According to the 2026 Global Data Center Outlook report released by Jones Lang LaSalle, the data center industry continues to experience robust growth, signifying that computing power infrastructure is becoming a strategic real estate asset class on par with office space and manufacturing plants. The report forecasts that global data center capacity will surge from 103GW to 200GW by 2030, representing a near doubling. Within this expansion, the Asia-Pacific region's capacity is projected to grow from 32GW to 57GW, achieving a compound annual growth rate of 12%. Hyperscale enterprises, employing a dual-track strategy of building their own facilities combined with leasing, are identified as the primary drivers of this growth.
Yao Yao, Head of Research for Jones Lang LaSalle China, commented that the Chinese data center market has undergone rapid expansion in recent years, signaling its entry into a super investment cycle alongside global trends. However, he also noted that energy infrastructure constraints—particularly limitations in grid capacity and supply chain risks—are posing challenges to development. Investors and enterprises must strike a balance between market speed and capital efficiency, accurately anticipate demand inflection points, and maintain sufficient flexibility to adapt to the rapid evolution of AI models and application scenarios.
Data from the China Academy of Information and Communications Technology (CAICT) indicates that in 2024, national data center electricity consumption exceeded 166 billion kilowatt-hours, accounting for approximately 1.68% of total societal electricity use. This represented a 10.7% year-on-year increase, significantly outpacing the 6.8% growth in overall societal electricity consumption. CAICT projections suggest that under a high-growth scenario driven by an AI boom, China's computing center electricity consumption could surpass 700 billion kWh by 2030, constituting 5.3% of total societal usage. In a lower-growth scenario with slower AI adoption, consumption would still reach around 300 billion kWh, or 2.3% of the total. Overall, the proportion of societal electricity consumed by data centers is rising.
To meet security, compliance, and performance demands, enterprises are increasingly adopting hybrid architecture models that integrate on-premise facilities, third-party colocation services, hyperscale cloud platforms, and distributed edge nodes. This combination allows companies to: retain sensitive core business operations in on-premise data centers; migrate stable, foundational workloads that don't require proprietary builds to flexible colocation spaces; leverage the elasticity and advanced services of hyperscale clouds for variable workloads and innovative applications; and deploy computing power at the network edge to meet ultra-low latency requirements for scenarios like IoT and real-time interaction. Under this model, companies are downsizing their on-premise footprints while retaining the capability to process critical workloads locally. Consequently, proactive planning is essential—for instance, integrating grid and distributed generation resources into early-stage corporate real estate decisions to secure long-term power capacity; using pre-leasing or phased leasing to hedge against rental and power scarcity risks; designing facilities with upgradeability for high-density computing and liquid cooling systems; and employing modular architectures to enhance resilience against construction delays, equipment delivery lags, and cost inflation, thereby achieving a dynamic balance between cost efficiency and operational resilience.
The Jones Lang LaSalle report anticipates that by 2030, AI workloads will constitute half of all global data center capacity, with inference workloads surpassing training as the dominant demand driver around 2027. Currently, AI accounts for roughly a quarter of all data center workloads. This significant shift is reshaping data center technology and spatial requirements, as AI training and inference demand far greater computing power and energy consumption than traditional loads. Power density per rack is escalating to 40-100+ KW, far exceeding the current average of under 20 KW. When density exceeds 20 KW, traditional air cooling becomes ineffective, making liquid cooling a necessity. Thus, the transition of liquid cooling from an "option" to a "standard feature" will be a notable change. Given that AI-dedicated data centers are major energy consumers, energy transition is an inevitable trend for future construction. The report notes that solar power is expected to account for nearly half of global renewable energy capacity by 2026. Simultaneously, the Asia-Pacific region is poised to lead the shift towards renewables in data centers, with nearly 4000GW of global solar capacity projected to originate from the region by 2030, with China contributing approximately 80% of this growth. This aligns with the broader energy transition landscape: according to the International Energy Agency, China maintains a dominant position in global new renewable capacity additions and is expected to contribute 60% of the world's new renewable power generation by 2030.
The data center construction boom has raised concerns about a potential bubble in the sector. However, the report indicates that relevant real estate metrics suggest the industry remains healthy. As of the end of 2025, global data center utilization rates stood at a high 97%, and 77% of capacity under construction was pre-leased, showing no signs of excessive vacancy or disorderly expansion. Furthermore, market demand is not solely reliant on AI; data generation and storage, network traffic, and cloud service revenue all maintain long-term compound annual growth rates exceeding 20%. These sustained demand factors are expected to drive structural upgrades and long-term expansion in the industry, rather than a transient, frothy boom.
From an investment perspective, data centers are currently one of the most attractive asset classes within commercial real estate. Since 2020, global direct investment transactions in data centers have reached approximately $75 billion, while merger and acquisition activity has surpassed $300 billion. Core fund sizes dedicated to data centers globally are projected to exceed $50 billion this year, with target returns above 10%. According to CAICT data, the market size of China's data center industry grew from RMB 87.8 billion in 2019 to RMB 304.8 billion in 2024, achieving a CAGR of 28.26%. As AI computing demand surges, underlying infrastructure needs continue to expand. In the first half of 2025, newly delivered data center capacity in China already exceeded the total for previous full years. By the end of 2024, the scale of operational data center racks in China reached 9 million (standard 2.5kW/rack), an 11.11% year-on-year increase. The compound annual growth rate from 2017 to 2024 was 23.35%, though the growth rate moderated in 2024, potentially due to a larger base and industry-wide optimization of computing resources. By the end of June 2025, the rack count had increased to 10.85 million, adding 1.85 million racks since end-2024, corresponding to an additional IT load of approximately 4.63GW.
Globally, traditional bank financing is struggling to meet the capital demands of rapid industry expansion, making asset securitization a critical pathway. The issuance of ABS and CMBS with data centers as underlying assets has nearly doubled annually since 2020 and is expected to reach $50 billion by 2026. In China, August 2025 saw the listing of the first batch of data center public REITs—China Southern Wanguo Data Center REIT and China Southern Runze Technology Data Center REIT—on the Shanghai and Shenzhen stock exchanges, respectively. Recently, two institutional REITs (holding-type real estate ABS) for 21Vianet's IDC assets also completed successful issuances. These developments validate the report's view that securitization products with data centers as underlying assets will see significant growth in the Asia-Pacific region, including China, reflecting long-term capital confidence in this asset class.