抢在美国新规前，英伟达过去几个月一直游说客户尽快下单

Core Points - The U.S. government is set to implement a new AI diffusion framework aimed at maintaining its competitive edge in AI development by restricting the export of advanced chips [1][2] - Companies like NVIDIA and Oracle are accelerating the shipment of chips and servers to data centers outside the U.S. to mitigate the impact of the new regulations [1] - The new rules require that 50% of a company's total computing power must be located in the U.S., and 75% must be in the U.S. or among 18 allied countries, limiting other countries to only 7% [2] Group 1 - The AI diffusion framework will require U.S. government approval for exporting NVIDIA's advanced chips to most countries [2] - Cloud service providers will need to closely monitor the destinations of their AI chip shipments due to the new regulations [2] - The complexity of the new rules lies in measuring the GPU capacity of cloud service providers across different countries, focusing on total processing power rather than the number of GPUs [2] Group 2 - The new regulations have faced criticism from companies, including NVIDIA, and some Republican senators, who argue it may harm U.S. leadership in AI and chip innovation [3] - NVIDIA's spokesperson stated that the rules benefit foreign competitors and do not encourage investment in U.S. technology [3] - The Trump administration is reportedly considering modifications to the new rules to alleviate their most severe impacts [4]