Tuesday, 8 September 2026
China’s artificial intelligence expansion is beginning to have consequences far beyond the technology industry. The enormous investment being directed towards computing capacity, advanced manufacturing and digital infrastructure is creating physical requirements for data centres, offices, research facilities, business parks and energy infrastructure. For commercial real estate investors, this is turning AI from a technology story into an increasingly important property-market question. Evidence emerging during the first half of 2026 shows that some of this demand is already measurable. Technology businesses are contributing to office and business-park leasing in Shanghai, companies connected with artificial intelligence and integrated circuits are taking space, and investors in Beijing are expanding into data centres alongside more traditional property sectors. At the same time, high vacancy across parts of the technology-property market demonstrates why investors need to distinguish genuine occupier demand from development based primarily on expectations of future growth.
Shanghai provides one of the clearest examples. Its office market recorded improving leasing activity during the second quarter, with technology among the industries supporting demand. Falling rents have also changed the economics of expansion, allowing businesses that might previously have found larger or higher-quality premises prohibitively expensive to consider a broader range of buildings and locations. AI should not be regarded as a solution to Shanghai’s wider office oversupply, as vacancy remains high and substantial new supply is still competing for tenants. Nevertheless, technology companies are providing an additional source of demand at a time when many conventional corporate occupiers remain cautious.
The connection is even clearer in Shanghai’s business parks. Net absorption reached approximately 205,000 square metres during Q2 2026, with artificial intelligence and integrated-circuit companies among the active technology occupiers. Research organisations and innovation-oriented businesses also contributed to leasing activity. These numbers provide tangible evidence that China’s technology investment is translating into demand for physical premises, yet they also demonstrate the limitations of the AI property narrative. Business-park vacancy remained approximately 31.7% at the end of the quarter despite the leasing activity, while rents declined again. Two realities therefore exist simultaneously: AI and related technology companies are creating genuine property demand, but there is still considerably more space available than the market currently requires in some locations.
That distinction is important because China is continuing to develop technology-oriented districts and research campuses. The presence of AI businesses in the economy does not automatically justify every new technology park. Successful occupiers still make conventional real estate decisions based on costs, transport, building quality, access to employees, infrastructure and proximity to customers, suppliers and research institutions. The strongest technology-property locations are therefore likely to be those where an existing economic ecosystem supports the buildings rather than those relying primarily on expectations that technology companies will eventually arrive.
Beijing provides another part of the story. Commercial-property investment strengthened during the first half of 2026, reaching its highest first-half level in several years, while buyers broadened their interest beyond conventional property sectors. Data centres were among the alternative assets attracting capital. The attraction is understandable. Artificial intelligence requires enormous computing capacity, and that capacity ultimately has to exist somewhere physically. Servers require buildings, cooling systems, electricity, telecommunications connections and highly resilient infrastructure. For property investors, however, a data centre is fundamentally different from an ordinary warehouse or office because its economic value is closely connected to the infrastructure supporting it. Access to sufficient electricity, network connectivity, cooling capacity and suitable land can be as important as the building itself.
This makes China’s AI property market inseparable from its energy strategy. China has been expanding computing infrastructure across a network that connects major eastern technology and economic centres with regions further west where land and energy can be more readily available. By the end of March 2026, more than 80% of the intelligent computing capacity within the national computing-hub system was concentrated in eight major nodes. By the end of June, China’s intelligent computing capacity had reached approximately 2,185 EFLOPS, while national computing-facility utilisation was reported at around 71%. The figures illustrate both the enormous expansion already achieved and the importance of examining how effectively that infrastructure is actually being used.
The geography of AI property could therefore look very different from China’s traditional office market. Beijing, Shanghai, Shenzhen and other major cities can remain centres for research, corporate management, software development and highly skilled employment, while some computing-intensive activities can be accommodated in areas where electricity and land are more abundant. This creates opportunities for regions that previously played a limited role in institutional technology property, but it also creates a new investment risk. Low land and electricity costs do not automatically guarantee successful data-centre demand. A facility still requires customers, reliable connectivity and sufficient utilisation to justify the capital invested in it. Investors therefore need to examine actual workloads and customer commitments rather than assuming that national AI growth will make every computing facility successful.
