In the era of digital economy, computing power is a new quality of productivity and the focus of scientific and technological competition among countries. Computing power, as the "storage capacity" of data storage technology and the "transportation capacity" based on network communication technology, is an important part of the information industry. At the same time, computing power is also a kind of computing-based service capability, with the help of the penetration of the online market, like water and electricity, it continues to enter thousands of households and serve thousands of industries. According to relevant estimates, the average increase in the computing power index will increase by 3.6‰ and 1.7‰ respectively for the digital economy and GDP.
Computing power helps the formation and growth of the online market
At present, the digital transformation of all walks of life has led to the continuous shift of economic activities from offline to online, and the ability of online markets to allocate resources has improved. On the one hand, the development of the digital economy has formed an e-commerce market with similar content to the traditional market, but it is unique. On the other hand, digital platforms have given birth to emerging markets such as search, instant messaging, online games, and data transactions. As the digital economy gradually extends from the consumer end to the industrial end, and from the low-end repetitive manual labor to the knowledge-intensive and human emotional field, the online market will play a greater role in the resource allocation process of all links of the market system, such as production, consumption, trading, and circulation. At the same time, financial activities in online marketplaces are becoming more extensive and complex, placing higher demands on the speed and efficiency of data processing.
The operation of the online market is premised on computing power. In the past, the offline traditional market was formed by the convergence of market segments, and a local equilibrium was formed through the matching of supply and demand and the price mechanism of each market segment, and then the upstream and downstream of the industrial chain and supply chain and the interaction between different industries were conducted, and finally the general equilibrium of a unified market was formed. This mainly relies on the power of the spontaneous order of the market, and also hides the calculations and considerations of thousands of merchants, enterprises and consumers on their own decisions. The online market not only retains this market power, but also relies more on the platform's unity and coordination with different market players. To a certain extent, the algorithm replaces the production function to allocate resources, push products to users, provide procurement information to suppliers, and feedback personalized needs to manufacturers. The algorithm is the operating rules set by the platform, and the actual execution of the algorithm needs to be based on computing power. In daily life, the optimal configuration of the logistics system (such as the design of distribution routes) and the interaction between online and offline market players (such as takeaway) are inseparable from the accurate calculation and prediction of computing power. There is no doubt that the large-scale coordination of resource allocation in the online market relies on superior computing power, which is beyond the capabilities of ordinary oral arithmetic, abacus, mental arithmetic and calculators, and requires the help of modern digital technology and infrastructure.
In particular, online marketplace pricing and payment need to rely on computing power. On the one hand, computing power can facilitate price discovery in online marketplaces. Computing power enables enterprises to better match market supply and demand, and dynamically adjust their own prices, so as to promote the formation of enterprise prices, market segment prices, and general equilibrium prices, and reduce price failures in local markets and overall market shocks; On the other hand, online payment requires computing power to coordinate pricing, implement preferential policies, and reasonably distribute revenue. The online marketplace has set clear pricing rules, and most of the parties to the transaction do not need to bargain, but realize one-click quotation under the computing power service. The platform has set reference or guide prices for online car-hailing, freight trucks, labor dispatch, takeaways, etc., and some platforms have set rules such as price protection and the lowest price on the whole network, which play an important role in the actual pricing of merchants, and few merchants can significantly deviate from the relevant prices set by the platform. In this process, the platform is not a planner who overrides the market, but simulates the market through historical data to arrive at a price that approaches the general market equilibrium. It is conceivable that without super computing power, it will be difficult for the platform to reasonably guide the price and financial payment of a wide range of products and diversified merchants.
Computing power can also unleash potential consumption, cultivate and expand the "lemon market" such as second-hand cars and idle transactions. Like electricity, there are peaks and troughs in the use of computing power. At a certain point in time, people may need to allocate a large amount of resources in the online market, which to a certain extent promotes the investment of platform enterprises in computing technology and related facilities. The investment in computing power during the peak period will release the potential consumption of the mainland, forming a virtuous circle in which consumption and investment promote each other. A common example of this is the rise of Singles' Day and the rapid development of cloud computing. The release of this consumption potential is also reflected in new business formats such as live streaming. It is particularly noteworthy that, according to traditional economic theories, the "lemon market" such as second-hand cars and idle transactions has problems such as information asymmetry, moral hazard, and adverse selection, which makes it difficult for the market to develop effectively. However, the online market relies on computing power to easily obtain product quality information and consumer evaluation information, and form reasonable valuation, certification, reward and punishment rules, so as to promote the healthy and orderly development of the market.
Facing the future, the expansion of the online market is mainly due to the continuous implementation of new technologies in new scenarios, which requires continuous development of computing power. With the development and application of general artificial intelligence, the application of intelligent computing power is becoming more and more extensive, which can not only provide massive data processing capabilities, but also support high-performance intelligent computing, forming a new quality productivity with higher energy level and higher quality. From the perspective of specific technologies and application scenarios, digital yuan, smart contracts, blockchain, and connected cars have high requirements for computing power, which will further improve the efficiency of resource allocation in the online market.
Problems with computing power applications
From mobile phones and personal computers to servers and supercomputers, computing power is everywhere, and the convenience it brings to people is also everywhere. However, in the online market, the role of computing power still has certain limitations and deficiencies.
