Why Nvidia’s Approved H200 AI Chip Sales to China Are Still Stalled

Nvidia has received U.S. approval to sell its H200 artificial‑intelligence accelerator to several major Chinese technology companies—but the deal remains effectively frozen. Even though export licenses have cleared key buyers, no H200 chips had been delivered to China as of mid‑May 2026, leaving one of the most closely watched semiconductor deals in limbo .

The standoff highlights how AI hardware has become a geopolitical bargaining chip between Washington and Beijing. Both governments are shaping how, when, and whether the transactions can actually occur.

Current Status of the H200 Sales

U.S. regulators have approved roughly 10 Chinese companies to purchase Nvidia’s H200 GPUs, the company’s second‑most powerful AI data‑center chip . Despite that approval, shipments have not started.

People familiar with the matter say the absence of deliveries reflects unresolved regulatory and political issues rather than a lack of demand. Chinese cloud and AI firms continue to want the hardware, but cross‑border approvals remain incomplete .

Which Chinese Companies Have Been Cleared

Reporting indicates that several of China’s largest internet and cloud firms are among the approved buyers, including:

• Alibaba
• Tencent
• ByteDance
• JD.com

These companies are major operators of AI infrastructure and large cloud platforms, making them key potential customers for high‑performance AI accelerators .

The full list of the approximately ten approved companies has not been publicly confirmed, so the named firms represent those identified by sources rather than a complete official roster .

Approved Distributors for the Chips

In addition to end customers, the U.S. has reportedly authorized Lenovo and Foxconn to act as distributors for H200 systems destined for China .

These companies could package the chips into servers or AI infrastructure products used by Chinese cloud providers and enterprise customers.

The Licensing Conditions Behind the Deal

The approvals are part of a broader shift in U.S. export policy toward advanced AI chips.

Instead of automatically denying exports, the U.S. Commerce Department’s Bureau of Industry and Security now evaluates shipments of chips such as Nvidia’s H200 on a case‑by‑case licensing basis .

Key conditions attached to the framework include:

• Individual export licenses for specific customers
• Security and compliance checks
• Limits and oversight on shipments
• Inspection requirements for exported chips

Nvidia has also reported obtaining a license for a limited number of H200 shipments to China, with exports subject to inspection and a 25% duty on sales under the current framework .

This system allows limited commercial sales while preserving U.S. oversight of advanced computing hardware flowing into China.

Why No H200 Chips Have Been Delivered Yet

Despite U.S. clearance, the main bottleneck appears to be on the Chinese side.

Reports indicate that Chinese authorities have tightened scrutiny of foreign technology dependencies and are slowing or blocking imports of advanced AI hardware while promoting domestic semiconductor alternatives .

That means the transaction now depends on approvals and policy decisions from both governments. Even when U.S. export licenses are granted, shipments cannot proceed without Chinese import clearance and regulatory alignment.

How the U.S.–China Tech Rivalry Is Shaping the Deal

The H200 situation reflects a broader strategic balancing act.

Washington wants to restrict China’s access to the most powerful AI hardware while still allowing some controlled commercial trade. Moving from a blanket ban to a managed licensing regime attempts to preserve U.S. oversight while protecting American semiconductor companies’ global revenue streams .

Beijing, meanwhile, is trying to reduce reliance on foreign chips and accelerate domestic alternatives. Policies that slow or scrutinize imports help encourage Chinese companies to adopt local hardware.

What the Delay Means for Nvidia and Huawei

For Nvidia, the stalled shipments create uncertainty in one of the world’s largest AI infrastructure markets. Companies planning new AI clusters cannot rely on hardware that might remain stuck in regulatory review.

The pause also creates an opportunity for domestic competitors. Huawei and other Chinese chip developers are pushing AI accelerators aimed at replacing U.S. technology in local data centers. As Nvidia’s deliveries remain uncertain, Chinese cloud providers may invest more heavily in those alternatives.

In practice, the H200 deal has become more than a semiconductor sale—it is now a test case for how far advanced AI hardware can still move between the United States and China in an era of escalating technology competition.