IDC According to the report , The scale of China's AI infrastructure market is 2020 reach 39.3 Billion dollars , Year-on-year growth 26.8%. among ,AI The size of the server market accounts for the overall AI infrastructure market 87% above . With AI The rapid development of algorithms , More and more model training needs huge amount of computing power to implement quickly and effectively , Computational power is the decisive factor for the breakthrough of artificial intelligence application in the future .IDC expect , China's AI servers will maintain rapid growth , And will be in 2024 reach 78.0 Billion dollars .
AI Computational power has become the decisive factor in the breakthrough of artificial intelligence
Among the three elements of AI development , Whether it's data or algorithms , They can't do without the support of computational force , Computing power is playing an increasingly important role . With the rapid development of artificial intelligence algorithm , More and more model training needs huge amount of computing power to implement quickly and effectively . at present , Some models have approached the computational limit of artificial intelligence ,OpenAI The most recently announced AI Language model GPT-3 It's not just that the size of the model has increased to 1750 Billion , And the amount of data is amazing 45TB, On the one hand, this evolution , For new tasks , There's no need to re collect a lot of tagged data , Data utilization efficiency has been further improved ; On the other hand , It can avoid over fitting in algorithm fine-tuning , As a result, the generalization ability of the model is reduced . And the key to this capability is , In summary ： Do wonders vigorously .
《2020-2021 China's AI computing power development assessment report 》 Release
AI Diversified computing power ,GPU The proportion of servers is still as high as 95%
The report presents five important trends in the development of AI computing power in the future ： First of all , Artificial intelligence chips will continue to show the development of diversity ,GPU Still the first choice for data center acceleration , occupy 95% Above market share ; second , China's AI servers will maintain rapid growth in the next five years , Is the core driver of the overall server market growth ; Third , AI computing power will gradually penetrate into the edge , To 2023 year , nearly 20％ Used for processing AI The servers for the workload will be deployed on the edge side . Fourth ,AI The cloud service （AIaaS） It has become an important driving force for the development of artificial intelligence market ,2018 to 2024 The compound annual growth rate is expected to reach 93.6%; The fifth ,AI The benchmark is gradually improved ,MLperf、SPEC ML、AI-Rank And other authoritative benchmarks have been launched in succession , Not only for the cost-effectiveness of enterprises to provide a reference , It is also an important factor for the sustainable development of artificial intelligence applications in the future .
Beijing 、 deep 、 Hang 、 Shanghai 、 Chongqing ranks first AI Computational City TOP5
In the report 2020 The list of AI cities in China , The top five cities in turn are Beijing 、 Shenzhen 、 Hangzhou 、 Shanghai 、 Chongqing , ranking 6-10 The city of Guangzhou is Guangzhou 、 Hefei 、 Suzhou 、 Xi'an 、 nanjing . And 2019 Compared to , Shenzhen surpassed Hangzhou in second place , Chongqing enters the first echelon , Xi'an surpasses Nanjing in ninth place . except TOP10 Outside the city , Many cities are driven by their own industrial advantages and various factors , Great progress has been made in the application of artificial intelligence , For example, intelligent manufacturing in Dongguan ; Smart medicine in Wuhan ; Smart agriculture in Hefei, etc , The development of artificial intelligence cities in China is blooming everywhere , In the future, there will be more and more artificial intelligence demonstration areas combined with urban characteristics , Set a benchmark for industrial development .
From the perspective of AI industry application penetration ranking , Same as 2019 Compared to , The Internet remains number one , The application scenarios of Telecom and manufacturing industries are more abundant , The market potential is expected to be greatly improved , stay 2020 In the first half of , Under novel coronavirus pneumonia, the medical industry has accelerated the landing of AI applications , Remarkable results have been achieved in many aspects , It also promotes the penetration of artificial intelligence applications in the medical industry to surpass that in the education industry, ranking seventh .
In addition to the industries where AI investment is relatively concentrated , Driven by business needs , A lot of fragmented applications are beginning to be widely used , And radiate to media entertainment 、 Modern agriculture 、 Smart home 、 Smart power and many other different fields . On the whole , Including biometrics 、 Intelligent customer service 、 The general application scenarios, including precision marketing, have been quite mature , The industrialization of artificial intelligence has gone from point to area , From general application scenarios to more industry-specific scenarios , industry AI From the early pilot has gradually become the development and survival of enterprises .
public AI Computing infrastructure has become 74.5% The enterprise just needs
IDC Research shows , More than 90% of enterprises are using or plan to use artificial intelligence in the next three years , and 74.5% In the future, we hope that we can adopt the public computing infrastructure of artificial intelligence with utility significance . future , With more scale 、 An inclusive AI infrastructure platform has been built , China's AI industry will enter another fast lane of development —— The firm needs of enterprises determine the speed of AI development , Computing power broadens the way for the development of new public infrastructure .
in addition ,IDC In this study, we also investigate the common needs and challenges of enterprises in the application of artificial intelligence , There is a lack of data for model training 、 There is a shortage of computing infrastructure 、 And the high cost of artificial intelligence applications is the main challenge facing most enterprises . therefore , The main body of social service represented by the government , It is very important to promote the construction of AI public computing infrastructure , We need to fully consider the current needs and pain points faced by enterprises , With AI ecological partners , Join hands with key AI infrastructure providers , act with united strength , Jointly build a platform to accelerate industrial development .
From the global AI From the development of computational power ,2020 In, China's AI servers accounted for about one third of the global AI server market , It is the backbone of the global AI industry .AI Computing power profile reflects a country's cutting-edge innovation capability . about AI The investment in computing power also shows that the country attaches great importance to artificial intelligence at the strategic level , And the urgent vision of enterprises to enhance their core competitiveness through the development opportunity of artificial intelligence .（ Xiabin ）
本文为[Blood wolf SEO]所创，转载请带上原文链接，感谢