Nvidia halts China focused H200 production and shifts capacity to Rubin

Nvidia has stopped producing its China targeted H200 Hopper GPU at TSMC after building a large inventory, as export and import restrictions from the United States and China slow deployment. The company is now reallocating some manufacturing and packaging capacity toward its next generation Rubin chips.

Nvidia has halted production of its China focused H200 Hopper GPU at Taiwan Semiconductor Manufacturing Company facilities after building a substantial buffer of inventory aimed at the Chinese market. The company has accumulated an inventory of 250,000 H200 GPUs, which will be available in the Chinese market for select applications that do not compromise United States national security. Nvidia began stockpiling these accelerators after securing export rights from Washington, intending to serve artificial intelligence laboratories across China while staying within tightened United States export controls.

Regulatory constraints on the Chinese side are now limiting how quickly this inventory can move. China has also restricted what its domestic companies and artificial intelligence labs can import, meaning that the import of H200 GPUs is still prohibited unless a company receives a letter of exemption from Beijing. As a result, Nvidia used its TSMC N5 5 nm node capacity to create about 250,000 units, which are now stored in a warehouse awaiting export approval from the U.S. administration and import approval from Chinese customs for artificial intelligence labs. This stockpiling strategy leaves Nvidia with a sizeable pool of ready hardware but dependent on dual government approvals before shipments can ramp.

Attention is now shifting to Nvidia’s next generation Rubin platform as production priorities evolve. The Financial Times and Reuters note that Nvidia will now “reallocate” capacity from H200 production to the new Rubin, even though the two generations differ in both process technology and packaging. For Hopper, Nvidia uses TSMC’s 5 nm node with CoWoS-S packaging, while Rubin uses a 3 nm node with CoWoS-L packaging. Analysts suggest these reports most likely refer to reallocating packaging or other secured capacity rather than directly converting mature 5 nm wafer lines, since it is unlikely that the 5 nm semiconductor node can be converted into a 3 nm node without significant line remodeling and changes to manufacturing equipment. Packaging can be adjusted more easily, which is likely the practical path for shifting resources from H200 to Rubin.

68

Impact Score

Adobe plans outcome-based pricing for Artificial Intelligence agents

Adobe is positioning its Artificial Intelligence agents around performance-based pricing, charging only when the software completes useful work. The approach points to a more results-oriented model for selling generative Artificial Intelligence tools to business customers.

Tech firms commit billions to Artificial Intelligence infrastructure

Amazon, OpenAI, Nvidia, Meta, Google and others are signing increasingly large cloud, chip and data center agreements as demand for Artificial Intelligence infrastructure accelerates. The latest wave of deals spans investments, compute purchases, chip supply agreements and data center buildouts.

JEDEC outlines LPDDR6 expansion for data centers

JEDEC has previewed planned updates to LPDDR6 aimed at pushing the memory standard beyond mobile devices and into selected data center and accelerated computing use cases. The roadmap includes higher-capacity packaging options, flexible metadata support, 512 GB densities, and a new SOCAMM2 module standard.

Tsmc debuts A13 process technology

Tsmc has introduced its A13 process at its 2026 North America Technology Symposium as a tighter version of A14 aimed at next-generation Artificial Intelligence, high performance computing, and mobile designs. The company positions the node as a more compact and efficient option with backward-compatible design rules for faster migration.

Contact Us

Got questions? Use the form to contact us.

Contact Form

Clicking next sends a verification code to your email. After verifying, you can enter your message.