NEW YORK, December 31, 2025, 05:51 ET — Premarket
- TSMC shares eased about 0.4% in U.S. premarket trading, hovering just under $300.
- Reuters reported Nvidia has approached TSMC about expanding H200 AI chip production for Chinese demand.
- Investors are also parsing TSMC’s 2nm production milestone ahead of its Jan. 15 earnings conference.
Taiwan Semiconductor Manufacturing Co’s U.S.-listed shares fell 0.4% in premarket trading on Wednesday after Reuters reported Nvidia was sounding out the chipmaker about expanding production of its H200 artificial intelligence (AI) chips for Chinese customers. Chinese firms have ordered more than 2 million H200 chips for 2026 while Nvidia has about 700,000 units in stock, and it has asked TSMC to start additional output in the second quarter of 2026 using TSMC’s 4-nanometer process, Reuters said, adding that Beijing has yet to greenlight shipments and sources put pricing at around $27,000 per chip. The stock was down $1.15 at $299.58. 1
The development matters because TSMC sits at the choke point of the AI hardware boom, manufacturing many of the most advanced chips designed by U.S. technology firms. When demand shifts for leading-edge processors, it can move not just one stock but expectations for factory utilisation, pricing and future spending across the semiconductor supply chain.
The backdrop is jittery, holiday-thin trading as investors reassess big tech positioning into year-end. “It’s just a healthy rebalancing of allocations more so than an emotionally driven sell-off,” said Mark Hackett, chief market strategist at Nationwide. 2
Separately, TSMC said on its website that its 2-nanometer (2nm) “N2” technology has started volume production in the fourth quarter, and said it will be the most advanced technology in the industry in terms of density and energy efficiency. A process node like 2nm is shorthand for a new generation of manufacturing that packs more transistors into a chip, typically improving speed while cutting power use. Volume production means the process is running at scale rather than on pilot lines. 3
For investors, the key question is how smoothly the company can ramp 2nm without sacrificing yields. Yields are the share of usable chips coming off a wafer, and small changes can swing margins when volumes rise.
The 2nm milestone also sharpens the competitive backdrop as rivals such as Samsung and Intel push their own next-generation manufacturing. In markets, a durable lead at the cutting edge often translates into stronger pricing power and more predictable demand from top-tier customers.
Semiconductor trading was mixed before the opening bell. Nvidia was down 0.3%, AMD slipped about 0.1% and the iShares Semiconductor ETF edged lower, while Intel rose about 1.7%, keeping moves modest as traders weighed headlines.
China demand stayed in focus after Reuters reported that ByteDance plans to spend about 100 billion yuan ($14.29 billion) on AI chips from Nvidia in 2026 if H200 sales are allowed, citing the South China Morning Post. Reuters said it could not immediately verify the report. 4
For TSMC, that kind of end-demand signal matters less for any single quarter than for how customers plan capacity a year or more out. Big AI deployments can pull forward orders for advanced manufacturing, while policy shifts can just as quickly delay them.
Investors will also watch for any sign that export rules and approval processes become a bigger bottleneck than factory output. Even a short pause can show up first in customer inventory moves and only later in foundry wafer starts.
The next company catalyst is TSMC’s fourth-quarter earnings conference on Jan. 15, when it is due to update its outlook and discuss spending plans. TSMC said the call will be held at 1:00 a.m. Eastern Time and that it will observe a quiet period from Jan. 5 to Jan. 14. 5
