Nvidia has shipped its first H200 artificial intelligence data-center processors to China under a newly approved U.S. licensing arrangement, marking a limited reopening of one of the world’s biggest AI markets for the chipmaker. However, the contribution remains extremely small: Nvidia said H200 sales to China accounted for less than 1% of its $89 billion Data Center revenue in the second quarter of fiscal 2027.
Despite the first shipments, Nvidia is still taking a conservative approach to China. The company has forecast $108 billion in third-quarter fiscal 2027 revenue without assuming any Data Center compute revenue from China. The cautious outlook reflects continued uncertainty over Chinese regulatory approvals, U.S. export restrictions and Beijing’s push to encourage domestic AI-chip alternatives.
Nvidia H200 China Shipments At A Glance
The first deliveries represent a policy shift, but China remains a relatively small part of Nvidia’s current Data Center business.
Key Details
| Particular | Details |
|---|---|
| Chip | Nvidia H200 |
| Market | China |
| First shipments | Q2 FY2027 |
| China contribution to Data Center revenue | Less than 1% |
| Nvidia Q2 Data Center revenue | $89 Bn |
| Nvidia Q3 revenue guidance | $108 Bn ±2% |
| China Data Center revenue assumed in Q3 guidance | $0 |
| Known early recipients | ByteDance, Tencent |
| Reported initial shipment | ~10,000 H200s each |
| Key competitors in China | Huawei and other domestic chipmakers |
Nvidia’s latest disclosure marks a significant change from the previous quarter, when the company reported no shipments of Data Center Hopper products to China.
Nvidia Ships H200 Chips To China For The First Time
The H200 shipments follow months of uncertainty over whether Nvidia would be able to resume meaningful sales of advanced AI processors in China.
Washington approved licenses for selected Chinese customers, but Beijing initially resisted purchases as it promoted domestically developed alternatives.
Nvidia has now confirmed that some H200 shipments took place during the latest quarter.
The company nevertheless said the amount sold was small.
U.S. Export Approval
↓
Chinese Regulatory Review
↓
Limited Customer Approvals
↓
H200 Shipments Begin
↓
Less Than 1% Of Nvidia
Data Center RevenueThe development therefore represents a reopening rather than a full return to the Chinese market.
H200 China Revenue Is Less Than 1%
Nvidia’s Data Center business generated $89 billion in revenue in Q2 FY2027, up 117% from a year earlier.
H200 sales to China accounted for less than 1% of that figure.
That means China’s current contribution remains relatively modest compared with Nvidia’s overall AI business.
Nvidia Data Center Revenue
| Metric | Q2 FY2027 |
|---|---|
| Data Center revenue | $89 Bn |
| YoY growth | 117% |
| Sequential growth | 18% |
| China H200 contribution | <1% |
The contrast is important: Nvidia’s global AI business is expanding rapidly even while its access to China remains heavily restricted.
Why Nvidia Is Still Excluding China From Its Forecast
Nvidia’s third-quarter guidance is $108 billion, plus or minus 2%.
However, the company is assuming no Data Center compute revenue from China in that forecast.
This conservative approach reflects the uncertainty surrounding future approvals.
Nvidia’s Q3 Outlook
| Metric | Guidance |
|---|---|
| Revenue | $108 Bn ±2% |
| Potential revenue range | ~$105.8–$110.2 Bn |
| China Data Center compute revenue | Not assumed |
| Gross margin | ~74% ±0.5 pts |
Nvidia is therefore treating any additional China sales as potential upside rather than as a dependable component of its base forecast.
ByteDance And Tencent Receive H200 Chips
Earlier reports indicated that ByteDance and Tencent had each received approximately 10,000 H200 processors in recent weeks.
Other Chinese technology companies could also receive similar shipments, although approvals remain subject to regulatory conditions.
Reported Initial Shipments
| Company | Reported H200 Shipment |
|---|---|
| ByteDance | ~10,000 |
| Tencent | ~10,000 |
| Total for two companies | ~20,000 |
These numbers should be treated as reported shipment estimates rather than Nvidia’s official customer-by-customer disclosure.
Why Chinese AI Companies Still Want Nvidia Chips
China has been investing heavily in domestic AI accelerators, particularly from companies such as Huawei.
Nevertheless, Nvidia’s hardware remains attractive because of its mature software ecosystem.
One of the biggest advantages is CUDA, Nvidia’s software platform used extensively across AI development and training.
For large AI companies, switching from an established Nvidia-based infrastructure to a different accelerator platform can require substantial engineering work.
Nvidia GPUs
↓
CUDA
↓
AI Frameworks
↓
Existing Models + Software
↓
Training Infrastructure
↓
AI ProductsThis creates significant switching costs for organizations that have spent years optimizing their AI systems around Nvidia hardware.
