What happened when Nvidia re entered the Chinese market by shipping a small number of H200 AI chips during its fiscal second quarter— including which Chinese companies were authorized to buy them and under what limits, why deliveries were delayed and remained far below the permit

Nvidia’s China re-entry was commercially minor: it shipped only a small number of H200 AI chips in fiscal Q2, while political resistance in Beijing kept sales far below the U.S.-licensed allowance. The episode contrasted with an otherwise strong quarter that exceeded Wall Street expectations and prompted a rebound in after-hours trading. 23

China H200 sales

• Available evidence indicates that ByteDance and Tencent were authorized buyers, with limits reportedly of up to 75,000 H200s per company; small shipments of about 10,000 units each were reported. The precise authorization list and shipment figures are not independently substantiated by the higher-authority sources supplied here, so treat them cautiously. 6
• Deliveries were delayed and volume remained below the permitted amount because Chinese authorities objected to or restricted purchases despite the U.S. export authorization. 2
• H200 sales to Chinese customers accounted for less than 1% of Nvidia’s fiscal-Q2 data-center revenue. 2
• Nvidia recorded a $400 million charge over the preceding six months for excess H200 inventory because demand for the China-targeted product had diminished amid those restrictions. 2

Outlook and competition

• Nvidia’s fiscal-Q3 outlook assumed no China data-center-compute revenue, reflecting continuing geopolitical uncertainty rather than a return to normalized China sales. 2
• The China shortfall also came as Nvidia faced strategic pressure from customers developing in-house AI chips and as production capacity and attention shifted toward the next-generation Vera Rubin platform; reporting later indicated Nvidia stopped China-bound H200 production and redirected TSMC capacity toward Rubin. 5
• The longer-term implication was that an uncertain, lower-volume H200 China opportunity was less important than successfully scaling Rubin and defending Nvidia’s position as large customers seek alternatives.

Broader earnings and market response

• Fiscal-Q2 revenue was $96.2 billion, up 106% year over year, while data-center revenue reached $89.0 billion, up 117%. 7
• Revenue beat the roughly $92.2 billion analyst forecast, and Nvidia guided fiscal-Q3 revenue to $108 billion, also above expectations. 8
• Investors initially sent the stock lower after the release, but shares subsequently rose in off-hours trading; one report described an after-hours gain of more than 4%. 36
• Nvidia had beaten Wall Street expectations for the 15th consecutive quarter, so the modest China sales did not outweigh the much larger demand story in its core data-center business. 3