New York, January 6, 2026, 17:34 (EST) — After-hours
Nvidia (NVDA.O) shares were down about 0.5% at $187.24 in after-hours trading on Tuesday after CEO Jensen Huang said “it’s just going to be purchase orders” that show whether Chinese customers can import the company’s H200 AI chip. CFO Colette Kress said the U.S. government was working “feverishly” on export licenses, while Huang said Nvidia had fired up its supply chain as demand in China was “quite high.” Reuters
The China update landed as Nvidia used the Consumer Electronics Show in Las Vegas to outline its next data-center platform, Vera Rubin, which Huang said is in full production and will arrive later this year. Huang said Rubin can deliver five times the AI computing of Nvidia’s previous chips in chatbot workloads and can raise the efficiency of generating “tokens” — small chunks of text that models process — by 10 times in large clusters. Nvidia faces tougher competition in running deployed AI services, from Advanced Micro Devices (AMD.O) and from customers such as Alphabet’s Google, even as it expects CoreWeave, Microsoft, Oracle, Amazon and Alphabet to adopt Rubin systems. Reuters
Chip stocks rallied in regular trading on Tuesday after Huang’s CES remarks sparked fresh interest in memory and storage plays, lifting SanDisk (SNDK.O), Western Digital (WDC.O), Seagate (STX.O) and Micron (MU.O) to record highs. The Philadelphia SE Semiconductor Index (.SOX) also hit an all-time high, and Argent Capital’s Jed Ellerbroek said he expects “a very strong earnings season for Big Tech” with capital spending forecasts revised higher. Reuters
In a company blog post, Nvidia said Rubin is a six-chip platform designed to lower the cost of generating tokens to roughly one-tenth of the prior generation, and it highlighted a new “inference” (running a model rather than training it) context memory storage product aimed at speeding up long prompts. Huang also used CES to push open AI models across several domains and introduced Alpamayo, an open reasoning model family for autonomous-vehicle development, the company wrote. NVIDIA Blog
Nvidia’s talk also rippled through the wider AI supply chain, as investors weighed whether more efficient chips will curb spending on parts of data-center infrastructure. Barclays analysts led by Julian Mitchell wrote that “given the primacy of Nvidia to the whole AI ecosystem, one should not take their comments lightly,” after Huang said future systems may need no water chillers for cooling. Reuters
Other AI-linked chip names were mixed in late trade, with AMD down about 3.1% and Broadcom (AVGO.O) little changed, while AI server maker Super Micro Computer (SMCI.O) was up about 1.5%. Traders said moves in the complex remain sensitive to shifts in expectations for cloud spending and to policy signals on cross-border chip sales.
But the path to fresh China shipments depends on when export licenses clear, and Washington could tighten rules again if politics shift. A slower pace of data-center buildouts, or faster gains by rival chips and in-house designs, would test Nvidia’s pricing power as investors head into the heart of earnings season.
Investors now turn to U.S. labor data that can sway rate bets and risk appetite for high-growth chip stocks, with the Job Openings and Labor Turnover Survey for November 2025 due on Jan. 7 at 10:00 a.m. ET and the December employment report set for Jan. 9 at 8:30 a.m. ET. Any update on H200 licensing — or more details on Rubin orders — could add to swings in Nvidia and its peers.