H200 Is An Older Generation, But Still Powerful
The H200 belongs to Nvidia’s Hopper generation and is not the company’s newest AI accelerator.
Nvidia has since moved to the Blackwell family and is now ramping its next-generation Vera Rubin platform.
However, the H200 remains a powerful processor for AI training and inference.
Its availability in China therefore gives Chinese AI companies access to Nvidia’s established ecosystem even though newer Nvidia architectures remain subject to tighter restrictions.
Nvidia AI Architecture
| Generation | Status |
|---|---|
| H100 | Earlier Hopper generation |
| H200 | China-approved limited shipments |
| Blackwell | Current advanced generation |
| Vera Rubin | Next-generation platform |
The gap between H200 and Nvidia’s latest products also means Chinese customers are not receiving the company’s most advanced computing technology.
China Is Trying To Build Its Own AI Chip Ecosystem
Beijing has been encouraging Chinese companies to use domestic processors instead of relying on foreign technology.
Huawei has emerged as one of the most important competitors in this effort.
The Chinese government has incentives to reduce dependence on Nvidia because advanced AI accelerators are strategically important for both commercial and national-security applications.
China’s AI Chip Strategy
U.S. Export Restrictions
+
Chinese Technology Policy
↓
Reduced Nvidia Access
↓
More Demand For Domestic Chips
↓
Huawei + Chinese AI Chipmakers
↓
Development Of Local EcosystemThe limited H200 approvals therefore represent a balancing act between immediate AI-computing requirements and China’s longer-term semiconductor ambitions.
Beijing’s Position Creates A Complicated Market
The situation is unusual because both Washington and Beijing have influence over Nvidia’s ability to sell H200 chips in China.
The U.S. controls whether Nvidia can legally export the technology.
China controls whether domestic companies can import and deploy it.
This creates a two-stage approval process.
Nvidia
↓
U.S. Export License
↓
Chinese Buyer
↓
Chinese Regulatory Approval
↓
Mainland DeploymentEven after receiving U.S. approval, Chinese customers may not necessarily receive permission to deploy the chips on the mainland.
Hong Kong Provides Another Route
Chinese regulators have reportedly allowed some companies to ship H200 processors to Hong Kong, which operates as a separate customs territory from mainland China.
However, Hong Kong’s ability to absorb large amounts of AI computing capacity is constrained by available data-center infrastructure and power.
This limits how useful Hong Kong can be as a substitute for mainland deployment.
Nvidia Previously Had No H200 China Revenue
The latest shipment disclosure is particularly important because Nvidia had previously said it had generated no revenue from H200 sales to China.
In May, CFO Colette Kress said Nvidia had not generated revenue under the H200 licensing program and remained uncertain whether imports would be allowed.
The company has now moved from uncertainty to actual shipments.
Nvidia’s China H200 Timeline
| Period | Development |
|---|---|
| Early 2026 | U.S. licensing begins |
| May 2026 | Nvidia reports no H200 China revenue |
| July 2026 | U.S. official confirms limited shipments |
| August 2026 | Small H200 batches reach mainland China |
| Q2 FY2027 | Nvidia reports China H200 revenue |
| Q3 FY2027 | No China compute revenue assumed |
The progression illustrates how slowly the policy change has translated into actual sales.
China Is Still A Small Part Of Nvidia’s AI Business
The limited H200 sales also demonstrate how much Nvidia’s growth has shifted toward markets outside China.
Nvidia’s Q2 Data Center revenue reached $89 billion, up 117% year over year.
Global demand from hyperscalers, AI laboratories, enterprises and sovereign AI projects is currently far larger than Nvidia’s China business.
Nvidia’s Global AI Demand
Hyperscalers
+
AI Labs
+
Enterprises
+
NeoClouds
+
Sovereign AI
↓
Global AI Compute Demand
↓
Nvidia Data Center Revenue
↓
$89 Billion Q2That helps explain why Nvidia can continue projecting extremely rapid growth despite uncertainty in China.
Nvidia Expects 70% Revenue Growth In Fiscal 2028
The China situation is also being overshadowed by Nvidia’s extraordinarily strong global outlook.
Nvidia expects revenue to grow by approximately 70% in fiscal 2028.
CEO Jensen Huang said actual demand is significantly higher than that figure and described the company’s outlook as supply constrained.
This means Nvidia expects to grow rapidly even without relying on a major recovery in China.
The China Market Could Still Become Significant
Although current sales are small, China remains a potentially important market for Nvidia.
China has a large technology industry and numerous companies developing large AI models.
If H200 restrictions are relaxed further, Nvidia could potentially increase sales.
However, the company faces a shrinking window because domestic competitors are continuing to improve.
