New AI customers have brought new demands for office buildings, and the property keywords have changed from cleaning, security, and reception to hardware, services, and operation and maintenance.
The development of AI industry rewrites the leasing logic, and commercial real estate welcomes opportunities for inventory transformation
Entering Hangzhou Zijingang Science and Technology City Cainiao Cloud Valley Industrial Park (hereinafter referred to as Cloud Valley), nearly 90% of the companies settled in the park cover large models, humanoid robots, and AI vision tracks. R&D computer rooms operating 24/7 can be seen everywhere. "A few years ago, investment in the park was still dominated by traditional Internet and cultural and creative enterprises, but now, more than 70% of the new leasing demand comes from AI industry chain enterprises." said Mr. Zhao, head of operations of Nandu Property Cloud Valley Project.
In recent years, with the large-scale implementation of the AI industry, the office market in domestic core cities has ushered in structural changes. Cases of AI technology startups expanding their entire floors and purchasing buildings for their own use continue to emerge. The AI industry is quietly reshaping the urban commercial real estate landscape.
AI companies reshape the office leasing landscape
At present, the office market has obvious regional differentiation characteristics. Taking Hangzhou as an example, the Hangzhou office building report for the first half of 2026 released by Cushman & Wakefield shows that the vacancy rate in the traditional business sectors of Wulin and Qianjiang New Town exceeds 27%, and rents continue to be under pressure; the vacancy rate in Future Science and Technology City and Binjiang AI Industrial Park is only 8.7%, and the rental premium of high-quality buildings is 15% to 20%, making it difficult to find a room in some high-quality buildings.
AI companies are expanding rents and the demand for single-family rentals continues to be released, which has injected strong impetus into the development of commercial real estate. Yushu Technology and Yunshen Technology successively rented single-family buildings in the industrial park in Hangzhou Future Science and Technology City; Alibaba Tongyi Qianwen continued to expand its lease of multi-story office space in Alibaba Cloud Valley; Zhipu Huazhang acquired the building in Zhongguancun Software Park in Beijing and used it as a large model R&D headquarters; ByteDance's large model team successively acquired multiple office spaces in core business districts such as Beijing, Shanghai, and Shenzhen; a large number of AI start-ups in Shenzhen poured into the OPC science and technology community, and the industrial agglomeration effect was highlighted...
A UBS research report pointed out that this round of domestic commercial real estate restoration does not rely on policy releases, but is supported by real office demand brought by the expansion of the AI industry. Different from the previous Internet expansion cycle, AI companies have smaller personnel, but are more willing to rent single buildings and entire floors, which puts forward higher requirements for building hardware and dedicated operation and maintenance services.
Start-up teams often rent 100 to 500 square meters of space to work independently, and leading companies frequently take over entire floors and thousands of square meters of space. AI science and technology tenants present distinctive new characteristics: short decision-making cycles, large lease area spans, strict hardware standards, and no fixed time limits for R&D work.
"The shortcomings of traditional office buildings are very prominent. The air conditioners are shut down at night, the power load is insufficient, and servers cannot be placed. They completely do not meet the needs of model training." The person in charge of a scientific and technological enterprise that is deeply involved in the research and development of multi-modal large models told reporters that when the company was selecting a site, it compared more than 10 office building projects and finally landed in Yungu. The main factors were the three hard conditions of the park's expandable power supply, 7×24 hours heating and ventilation, and reserved equipment rooms.
"Hardware transformation + service upgrade" adapts to new business formats
"The location selection logic of AI companies is completely different from that of traditional companies." Mr. Zhang, the head of a local commercial real estate agency in Hangzhou, said frankly. "In the past, the key words for investment were subways, business districts, and supporting facilities. Now they must be marked with 'adaptive computing room and dual-circuit power supply' to attract AI customers."
Faced with the differentiated needs of AI companies, Zhejiang office property management no longer only focuses on the traditional model of "cleaning, security, and basic services", but instead focuses on systematic upgrades in three aspects: building hardware renovation, exclusive industrial services, and smart park operation and maintenance, to create a new office carrier suitable for the artificial intelligence industry.
