Preface
The United States has imposed restrictions on China's export of high-end chips, such as TSMC, and the United States has asked it not to supply to Huawei. This rule has a greater impact on China's high-tech companies, such as Alibaba's cloud business and high-end computing service delivery capabilities, as well as Huawei, ZTE and other companies.
However, there is plenty of inventory and the training of large language models can continue in the next year or 18 months.
In addition, when it comes to inference, high-power and high-end chips are not necessarily required. So, is it limited by the chip shortage?
First, China will develop its own ability to manufacture high-end GPUs
At present, the global chip shortage has not been alleviated, especially for chips manufactured in advanced processes.
However, China has already begun to move and plans to put a self-developed high-end GPU product line into production around next year.
The CPU processors independently developed by Continental have begun to be applied on a large scale, and have achieved good results in both the workstation and server markets.
Therefore, it is believed that the mainland is not a problem to develop and manufacture high-end GPUs, especially the current chip shortage, and it is necessary for the mainland to step up and improve its ability to innovate independently.
Second, the shortage of chips is a big problem, but there are already people who have stockpiled inventory to deal with it
The chip shortage has always been a headache, especially for some companies, which may directly affect their production and business.
However, even with the chip shortage, we need to see some of the positives.
First of all, the shortage of chips has also directly led to the rise in its price, which may affect the procurement of some enterprises on the one hand, but on the other hand, it has also brought more profits to some chip companies, and has also encouraged more enterprises to increase investment in chip research and development and improve their independent research and development capabilities.
Secondly, for the shortage of chips, to a certain extent, it will also prompt some companies to make reserves in advance to ensure that their production and business will not be greatly affected.
Of course, we can't expect to solve the problem of chip shortage in the short term, and some deep-seated problems hidden behind it, such as the layout and division of labor in the industrial chain, require us to do long-term planning and adjustment.
Third, in terms of inference, you don't necessarily need high-power and high-end chips like the latest models of Nvidia
The phenomenon of chip shortage has actually brought some enlightenment to some enterprises.
For example, in terms of inference, it may be true that high-power and high-end chips like the latest models of Nvidia were needed to have better performance, but due to the shortage of chips, some companies have begun to try some new ways.
They found that in terms of inference, it is not necessarily necessary to have such a high performance, and through some software and hardware optimization, good results can be achieved on ordinary chips.
Therefore, for some enterprises, they may take the initiative to slow down the pace of procurement of high-end chips, start to try to balance performance and power consumption, and find chip products that are more suitable for themselves, which can also bring more reference and inspiration for the development of the entire industry.
epilogue
The problem of chip shortage is not clear in one or two sentences, which involves many factors and is more complex, which requires us to have a more comprehensive and in-depth understanding.
Of course, we also need to be cautious and optimistic when facing such problems.
We can't blindly fall into the negativity of it, but take the initiative to find some ways and methods to solve the problem.