How will AI develop in Shanghai in 2025? This event is about to give the answer, and has issued a "hero post" to star enterprises

On February 8th, the 2025 Global Developer Pioneer Conference will be held in Xuhui, Shanghai from February 21st to 23rd, according to the press conference of the Global Developer Pioneer Conference on the 7th. (Reporter Zhang Yangyang)

As a major conference in the AI industry, especially at present, Chinese companies represented by DeepSeek are once again setting off a new wave of global AI. The participation of companies in the conference, especially whether DeepSeek will show up, is a hot topic of discussion on site.

When asked if DeepSeek will participate in this year’s conference, the organizers said, ‘We have been in contact with DeepSeek and have received a very positive and affirmative response.’ ‘The Shanghai Developer Conference is a big ecosystem that they (DeepSeek) really need, and they (DeepSeek) will actively participate in the conference.’

It is understood that the 2025 Global Developers Pioneer Conference will closely link the entire industrial chain of artificial intelligence at home and abroad, forming an end-to-end organic synergy of the industry chain. This conference will invite participants such as basic models, vertical models, application scenarios, computing power, corpora, funds, tools, software service companies, and developer communities.

Including top domestic and international communities such as Hugging Face, Microsoft Developer Community, CSDN, Alibaba Moda Community, Linux Foundation, Apache Foundation, Huawei Community, etc., enterprises like Jieyue Xingchen, MiniMax, etc., will participate in this conference.

▍Shanghai reinforces four major basic bases

The artificial intelligence industry is one of the three leading industries in Shanghai. By the end of 2024, Shanghai issued the “Implementation Plan for ‘Modeling Shanghai’ in Artificial Intelligence” to accelerate the construction of the artificial intelligence “Shanghai Heights” and create a world-class industrial cluster for artificial intelligence.

At yesterday’s press conference, according to Yang Haijun, the chief engineer of the Shanghai Municipal Committee’s Internet Information Office, as of the end of 2024, Shanghai has completed the filing of 60 large-scale AI service models and registered 63 generative AI services, forming a batch of innovative applications that empower various industries through AI technology.

So, what are Shanghai’s plans for the development of the artificial intelligence industry in the new year?

During an interview with reporters, Pan Yan, director of the Artificial Intelligence Development Department of the Shanghai Economic and Information Technology Commission, said, “In addition to the ‘5+6’ key areas and industries in the previous ‘Implementation Plan’, we also need to further strengthen the construction of four basic foundations in our work this year, and build a solid foundation in terms of computing power, corpora, virtual and real integration, and large model base.”

The first is to build a super-large-scale autonomous intelligent computing cluster. With the in-depth development and large-scale application of artificial intelligence technology, the demand for post-model training and inference may soon exceed the demand for pre-training computing power. It is necessary to prepare for the arrival of a new round of industrial and technological revolution in advance for the guarantee of computing power and stable supply. Efforts will be made to promote the integration of infrastructure IaaS, software tool platform PaaS, corpus data layer DaaS and model layer MaaS capabilities, provide intelligent computing cloud services with “integrated training and promotion and hybrid deployment”, accelerate the transformation to a first-class intelligent computing cloud public service provider in China, and support the vertical application of the city’s large model industry and the digital transformation of thousands of industries.

Second, build a multi-level corpus supply system. In AI applications, the quality of the corpus directly determines the performance of the model. High-quality training data can effectively reduce the error rate of the model in the reasoning process, improve the creative ability and generation quality of the model in practical applications. Last year, nearly twenty companies jointly released the group standard “Corpus Construction Guidelines.” The plan is to establish a batch of general and specialized corpora, build a multi-level corpus system, support the development of basic large models and vertical applications. Focus on the training needs of cutting-edge large models and promote the creation of basic large model training corpora.

Third, it is to build a virtual and real integration super large-scale practical training field. On the technical side, the fusion of multimodal data standards, training systems, and deep reinforcement learning technology should be strengthened, and a cloud platform-based virtual and real integration testing system should be developed to build relevant data exchange protocols and enhance the ability to generate complex scenarios. On the industrial side, it is necessary to synchronously build a virtual and real integration testing standard system and policy system, and promote the diversified development of virtual and real integration testing scenarios by new research and development institutions and public technology platforms.