The EFC European and American Financial City Science and Technology Group, operated by Narada Property Management, invested tens of millions of yuan this year to upgrade the power distribution and air-conditioning systems. "The park has separately added transformers for AI tenants, and the air conditioners in the computer room area operate at a constant temperature throughout the year. Yunxiang cleaning robots and AI video inspection systems are also deployed in public areas, and blind spots in manual inspections are automatically filled by machines." Li Ting, project operations manager, said that in response to the late-night research and development and weekend office needs of AI companies, the property has launched three value-added services: 24-hour maintenance, exclusive conference services, and computing power supporting coordination. The property fee of the Kechuang Group has been increased to 4.2 yuan/square meter, and it remains fully rented.
Property management companies are also using AI technology to complete digital transformation, reduce operating costs and improve service accuracy. Narada Property has deployed AI patrol robots and intelligent inspection drones in Yungu to automatically identify damaged exterior walls and fire hazards; Binjiang Service United Technology Enterprise has implemented an AI energy consumption control system to dynamically adjust lighting and air-conditioning operations, reducing the overall energy consumption of the park by 30%; many office buildings in Zhejiang have enabled senseless traffic and AI intelligent repair reporting, achieving "one-click response" to corporate needs and reducing labor management costs by more than 30%.
For AI companies at different stages of development, some properties have also launched tiered space solutions: providing shared workstations for start-up teams, significantly lowering the threshold for starting a business; providing divisible whole-floor offices for growth-stage companies, and flexibly adjusting the rental area; for leading unicorns, connecting to independent headquarters buildings, equipped with exclusive naming and independent research and development laboratories. Flexible and diverse space supply covers the full cycle needs of AI companies from incubation to large-scale development.
Property management shifts from space management to service operation and maintenance
While the AI industry has created new market demands for commercial real estate and property services and reshaped the industry landscape, it has also brought new challenges to the structural transformation of the industry.
"In the future, the differentiation of the office building market will further intensify. The asset value of science and technology parks with AI industry supporting facilities, hardware adaptation, and industrial operation capabilities will continue to rise, while the value of single traditional commercial office buildings will continue to shrink." Li Congyue, head of property and asset management for East China at Jones Lang LaSalle, believes that "in the past, the core competitiveness of properties was the environment, security, and reception. In the future, the core barriers are computing power supporting capabilities, mechanical and electrical special operation and maintenance, and industrial resource integration capabilities."
Some places have already taken action. "Traditional properties only manage space, while AI park properties must serve industries." Mr. Liu, the person in charge of Binjiang Service Business Management Section, said that in order to better serve customers, the park has specially set up science and technology innovation service specialists to assist settled companies in applying for digital economy policies, docking computing resources, organizing industry chain salons, and providing supporting services to companies.
"Currently, the industry's biggest pain points are focused on renovation cost pressure and compound operation and maintenance talent gaps." Peng Yu, deputy director of research at the Property Division of the Zhongzhi Research Institute, suggested promoting industry upgrading through gradient transformation: large industrial parks can complete dual-circuit power supply and machine room load-bearing transformation at one time; small and medium-sized existing office buildings should prioritize the implementation of lightweight AI inspection and energy consumption control equipment to control early investment. At the same time, property management can cooperate with local universities and AI companies to cultivate comprehensive operation and maintenance personnel with both mechanical, electrical and computing power foundations to solve the problem of professional talent shortage.
Yu Xiaofen, dean of the China Housing and Real Estate Research Institute of Zhejiang University of Technology, said that office properties in various regions cannot apply a unified renovation template, and hardware and services must be adjusted to fit the local AI business format. "For example, local companies in Hangzhou need small model laboratories and real-life robot training venues; industrial AI companies in Ningbo need supporting sample testing areas and large-scale equipment freight channels. Only by misaligned layouts in various places can we avoid a surplus of homogeneous AI-themed buildings."
"AI is restructuring the demand for office buildings, forcing the property industry to shift from a mere 'space manager' to an 'industrial operation service provider.' This is a key opportunity for commercial real estate to achieve high-quality transformation," Yu Xiaofen said.
Reporter Wang Jiajun
"Worker Daily" (September 07, 2026, Page 04)
