Before reading this article, please click "Follow", which is convenient for you to discuss and share, and can bring you more participation, thank you for your support.
The lithography machine conundrum
As we all know, ASML lithography machine, the world's most sophisticated instrument, can manufacture 3nm process chips, and this lithography machine, only the Dutch ASML company can produce
Mainland China has been slow to develop in this regard due to problems such as insufficient technical experience, long research and development cycle, huge investment, and precision requirements
Optical convolution processing technology
Recently, the Chinese Academy of Sciences has developed an optical convolution processing technology that does not require lithography processing. Its computing and processing speed is 1,000 times that of traditional electronic chips, which can not only greatly improve the computing system of artificial intelligence systems, but also solve the problem of the neck of the development of mainland semiconductors.
As we all know, NVIDIA is the most valuable chip manufacturer in the United States, covering more than half of the world's chip business, and China's high-end chips have always been provided by it. However, the United States restricted Huawei's development in the field of high-end mobile phones on the grounds of endangering national security, and American chip companies also lost the Chinese market. After weighing the pros and cons, NVIDIA announced that it would no longer export high-end chips to China, which hindered the development of China-related enterprises, and the price of high-end chips rose again and again. Now that the Chinese Academy of Sciences has developed optical convolution processing technology, the mainland no longer has to be subject to people, and can greatly promote the development of artificial intelligence and related industries.
Artificial intelligence is a technical science used to simulate, extend and extend human intelligence. Because it is very computationally intensive, it cannot be used in major industries by ordinary chips alone, so scientists have developed special accelerators to support its development, such as NVIDIA's GPU and Google's TPU. But these accelerators are very expensive and small and medium-sized enterprises cannot afford them, and they can only access AI capabilities through cloud computing.
Traditional electronic chips are achieved by characterizing microcircuits on tiny semiconductors, which rely on electron streams for calculations, so the operation speed is slow. The optical convolution processor uses a new laser beam to process information, which solves the bottleneck of the computing speed of traditional electronic chips. According to the Chinese Academy of Sciences, this technology uses the coherence and nonlinear optical materials of lasers to complete the real-time convolution of images, which makes its computing speed 1000 times faster than traditional chips, thereby greatly improving the computing efficiency and stability of artificial intelligence systems.
People who want to produce this technology do not need to resort to complex lithography, just convenient lamination technology. In this process, nonlinear media and other components are stacked into multilayer constructions that can be easily engraved on a flying silicon substrate and extended to different sizes and different kinds of nonlinear media.
After the technology is announced
As soon as this technical theory was released, it attracted the attention of major companies in the world, and high-end chips can be produced without lithography technology, and everyone is excited. But science and technology are priceless for one country and can never be sold to other countries, just like NVIDIA's CPU technology. As for the online statement that the American business bigwigs lined up to cooperate with the Chinese Academy of Sciences, it is actually nonsense, they know the value of chips, and they know that China will not sell them, so they come to China not to ask for optical convolution processors, but to improve Sino-US relations, because they do not want to lose the Chinese market and interests. Although they can still make money without a Chinese market, but the Chinese market is large, instead of picking up a little here and there, it is better to concentrate on making money.
What's more, the optical convolution processor is still in the theoretical stage, and there is still a long way to go before it can be applied in practice, and capitalists do not have so much time to wait for it. But it can be expected that once this technology lands, we will no longer be subject to people!