' leaving Rubin Ultra with a 'limited scale-up domain.'

Analyst Outlook and Supply Chain

The following table summarizes the key analyst updates and supply chain details mentioned in the reports:

Nvidia Corp has officially begun shipping its powerful H200 artificial intelligence chips to China, signaling a resumption of advanced hardware exports under strict case-by-case government licenses. A top U.S. trade official confirmed the development, while KeyBanc maintained an Overweight rating on the stock and raised its price target to $330 from $310, citing robust global demand and supply chain progress. The regulatory approvals and analyst optimism highlight Nvidia's strategic position despite geopolitical tensions and hardware rollout delays.

H200 Shipments Remain Limited

Jeffrey Kessler, Under Secretary of Commerce for Industry and Security, stated that early volumes remain heavily restricted, describing initial deliveries as "trivial" and "very few". Despite tight restrictions, recent U.S. approvals have cleared around 10 Chinese firms to receive advanced AI hardware from Nvidia and Advanced Micro Devices Inc. Major tech conglomerates—including ZTE Kangxun Telecom, a unit of ZTE Corp, Alibaba Group Holding Ltd ADR, Tencent Holdings Ltd, ByteDance, Maginfra, and a Kingsoft subsidiary—are among the entities cleared to purchase the chips. Kessler emphasized that applicants must satisfy rigorous national security benchmarks and submit to inspections to guarantee compliance.

China Revenue Declines Amid Policy Hurdles

Nvidia sharply reduced its China exposure over the past year, even as its overall AI-driven revenue surged. Revenue from customers based in China, including Hong Kong, fell about 53% to $4.55 billion in the first quarter from $9.66 billion a year earlier, according to Nvidia's Form 10-Q. Despite that drop, total revenue jumped 85% year over year to a record $81.6 billion. Chinese authorities have slowed local purchases of U.S.-designed AI chips, partly because Beijing wants to support its domestic chip industry. Nvidia has told investors it does not expect additional near-term AI processor revenue from China, even after U.S. approval for H200 sales.

KeyBanc Raises Nvidia Forecast

KeyBanc analyst John Vinh described the company's supply landscape as "mixed but mostly positive." While Asia field checks reveal that the upcoming Vera Rubin chip architecture faces slight production delays due to thermal heat lid issues and HBM4 qualification delays with SK Hynix Inc ADR, KeyBanc foresees "minimal risk to estimates." To mitigate the delay, Nvidia is expected to ship more B300 GPUs in place of its R200 models. Vinh expects Nvidia to ship 5.5 million to 6 million Blackwell GPUs this year, along with 1 million Hopper GPUs. He also expects Nvidia to ship 70,000 to 80,000 total racks this year, including 5,000 to 6,000 Vera Rubin racks.

CoWoS Supply Supports AI Demand

The firm highlighted a significant increase in its supply forecast. Nvidia’s 2026 CoWoS supply outlook remains unchanged at 650,000 interposers, while 2027 supply has been revised significantly higher to 1.1 million interposers. This increase reflects strong demand and a full-year Rubin ramp. Vinh said Nvidia remains uniquely positioned to benefit from secular growth in data center AI and machine learning, pointing to Nvidia’s CUDA software stack as a major barrier to entry.