The connection between AI and energy infrastructure is becoming particularly important. Computing facilities consume large quantities of electricity, and increasingly sophisticated AI systems are intensifying the relationship between digital infrastructure and power availability. Chinese authorities are consequently placing greater emphasis on coordinating computing investment with electricity supply. This could make access to power one of the most important location factors in China’s emerging digital-property market. For conventional real estate investors, that represents a significant change. Office investment has traditionally been dominated by questions such as location, transport, tenant quality and rent, while logistics investment depends heavily on transport networks and access to consumers. Data centres add another fundamental constraint: the property may have limited economic value if sufficient power cannot be secured.
AI is also creating a much broader industrial-property story. The technology depends on an ecosystem that extends through semiconductors, telecommunications equipment, robotics, automation, research and advanced manufacturing. These activities require a wide variety of buildings, including laboratories, production facilities, research premises, specialised industrial buildings and supporting offices. The wider AI economy therefore has the potential to create property demand well beyond data centres and conventional technology offices. This is particularly relevant for business parks capable of combining research, office and light-industrial functions. A traditional central office tower may work well for software and management teams, but businesses involved in engineering, semiconductor development, robotics or testing can require substantially different environments. As AI companies move beyond research towards broader commercial operations, their property requirements can consequently extend into engineering, testing and supporting functions.
That does not mean every specialised technology building represents an attractive investment. Highly customised property can become difficult to reuse if the original occupier leaves. The more specialised the infrastructure, the more important it becomes to understand the financial strength of the tenant and the likelihood that alternative users could occupy the facility. Flexibility could therefore become an important dividing line within technology property. Buildings capable of accommodating several types of research, office or light-industrial occupier may provide greater protection against technological change than facilities designed around one highly specific activity.
Talent represents another important factor in determining where these clusters develop. Artificial intelligence remains heavily dependent on skilled employees, researchers and engineers. Areas close to universities, laboratories and established technology companies can therefore possess advantages that cheaper locations struggle to reproduce. This helps explain why China’s largest technology centres remain important despite comparatively high land and occupancy costs. Property is only one component of a company’s location decision, and access to employees and research networks can be considerably more valuable than obtaining cheaper space elsewhere. Successful technology districts can also become self-reinforcing as established companies attract employees and suppliers, research institutions create new businesses, and investors and service providers follow the growing cluster. Each additional occupier can strengthen the surrounding ecosystem, but buildings alone cannot manufacture that process.
Shanghai’s business-park vacancy illustrates the problem particularly clearly. AI and semiconductor businesses are actively leasing space, yet almost one-third of the relevant stock remains vacant. New development therefore still needs to be tested against actual occupier requirements regardless of how attractive the technology story appears. The same principle applies to data centres and advanced industrial projects. A facility supported by identified customers or occupiers presents a different investment proposition from one built largely because future AI demand is expected to materialise.
That distinction could become increasingly important as Chinese cities compete for technology investment. Artificial intelligence has become strategically important, giving local governments powerful incentives to develop computing infrastructure, innovation districts and advanced industrial clusters. Some will succeed in attracting sustainable occupier ecosystems, while others may find themselves competing for the same relatively limited group of expanding companies. For institutional investors, the strongest opportunities are consequently likely to be where AI-related demand can already be demonstrated. Offices occupied by expanding technology companies, data centres with established customers, research parks connected to functioning technology clusters and industrial facilities supporting active manufacturing operations provide clearer evidence of underlying property demand. The further an investment depends on forecasts of future AI growth rather than current occupiers and utilisation, the greater the development and leasing risk becomes.
This makes China’s AI property story considerably more interesting than a simple forecast of expanding data-centre demand. The technology is beginning to influence several layers of the built environment simultaneously. It is contributing to leasing in Shanghai’s technology districts, broadening institutional interest in data centres, supporting demand from semiconductor and advanced manufacturing businesses and creating stronger links between commercial property and electricity infrastructure. The resulting investment map could extend from the office towers and research campuses of China’s largest cities to industrial facilities, computing centres and power-intensive infrastructure across entirely different regions.
The winners, however, will not necessarily be the places constructing the greatest amount of AI-labelled property. They are more likely to be locations where companies, skilled workers, infrastructure, electricity and customers combine to create sustainable demand. For commercial real estate investors, that is the critical distinction. China’s AI expansion is producing a genuine physical property footprint, but the investment challenge is determining which buildings are being supported by the new economy and which have simply been developed in anticipation of it.
Source: CIJ.World Research & Analysis Team