First, technical bottlenecks and shortcomings in the computing power industry chain. If computing power is the premise of the operation of the online market, then the chip with high computing power is the engine of computing power. However, compared with foreign countries, high-end chips, especially artificial intelligence chips suitable for large language model (LLM) training, still have shortcomings. It is imperative for mainland China to promote the independent and controllable localization of artificial intelligence chips, and it is urgent to break through the bottleneck of computing power and solve the "bottleneck" technology.
Second, the cost of computing power is high. Unlike hydropower, which has entered thousands of households, computing power requires a large amount of sunk cost investment, which is difficult for small and medium-sized enterprises to bear. In addition, the running costs of new technologies such as ChatGPT (a dialogue model product of Ope-nAI, an American artificial intelligence research company) are quite staggering, and even large enterprises are not easy to bear. Therefore, the application of new technologies urgently needs to solve the problem of computing power cost, and only by using a more cost-effective computing platform can we provide a lighter model for ordinary individual users and enterprise users.
Third, it is difficult to promote the depth of computing power in secondary training and vertical market segments. Computing power may promote the construction of a new technology ecology in the online market, but at present, artificial intelligence, machine learning and other technical models are still learning public knowledge on the Internet, and cannot solve some personalized problems of specific industries and enterprises, so enterprises need to conduct secondary training in relevant in-depth industries and vertical subdivided industries. For example, for enterprises in a certain industry, some vertical solutions are formed separately, and intelligent technology is used to iterate on professional privatization knowledge, so that it has the ability to solve practical problems. It is foreseeable that as artificial intelligence enters the era of large models with huge parameters, the growing demand for computing power has ushered in huge opportunities for the AI chip and server market, and the landing and application of AI in vertical fields such as education, autonomous driving, and device-side equipment will be the main line in 2024. This requires a large number of excellent companies to increase investment in computing power, strengthen the research and development of computing technology, improve computing capabilities, and launch some products with solutions that are closer to customer needs and pain points, so as to increase the market penetration rate of computing power.
Suggestions for further relying on computing power to promote the development of the online market
Computing power has become a key driving force in the digital world, driving online economic activities to continue to unleash development vitality. Based on this, more attention should be paid to the layout of computing power, and the construction of computing power should be regarded as the policy focus of economic growth, especially the prosperity and development of the online market.
First, improve the computing infrastructure and increase the effective supply. Some people compare data to the "oil" of the information age, and compare computing power to the most important "infrastructure", which is enough to show the importance of computing power. It is necessary to strengthen institutional reform and relax the infrastructure access and industry access of relevant industries. The main contents of data infrastructure regulation are infrastructure co-construction and sharing, inter-network interconnection and fair access, business license access, service quality standardization, and fair competition order maintenance. Just as the per capita electricity consumption is valued in the era of industrial economy, the per capita computing power should be regarded as an important indicator to measure the comprehensive competitiveness of industries in various regions in the era of digital economy. More computing infrastructure will be deployed in backward areas, and financial subsidies will be provided to enterprises in backward areas to use computing power.
Second, improve the computing power industry chain and overcome technical shortcomings. Focusing on key links such as computing, network, and storage, we will guide the integration and development of upstream and downstream enterprises in the computing power industry chain, promote the implementation of cutting-edge technologies such as smart computing, and empower the breakthrough development of artificial intelligence in mainland China with high computing power. Cultivate a group of computing power service providers, develop and promote computing power products and services, and build a complete ecological partner system for small and medium-sized enterprises to actively use computing power. By increasing R&D investment and financial subsidies, we will increase scientific and technological innovation and talent cultivation, deepen school-enterprise collaboration, and strengthen the research and development of key technologies for computing power networks.
Third, strengthen data sharing, feeding and training of new technologies, and guide the allocation of resources in online market segments. On the one hand, establish a data governance structure, improve the data sharing and coordination mechanism, promote the orderly sharing of data, reduce semi-structured and unstructured data, eliminate different types of data barriers, and strengthen the correlation and interoperability between data. On the other hand, there are different requirements for text generation technology for different application scenarios, such as the financial field needs to generate specialized and technical texts, while literary works focus more on generating emotional and creative texts, and optimize relevant technical and policy details. In addition, it is necessary to promote the implementation of major projects such as data training bases and national blockchain hub nodes, especially to strengthen the application of high-quality Chinese data in artificial intelligence technology feeding and training.
Fourth, we need to coordinate development, equity and security to enhance social benefits. In the future, advanced computing technologies such as cloud computing, quantum computing, and edge computing will continue to improve the efficiency of resource allocation and promote economic development. Actively consider the social equity of computing power, gradually promote computing power to become a social-level service that can be "one-point access, ready to use" like water and electricity, and make detailed provisions on the protection of workers' rights and interests, consumer protection, and the goodness of scientific and technological products after the use of computing power. Laws and regulations such as open sharing of computing power data and cross-border data circulation have been introduced to improve the level of risk governance in the process of computing power use and effectively avoid data security risks.
(The author is an associate researcher at the Chinese Academy of Social Sciences)