NEW YORK, January 1, 2026, 19:27 ET — Market closed
- Nvidia (NVDA) last closed down 0.55% at $186.50 on Dec. 31, the final session of 2025.
- Reuters reported Nvidia has approached TSMC to ramp production of its H200 AI chips after Chinese firms ordered more than 2 million units for 2026.
- Investors are watching China’s approval process for H200 shipments, Nvidia’s Jan. 5 CES presentation, and its Feb. 25 earnings date.
Nvidia shares last closed at $186.50 on Wednesday, down 0.55%, with U.S. markets shut on Thursday for New Year’s Day.
The focus is on a Reuters report that Nvidia has sounded out Taiwan Semiconductor Manufacturing Co about expanding production of its H200 chips as orders from China surged, raising fresh questions about supply allocation.
That matters now because the H200 is a key data-center graphics processing unit (GPU) — a chip used to train and run AI models — and any incremental demand from China lands in a market already sensitive to capacity constraints. The report also underscored regulatory risk: Beijing has yet to approve H200 shipments, even as Washington has recently moved to allow exports under new terms.
Chinese technology companies have placed orders for more than 2 million H200 chips for 2026, while Nvidia holds about 700,000 units in stock, Reuters reported, citing sources. One source said expanded output at TSMC would start in the second quarter of 2026, though volumes were not clear.
Sources said Nvidia has priced the chips at about $27,000 each for China and plans to offer both standalone H200 chips and GH200 “Grace Hopper” superchips that pair a Grace CPU with Hopper GPUs. Reuters also reported Nvidia has been focused on ramping its newer Blackwell line and its upcoming Rubin chips, putting attention on whether any extra H200 production competes for manufacturing resources. Reuters
Reuters said an eight-chip module is expected to cost about 1.5 million yuan, and sources described the H200 as roughly six times the performance of the H20 — a lower-spec chip Nvidia designed for China that later became unavailable there. Reuters also reported Chinese officials are weighing whether to require buyers to bundle purchases of domestic chips with H200 orders.
A separate Reuters report cited the South China Morning Post as saying ByteDance plans to spend about 100 billion yuan ($14.29 billion) on Nvidia AI chips in 2026, contingent on China allowing H200 sales; Reuters said it could not immediately verify the report.
Supply-chain attention has also turned to high-bandwidth memory (HBM), the high-speed memory stacked alongside AI chips. Samsung Electronics co-CEO Jun Young-hyun said in a New Year address that customer feedback on its next-generation HBM4 chips has been strong, with some customers saying “Samsung is back,” as the company seeks to supply HBM4 to Nvidia and catch up with rival SK Hynix. Reuters
Broader risk appetite was already fragile into year-end. The S&P 500 fell 0.74% on Dec. 31 and trading volumes were thin in the holiday-shortened week, Reuters reported.
Before the next session, traders are likely to watch for any follow-through on China export approvals and whether more production planning for H200 chips tightens expectations for availability elsewhere, including for Nvidia’s latest-generation offerings.
The next near-term catalyst is Nvidia’s scheduled CES press conference on Monday, Jan. 5, where founder and CEO Jensen Huang is slated to present updates in Las Vegas.
Further out, Nvidia’s investor relations calendar lists its fourth-quarter fiscal 2026 financial results for Feb. 25. Investors typically focus on data-center demand, supply constraints and any commentary on China exposure when the company updates guidance.
From a chart perspective, the stock’s Dec. 31 range — a low of $186.49 and high of $190.56 — is likely to be in focus when trading resumes.