Potential China Outcomes
| Scenario | Nvidia Impact |
|---|---|
| More H200 approvals | Higher China revenue |
| Continued limited approvals | Modest contribution |
| Tighter U.S. restrictions | China sales remain constrained |
| Stronger Huawei adoption | Nvidia market share declines |
| Broader Chinese access | Potential significant upside |
The outcome will depend heavily on U.S.-China technology policy.
Nvidia Faces A Competitive Clock In China
The longer Nvidia remains restricted from China’s market, the more time domestic competitors have to develop their own ecosystems.
This is one of the risks Nvidia itself has previously highlighted.
In its regulatory filings, Nvidia warned that being effectively shut out of China’s Data Center compute market had allowed competitors to build larger developer and customer ecosystems.
That could make it harder for Nvidia to regain market share even if restrictions are later eased.
Nvidia’s China Strategy Is Becoming More Selective
Nvidia has also been exploring products that can comply with U.S. export rules.
However, the company recently denied a report that it was developing a China-specific language-processing unit for shipment later this year. Nvidia said it has no China-specific LPU product in its roadmap.
That leaves H200 as one of the most important Nvidia AI products currently capable of receiving limited China access.
The H200 Deal Has Geopolitical Implications
The shipments are about more than Nvidia’s revenue.
Advanced AI processors have become strategically important in the U.S.-China technology competition.
Washington wants to prevent China from gaining unrestricted access to the most advanced computing technology, while Beijing wants to accelerate domestic semiconductor development.
Nvidia is caught between the two policies.
United States
Restrict Advanced AI Chips
↕
Nvidia
↕
China
Secure AI Compute + Build Domestic ChipsThe H200 represents a compromise between those competing objectives.
What The H200 Shipments Mean For Nvidia
For Nvidia, the shipments provide three potential benefits.
First, they reopen at least part of a previously inaccessible market.
Second, they allow Nvidia to generate some revenue from existing H200 inventory.
Third, they give the company an opportunity to maintain relationships with major Chinese AI developers.
However, the immediate financial impact remains small.
What The Shipments Mean For China
For Chinese AI companies, access to H200 processors provides additional computing capacity for advanced AI development.
It also gives companies access to Nvidia’s established software ecosystem.
But the limited quantities mean Chinese firms cannot simply rely on H200s to satisfy all future AI demand.
They will continue to need domestic accelerators and alternative computing platforms.
What The Deal Means For Huawei
Huawei could face increased competition if Chinese companies receive broader access to H200 chips.
However, the limited nature of the shipments means domestic chipmakers retain significant protection from foreign competition.
The Chinese government appears to be balancing two objectives:
- Give AI companies access to useful computing power.
- Continue developing domestic semiconductor alternatives.
The current H200 policy appears designed to achieve both.
Nvidia’s Supply Constraints Add Another Complication
Nvidia is already dealing with a global supply shortage.
The company says demand for its AI infrastructure is significantly greater than available supply and expects approximately 70% revenue growth in fiscal 2028 despite those constraints.
That means Nvidia must decide how much production capacity to allocate to China while global customers are competing for the same infrastructure.
Nvidia Supply Allocation
Global AI Demand
↓
Limited Nvidia Supply
↓
Hyperscalers
AI Labs
Enterprises
Sovereign AI
China
↓
Allocation DecisionsChina therefore competes with a very large global customer base for Nvidia’s production capacity.
The Bigger Picture
Nvidia’s first H200 shipments to China represent a meaningful policy shift, but they are not yet a major financial development for the company. H200 sales to Chinese customers accounted for less than 1% of Nvidia’s $89 billion Data Center revenue in the second quarter of fiscal 2027. At the same time, Nvidia’s $108 billion third-quarter revenue forecast assumes no Data Center compute revenue from China, showing that management is not relying on a significant China recovery in its baseline outlook.
The longer-term importance is strategic. Chinese AI companies such as ByteDance and Tencent are receiving limited access to Nvidia’s H200 hardware even as Beijing encourages the development of domestic alternatives such as Huawei’s AI accelerators. Nvidia therefore has an opportunity to rebuild relationships with Chinese customers, but it also faces the risk that prolonged restrictions will allow domestic competitors to strengthen their hardware and software ecosystems.
Looking Ahead
The key question is whether Beijing will allow H200 purchases to expand beyond the initial limited shipments. If more Chinese companies receive approval and larger quantities can enter the mainland, Nvidia could eventually generate a more meaningful revenue contribution from China. However, the company is unlikely to depend on that market while U.S. export controls and Chinese regulatory policies remain unpredictable.
For investors, China’s importance should therefore be viewed as potential upside rather than the central driver of Nvidia’s current growth story. Nvidia’s global Data Center business is already generating $89 billion per quarter, while management expects approximately 70% revenue growth in fiscal 2028 despite supply constraints. The H200 shipments could reopen a strategically important market, but Nvidia’s broader AI infrastructure boom remains the much larger source of its near-term growth
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.
