Stock Markets June 9, 2026 09:17 AM

Citi Identifies winners from China’s planned AI data center buildout

Bank highlights data center operators and localized server suppliers set to benefit from a multi-trillion-yuan national initiative

By Leila Farooq
Share
Twitter Reddit Facebook LinkedIn
GDS VNET

China's reported plan to deploy roughly 2 trillion yuan over five years to build a nationwide AI-ready data center network is likely to lift demand across data center operators and a localized AI server supply chain, Citi says. The National Development and Reform Commission-led program will be run mainly by state-owned operators and requires at least 80% use of domestic AI chips, an arrangement Citi described as "a monumental catalyst for China's digital infrastructure." The bank pointed to specific beneficiaries on both the facilities and equipment sides of the market.

Citi Identifies winners from China’s planned AI data center buildout
GDS VNET
Summarize with
ChatGPT Perplexity Claude Grok Gemini

Key Points

  • China is reportedly planning to invest about 2 trillion yuan over five years to build a national AI-focused data center network, led by the NDRC and primarily run by major state-owned operators.
  • Citi identifies GDS and VNET as direct beneficiaries on the data center side, citing roughly 200MW of new bookings for GDS and over 500MW for VNET in Q1 2026.
  • On the supply chain front, Citi names IEIT, Lenovo and ZTE as key local AI server ecosystem proxies, with Accelink and Chinasoft also highlighted for component and software roles.

China is reportedly preparing to commit about 2 trillion yuan - roughly $295 billion - over the next five years to establish an extensive national data center network focused on AI workloads, and Citi has flagged a set of companies it expects to be among the primary beneficiaries.

The program, led by the NDRC, will be chiefly operated by major state-owned operators and includes a requirement that at least 80% of deployed AI chips be domestically sourced. Citi described the initiative as "a monumental catalyst for China's digital infrastructure."

On the facilities side, Citi pointed to two listed data center operators - GDS and VNET - as already showing momentum ahead of the rollout. The bank cited record wholesale bookings in the first quarter of 2026, with GDS recording about 200 megawatts of new bookings and VNET signing in excess of 500 megawatts of new orders during the period. Citi said these trends are directly aligned with the accelerating demand the national plan is expected to generate.

Using an assumed all-in unit cost of roughly 200 million yuan per megawatt, Citi calculated that the investment implies approximately 10 gigawatts of additional AI data center capacity over five years - an average pace of about 2 gigawatts per year. Against an estimated current installed base of 20 gigawatts of AI-grade capacity, that incremental build would represent an expansion of roughly 50%.

Beyond operators and colocators, Citi highlighted beneficiaries along the localized supply chain. The bank identified IEIT, Lenovo and ZTE as key plays representing a local AI server ecosystem proxy, noting that the domestic chip mandate is expected to drive AI server demand for these companies, which the report describes as supply-constrained in their server businesses.

Additional names cited by Citi include Accelink - singled out as a leading domestic optical transceiver maker - and Chinasoft, described as a partner to Huawei in ITS and ISV capacities. Together, the bank's list spans both the data center infrastructure and the equipment and software ecosystem that supports server deployments and networking.

The plan's scale and the explicit push for localized chips and components create a concentrated demand signal for infrastructure and suppliers. Citi's capacity and cost assumptions are central to its estimate of the program's output, and its selection of favored names reflects companies that stand to capture direct orders or component demand tied to the initiative.


Key beneficiaries named by Citi:

  • Data center operators: GDS, VNET
  • Server and equipment ecosystem: IEIT, Lenovo, ZTE
  • Component and software partners: Accelink, Chinasoft

Risks

  • The estimate of roughly 10 gigawatts of new AI data center capacity depends on an assumed all-in unit cost of about 200 million yuan per megawatt - if actual costs differ, capacity outcomes would change.
  • The plan mandates at least 80% reliance on domestic AI chips, which could put pressure on local chip production capacity and affect the pace at which AI servers and data centers can be deployed.
  • Citi describes certain server businesses as supply-constrained, indicating that supplier capacity limitations could restrain companies identified as beneficiaries from fully capturing demand.

More from Stock Markets

Survey: Oral GLP-1 Obesity Pills Gain Early Traction but Use Remains Limited Jun 9, 2026 Newell Brands Jumps 8.1% After Q1 Beat and Upgraded Guidance Jun 9, 2026 SpaceX IPO Demand Tops $250 Billion, Oversubscription Nears Fourfold Jun 9, 2026 HSBC Identifies Top Picks in Asia’s Emerging Markets Focused on AI, Energy Transition and Healthcare Jun 9, 2026 Big Tech and Small Caps See Wide Swings; Marvell, Arm Slide While Credicorp and Nuvalent Rally Jun 9, 2026