THE APEX TIMES
Treasury Secretary Scott Bessent said President Trump made it “very clear” that China should not receive NVIDIA chips positioned above the H100 tier, a restriction that builds on an earlier permission for a step lower, the H200. NVIDIA shares trade under NVDA.
U.S. Treasury Secretary Scott Bessent said President Trump made it “very clear” that China should not get NVIDIA AI chips “above H100s,” reinforcing a high-end ceiling on what the country can obtain. The comment, reported by Yahoo Finance, suggests Washington is continuing to calibrate export limits around the most advanced AI accelerators used to train and run large-scale models.
Bessent’s remarks refer to a tiered product boundary. In the way NVIDIA markets its data center GPUs, the H100 is positioned as a flagship-class accelerator for training and inference. The H200 is described as a step below the top H100 tier, and the report says President Trump previously allowed exports of H200-class chips last December.
The implication of a strict cutoff at the H100 level is that the most capable NVIDIA devices would remain out of reach for Chinese buyers, while some narrower access could continue for lower-tier systems. The distinction matters because performance and cost efficiency can change markedly across accelerator generations, affecting the ability of customers to scale AI workloads under tighter restrictions.
The announcement did not provide granular details in the way export rules are often implemented, such as which specific SKUs would be covered, whether the restriction applies to direct sales only or also to shipments via intermediaries, or what enforcement mechanisms or licensing procedures would apply. The report frames the shift largely as a political instruction on the intended direction of export policy rather than an administrative update with new technical definitions.
For NVIDIA, the company’s data center business is closely tied to demand for high-end GPUs used in AI infrastructure. If buyers in China are constrained to less advanced tiers, that can shape order patterns, product mix, and revenue timing, even if a subset of chips remains available.
Sector context is also important. The United States and other governments have increasingly treated advanced AI chips as strategic technologies because they can accelerate model training and automation. Export restrictions have become a lever for governments to manage both national security concerns and competition dynamics in frontier AI.
As of the report’s publication, NVIDIA itself did not make the policy decision in the cited statement, and the company did not provide additional information about how it would interpret or apply the tier ceiling in that specific post. Traders and analysts would typically look for follow-through through export-control guidance, licensing documents, or NVIDIA product availability announcements, none of which were outlined in the reported comments.
Investors will likely watch for whether the H100 ceiling becomes formalized through updated U.S. export-control regulations, additional licensing terms, or clearer compliance instructions to the market. Another key watch item is whether the earlier H200 carve-out holds, expands, or contracts, since the balance between restricted top-tier devices and allowed lower-tier alternatives can drive both customer behavior and NVIDIA’s shipment cadence.
A city policy advocate criticized Amazon over what he described as a new $5 express-delivery fee in New York City, adding pressure on the retail giant to justify added charges to consumers.
A two-day drop in AMD shares followed a claim by short-seller Hunterbrook that a blacklisted Chinese server maker kept using newer AMD firmware for EPYC chips. Jim Cramer said investors should “thank the shortseller for a discount,” pushing the question to whether the China concern is overstated or just finding a market price.
The New York University professor and valuation researcher said he set a $140 limit buy for NVIDIA (NVDA) after the stock closed at $125.53 on a recent November session, according to a new report. The accompanying note emphasizes the valuation logic behind the implied target price.
The analyst says improving results do not automatically erase the risk premium investors apply after a sharp run-down, even when companies beat expectations and raise guidance. Nvidia’s trajectory, he argues, resembles Apple’s earlier down cycle.
Traders are re-examining how restrictions on Nvidia’s China-bound products could ripple through the AI infrastructure trade, as the market tries to map demand, substitution risk, and revenue timing.
The company disclosed an SEC probe focused on revenue recognition tied to a major customer, and its board said it had received Huang’s resignation.
A review of 13F filings for the period ending Sept. 30 found 17 funds added Palantir while 6 reduced holdings, including Stanley Druckenmiller’s family office among the buyers.
A sharp late-week swirl of investor attention tied new AI developments at Anthropic to the market’s ongoing obsession with Nvidia’s chips.
A Reuters report said the SEC is examining possible accounting issues, prompting renewed attention to how Nvidia booked revenue connected to Singapore. In a prior comment when Nvidia CFO, interim CEO Colette Kress said more than 99% of Singapore billings were for U.S. customers.
The analyst raised its Amazon (AMZN) target to $310 from $285 and kept an Outperform rating, pointing to drivers that it says are not primarily about shopping or cloud.
Comments
Community guidelines. Comments may be removed for violations.
Loading comments...