PRESIDENT Xi Jinping went to the coalface of China’s ambitions to lead development of artificial intelligence with a visit last week to a Shanghai site that is nurturing new applications for the technology.

“AI technology is evolving rapidly and entering a phase of explosive growth,” Xi said during his visit to the Shanghai Foundation Model Innovation Center, emphasizing the nascent yet pivotal nature of the industry and its connection to the younger generation.

The center is not just any other incubator. It aims to be the world’s biggest incubator of AI advancement and already has a roster of more than 100 startups involved in projects ranging from humanoid robots to breakthrough AI models. It sort of creates a unique lifestyle in which different firms and entrepreneurs share knowledge and find ways to cooperate with each other. There goes a popular saying about the center: You go upstairs and you can find your upstream partners; you go downstairs and you may find your downstream partners.

The visit generated a burst of enthusiasm at the center in Xuhui District, encouraging innovators in their pioneering work developing everything from robots for factory and home use to AI technology already impacting the lives of millions of users.

However, these startups face competition from established international players such as Tesla, Google and OpenAI — the developer of ChatGPT — creating both pressure and opportunities, particularly amid the complexities of current Sino-US trade tensions.

Infinigence AI is a startup that owes much of its fast growth to the center.

Infinigence AI was founded in May 2023. When it joined the Shanghai Foundation Model Innovation Center in September the same year, it had only seven staff members. With the help of the center, Infinigence AI quickly registered in Shanghai. Other things also went well with the firm, such as those related to patent application.

So far, Infinigence AI has raised about 1 billion yuan (US$139 million) in several financing rounds and its staff now number 200, said a report published by The Paper, a Shanghai-based news portal.

Infinigence AI’s inclusive development of AI — one that makes AI affordable and available to as many users as possible — is all the more important as it tackles one of the most fundamental infrastructures of AI — computing power.

“We’ve reduced the cost of our AI compute power by 10,000 times,” said Dai Guohao, co-founder and chief scientist of Infinigence AI.

He was referring to “a computing power supermarket” jointly launched in February by the Shanghai Foundation Model Innovation Center, Infinigence AI and INESA. A much lower cost of compute power would enable AI to access and benefit various users, Dai explained.

You would better understand his remarks if you know how important computing power is to AI development and application and how expensive such power can be.

On April 28, McKinsey & Company published an article, titled “The cost of compute: A US$7 trillion race to scale data centers,” which describes computing power as one of this decade’s most critical resources amid an AI boom.

The article estimates that nearly US$7 trillion in capital outlays will be needed by 2030 to keep pace with the demand of computing power — a staggering number by any measure.

In this context, whoever can meaningfully cut the cost of computing power will gain a competitive edge in AI development while at the same time making AI more affordable and available to the largest possible groups of users.

During Xi’s tour, Peng Zhihui, co-founder and chief technology officer of AgiBot, showcased the company’s humanoid robots, including the recently unveiled Lingxi X2. This general-purpose robot, launched in March, demonstrated improved motion control and intelligence by riding a bicycle.

“Humanoid robots are the most promising physical form leading to generalized artificial intelligence in the future and have become a hotspot for international competition,” said Peng, a former Huawei Technologies employee and prominent online figure.

AgiBot forecasts substantial revenue growth in 2025 from production of thousands of humanoid robots this year, following the milestone of its 1,000th robot rolling off the production line in January.

The sector’s potential is further underscored by technology giant Tencent’s raising its stake in Agibot to 2.7 percent from 2.06 percent in April.

Leveraging Shanghai’s abundant data resources, Kupas, a company founded in 2024, has focused on building a high-quality corpus database to better serve the AI industry, particularly smaller companies and innovative entrepreneurs.

Corpus, meaning “body” or collection of data in AI industry, is crucial for enhancing AI model training, which is vital for applications across various specialized fields, including physics and medicine.

Another startup within the center, Shanghai Quality Intelligence Co, is concentrating on developing self-reliant AI infrastructure products and solutions — a strategic focus given the current US sanctions. Zhu Xin, the company’s chief executive, said the mission is to provide “high-quality AI for critical sectors.”

A booming AI incubator

Prior to Xi’s visit, Shanghai had already outlined ambitious plans to parlay the center into world prominence, underscoring the city’s determination to become a global leader in AI innovation and application.

The incubation center has expanded in size almost sixfold since its inception in 2023. Many of the companies operating there, such as Kupas, were founded in the last one to two years, highlighting the rapid growth of the ecosystem.

According to Yu Linwei, vice director of Xuhui District, the expanded incubator encompasses a comprehensive, vibrant ecosystem of online community involvement, policy support and fundraising channels. The district hosts about 255 AI companies, including notable players like MiniMax and StepFun, as well as more than 100 investment institutions and 10 “unicorn” companies, or startup private firms valued at over US$1 billion. The center benefits from city-level subsidies exceeding 130 million yuan and offers open-source “test corpus” capacity exceeding 180 terabytes.

Collaborative effort is part of the center’s ethos, with participants like Quality Intelligence partnering with Kupas and Midu on text corpus, models, systems and application development.

AI sits alongside biomedicine and integrated circuits as primary pillars of Shanghai’s “pioneer industries.” In the first quarter, revenue from the city’s AI industry grew by over 13 percent, outpacing the 5.1 percent growth in municipal gross domestic product.

Shanghai officials have said the city aims to create a world-class AI industry ecosystem by 2025, with the industry’s scale surpassing 450 billion yuan in 2024.

Major tech companies like Baidu, Huawei, SenseTime and Tencent have established regional headquarters or functional centers in Shanghai, contributing to a highly coordinated AI industry chain.

The Shanghai Data Exchange facilitates data sharing, while SenseTime’s AI Data Center, the largest of its kind in Asia, provides substantial computing power for training advanced AI models.