AI and Large Language Models: Competition and future trends in ChinaSource:原创 Analysis on China's Policy Environment and Market Competition in AI and Virtual reality
With the continuous evolution of AI (artificial intelligence) technology, LLMs (large language models), generative AI, and DLRM (deep learning models) have gradually become the digital engines of modern economic development. LLMs are shaping the future trends of AI, as new business scenarios emerge and extensive datasets accumulate, the requirements for computing power and architecture are on the rise. China is making rapid development in the field of digital technology. For instance, China accounts for approximately 25% of the global AI chip market, and in the realm of cloud computing, China's market size is steadily expanding and projected to reach 1.2 trillion-yuan RMB by 2025. What is the present state of China's market development in this sector? What opportunities and challenges are suggested by the overarching AI trend? 01 In early October, six national level government departments, including the MIIT (Ministry of Industry and Information Technology) jointly released the 'High-Quality Development Action Plan for Computing Power Infrastructure.' The plan sets forth ambitious goals to be achieved by 2025, including reaching a computing power scale exceeding 300 EFLOPS, a target of 35% for the proportion of intelligent computing power, a balanced development of computing power in eastern and western regions, and a storage capacity goal of over 1800EB. This initiative aims to drive technological advancements and applications in various sectors including industry, finance, energy, and education. This demonstrates China's vision and determination to take a leadership role in the global stage in the field of computing. As of September 2023, Huawei's Atlas 900 SuperCluster represents one of the pinnacle achievements in global computing power, equivalent to the computing capacity of 500,000 PCs, providing robust support in addressing this challenge together with other top super computers such as Sunway TaihuLight and Tianhe. Image credit: Huawei 02 — The key position of Metaverse in China’s economic development In recent years, China has made significant advancements in the fields of Virtual Reality (VR), Augmented Reality (AR), AI, cloud computing, and high-speed internet infrastructure, which have propelled the widespread application and development of virtual world technologies. On September 8th, the MIIT in collaboration with five other national-level departments introduced the 'Metaverse Industry Innovation and Development Three-Year Action Plan (2023-2025)' as a strategic framework to expedite significant breakthroughs, support China's ambition to excel in manufacturing, network infrastructure, and cultural development. This action plan underscores the pivotal role of metaverse technologies in China's future economic development and provides clear direction for the industry's growth. In the era of Web3.0, the global metaverse industry market is rapidly growing. According to Statista's data, it is expected that the global metaverse market will reach 936.6 billion US dollars by 2030. According to a report by the China Industrial Research Institute, the size of the Chinese metaverse industry was approximately 42.5 billion Chinese yuan in 2022. Over the next 10 years, the global metaverse market is projected to grow at an average rate of 50%, and the domestic metaverse market will surpass at least 600 billion yuan by 2027, approaching 100 billion US dollars. The substantial amount of content to enrich metaverse can be generated by AIGC (Artificial Intelligence Generated Content) technology. Research shows that the Chinese AIGC market is expected to reach 17 billion Chinese yuan in 2023. As a result of open source of LLMs and improvement of the ecosystem, the industry is expected to experience rapid growth with a projected CAGR (compound annual growth rate) exceeding 70% by 2025. It is estimated that by 2027, the Chinese AIGC industry will exceed 60 billion yuan, and by 2030, the market size will exceed one trillion yuan. Image credit: Airui research — The "Action Plan" addresses the importance of establishing an industrial ecosystem for coordinated growth, with a specific emphasis on nurturing both prominent companies in the metaverse and specialized small and medium-sized businesses. Regional Governments’ policies for the metaverse industry As of now, data from "China Electronics News" reveals that almost 30 regions, including Beijing, Shanghai, Sichuan, and Henan, have systematically introduced regional-level metaverse industry action plans and guidance. These strategic measures are designed to facilitate metaverse development and applications across diverse industries by emphasizing technological innovation, infrastructure for platforms, and the augmentation of international partnerships. For instance, Beijing's Tongzhou district is actively evolving into a metaverse application showcase zone, with a particular emphasis on its unique cultural and tourism content. Meanwhile, Shanghai is focusing on the pivotal role of government investment funds and the establishment of an industry fund dedicated to the "metaverse." This fund is designed to steer private capital towards supporting the growth of this industry. With the unprecedented popularity of ChatGPT, in China, AI large models have rapidly become an integral part of daily lives. Data shows that there are at least 130 domestic companies developing large model products, with 78 of them focusing on general-purpose large models, and over half of these large models are open-source, with universities and research institutions leading the way in contributions. Image credit: Zhipu AI On September 8, 2023, Alibaba Ant Group unveiled the CodeFuse large model at the Bund Summit in Shanghai, available on the Wisemodel community. CodeFuse is an open-source code generation model designed specifically to support the entire software development lifecycle. It’s Code LLMs aiming at enhancing programming efficiency. On October 17, Baidu Intelligent Cloud launched the "Qianfan AI Native Application Workspace," a one-stop, cloud-based platform for large model development and application. It supports the customization and development of large models, including third-party models. Baidu also introduced the latest iteration of its generative AI language model, ERNIE 4.0, claiming that its overall performance is on par with OpenAI's GPT-4. In the wave of generative AI technology revolution, Baichuan Intelligence has achieved the record for the fastest ascent to unicorn status among domestic large model startups. They recently concluded a $300 million strategic Series A financing round, with participation from tech giants such as Alibaba, Tencent, Xiaomi, and several top-tier investment firms. Zhipu AI, a team originated from Tsinghua University, has amassed over 2.5 billion Chinese yuan in financing this year. 05 A new approach for foreign investment in the Chinese market Additionally, the establishment of the Sino-U.S. Economic, Financial Working Groups at the end of September signals a shift from confrontation to long-term strategic competition between China and the US, despite future uncertainties. On October 24th, the Sino-U.S. Economic Working Group held its first meeting via video, indicating that the world's two largest economies remain closely interconnected, and the field of AI will inevitably continue to advance within the context of this competition. email:info@sirmionecg.com; Mobile:+8613810284967 |