Limited batches of Nvidia Corp.’s powerful H200 AI chips have been approved for shipment into mainland China, signaling a limited easing of restrictions on advanced semiconductor hardware. This development allows key Chinese technology firms to access critical AI training infrastructure despite broader US export controls that block Nvidia’s latest-generation processors.

ByteDance and Tencent Holdings Ltd. have each received around 10,000 H200 processors in the past few weeks. The Financial Times reported that other Chinese tech groups may soon receive approval for similarly sized shipments. These H200 chips are older than Nvidia’s current flagship AI processors, which remain inaccessible to Chinese customers due to stringent US export regulations.

Regulatory and Operational Constraints

Purchases of H200 chips require separate approval through an application process overseen by China’s state planner, the National Development and Reform Commission (NDRC). Last week, several Nvidia partners, including Lenovo, informed Chinese customers they could resume ordering products featuring H200 chips. These partners integrate Nvidia chips with CPUs and networking equipment to build AI servers.

Chinese regulators have also permitted companies to ship H200 chips to Hong Kong, which operates under a separate customs regime. However, access to Hong Kong offers limited relief for Chinese tech companies. The city lacks sufficient data-center capacity to accommodate the chips, and power constraints are expected to limit any rapid expansion of deployment there.

Strategic Implications for Chinese Tech

In July, Nvidia officially began shipping its H200 artificial intelligence chips to China under strict case-by-case government licenses. Initial volumes remained heavily restricted. Recent US approvals have cleared around 10 Chinese firms to receive advanced AI hardware from Nvidia and Advanced Micro Devices Inc.

This follows the Trump administration’s authorization in May for the sale of Nvidia’s H200 AI chips to 10 Chinese tech giants, including Alibaba Group Holding Ltd. and Tencent Holdings. Each approved customer can buy up to 75,000 chips. However, Beijing’s intervention halted the deliveries at that time.

China’s ongoing import controls underscore its push to strengthen domestic chipmakers, particularly Huawei, as they seek to capture a larger share of the world’s second-largest semiconductor market. Nvidia remains largely shut out by US export restrictions. Nevertheless, China’s limited access to the most advanced chipmaking equipment could constrain its ability to meet rapidly rising domestic demand in the near term. This constraint is likely to sustain some demand for Nvidia chips, particularly for AI training workloads.

Nvidia did not immediately respond to requests for comments.

Elon Musk raised doubts about the construction timeline for a major Ohio data center project backed by Nvidia Corp (NASDAQ: NVDA) and intended to support OpenAI’s growing artificial intelligence infrastructure needs. Musk posted on X that the facility will take "much longer to bring online than they think," casting skepticism on the aggressive rollout plans tied to a massive financial commitment from the chipmaker.

Nvidia Caps Ohio Data Center Commitment at $105 Billion

According to documents filed with the Securities and Exchange Commission, Nvidia’s aggregate payment obligations under the initial commitment are capped at $105 billion. These payments are subject to specific conditions, including the data center meeting applicable ready-for-service requirements. The first commitments are expected to begin in 2028.

The sprawling data center is being developed by SB Energy, an entity owned by SoftBank Group (OTC: SFTBF) (OTC: SFTBY). OpenAI has agreed to lease the site for 20 years, while Nvidia will serve as its exclusive chip provider. In addition to the lease guarantee, Nvidia plans to invest $1.5 billion in SB Energy. This follows a $1 billion investment from OpenAI and SoftBank earlier this year.

The Ohio campus is expected to eventually support as much as 8 gigawatts of power capacity. The initial phase targets 800 megawatts coming online in 2028.

Why Musk’s Warning Matters

Musk’s skepticism arrives as his own companies accelerate the build-out of massive AI computing infrastructure. Earlier this month, Musk stated that Space Exploration Technologies Corp (NASDAQ: SPCX) would exclusively use Nvidia GPUs. SpaceX is targeting roughly 10 gigawatts of AI compute capacity by the end of 2027, a significant increase from its current capacity of about 1.4 gigawatts.

Nvidia’s AI Infrastructure Push

The Ohio project underscores Nvidia’s increasingly central role in the AI infrastructure boom. The company’s data center revenue reached $75 billion last quarter, marking a 92% year-over-year increase. Last week, Nvidia announced a partnership with six major Wall Street asset managers to unlock more than $500 billion in financing for AI infrastructure.

Nvidia closed Tuesday at $219.74, down 2.34%. The shares slipped another 0.21% to $219.28 in after-hours trading, according to Benzinga Pro. According to Benzinga Edge Rankings, Nvidia ranks in the 99th percentile for growth and maintains positive short-, medium- and long-term price trend ratings.