The fourth is to accelerate the innovation of the industry platform large-scale model system. Accelerate the integration and innovation of general and special large-scale models, and build a group of industry platform large-scale models. Support industry enterprises to strengthen the basic large-scale model and open-source ecology, promote the application of technical innovations such as reinforcement learning, instruction fine-tuning, and thinking chains. Speed up the cultivation of a group of industry platform large-scale models that integrate industry data sets, and encourage the research and development of technologies such as knowledge distillation, pruning, quantification, and parameter sharing. Focus on the development and application of scenario application large-scale models and intelligent agents, cultivate a group of industry application developers and integrators, encourage enterprises to build model-as-a-service platforms, and promote the landing and application of large-scale models.

▍Xuhui’s 18 AI industry policies are already on the way

As the main venue of this year’s conference, Xuhui District has been making continuous efforts in the field of artificial intelligence industry in recent years. In response to the relevant industries, Xuhui District has also successively introduced a number of policies.

According to Yu Linwei, member of the standing committee of the Xuhui District Committee and deputy district chief, as of the end of last year, 255 large-scale model enterprises, 34 registered large-scale models, and more than 100 investment institutions have successfully landed in Xuhui.

In an interview with reporters, Yu Linwei mentioned that artificial intelligence technology has made key progress, but many aspects, including the competitive landscape of enterprises, are still in the process of development. The supply and demand of artificial intelligence technology still needs to be strengthened, and there are many unsatisfactory aspects of the products; in addition, artificial intelligence companies urgently need application scenarios to solve the problem of ‘high valuation, low revenue’ development.

“Xuhui has a long-standing determination to support industrial development, so our biggest expectation for the conference is the world’s top enterprises and teams, whether they are university professors, entrepreneurial teams, or traditional industries embracing AI technology. As long as there are good ideas and thoughts in the field of artificial intelligence applications, they can come to this grand event to share together,” Yu Linwei said.

“We will position ourselves as the ‘world’s largest artificial intelligence incubator’ and proudly recommend Mosu Space to the world,” said Yu Linwei. On one hand, this year Mosu Space will rely on the seven major clusters of Xuhui Binjiang to launch industrial carriers covering over 100,000 square meters. At the same time, the ‘mechanism + policy’ of Mosu Space will be replicated throughout the entire region, forming a ‘one core leading, three circles radiating, and multiple points supporting’ development pattern of ‘1 + 3 + X’.

On the other hand, deepen cooperation with OpenAtom, Moda Community, and others to explore the online ‘Modo Space’, fully implement the upgraded incubation paradigm, promote the construction of basic ecosystem zones, joint innovation centers in universities, leading enterprise AI innovation centers, etc., with better policy support and closer ecological linkage, to build a more mature and competitive industrial development ecosystem.

Yulinwei also revealed that Xuhui District will introduce more powerful 18 policies in the near future. “We are determined to learn from the highest level in the world and have made new attempts in institutional and conceptual mechanisms by benchmarking and studying more than ten top science and technology innovation incubators worldwide. We will also comprehensively promote the new positioning of Modu Space as the ‘world’s largest artificial intelligence incubator brand’ and discover and welcome more artificial intelligence entrepreneurs to come to Modu Space and Xuhui for development.”

YU Linwei said that Xuhui District will take the lead in seizing the opportunity of large-scale model and use artificial intelligence to empower economic development, urban management, and livelihood services. At the same time, with the brand radiation effect of the conference, it will give full play to the scene resource advantages of the super-large urban central area, and promote the penetration of artificial intelligence technology into more high-frequency life scenarios of users in the vertical fields such as education, medical care, and urban governance, making Xuhui a phenomenal district for the application of artificial intelligence.

Source: Science and Technology Innovation Board Daily

Author: Science and Technology Innovation Board Daily