“The US is imposing even stricter restrictions on the export of chips and artificial intelligence technology, seeking to preserve advanced computing developments for its own needs and allies, and to make it more difficult for China to access them. Reuters writes about it. The new laws provide for the introduction of global licensing requirements for certain types of chips, with few exceptions. All countries were divided into three categories. The first group includes 18 countries, including Japan, Great Britain, South Korea and the Netherlands, and they will be practically unaffected by the restrictions. About 120 countries, including Singapore, Israel, Saudi Arabia and the UAE, will face moderate restrictions. At the same time, third countries such as China, Russia and Iran, which are subject to arms embargoes, will be completely deprived of access to these technologies.Advertisement: US cloud providers, including AWS and Microsoft, plan to allow only 50% of computing resources to be deployed outside the US AI capabilities. Of these, no more than 25% can be directed to the countries of the second category, and for any single country not included in the first group, this indicator will be limited to 7%. Limitations will also affect advanced graphics processors (GPUs), which are necessary for training AI models and the operation of data centers. The lion’s share of such chips is produced by Nvidia, and some by Advanced Micro Devices. At the same time, leading providers of cloud services such as Microsoft, Google and Amazon will have the opportunity to build data centers with global licenses that will exempt their projects from national quotas for artificial intelligence chips.”, — write: epravda.com.ua
The US is imposing even stricter restrictions on the export of chips and artificial intelligence technology, seeking to preserve advanced computing developments for its own needs and allies, and to make it more difficult for China to access them. Reuters writes about it. The new laws provide for the introduction of global licensing requirements for certain types of chips, with few exceptions. All countries were divided into three categories. The first group includes 18 countries, including Japan, Great Britain, South Korea and the Netherlands, and they will be practically unaffected by the restrictions. About 120 countries, including Singapore, Israel, Saudi Arabia and the UAE, will face moderate restrictions. At the same time, third countries such as China, Russia and Iran, which are subject to arms embargoes, will be completely deprived of access to these technologies.Advertisement: US cloud providers, including AWS and Microsoft, plan to allow only 50% of computing resources to be deployed outside the US AI capabilities. Of these, no more than 25% can be directed to the countries of the second category, and for any single country not included in the first group, this indicator will be limited to 7%. Limitations will also affect advanced graphics processors (GPUs), which are necessary for training AI models and the operation of data centers. The lion’s share of such chips is produced by Nvidia, and some by Advanced Micro Devices. At the same time, leading providers of cloud services such as Microsoft, Google and Amazon will have the opportunity to build data centers with global licenses that will exempt their projects from national quotas for artificial intelligence chips.