Artificial intelligence news.
Straight from the original announcements.

ZTE Unveils 800V Power and Liquid Cooling Systems for AI Data Centers

November 16, 2026, 11:03By the AI News Bites Editorial TeamSource: PR Newswire
ZTE Unveils 800V Power and Liquid Cooling Systems for AI Data Centers
Image: PR Newswire

ZTE has outlined an energy infrastructure strategy for AI data centers, combining an 800V high-voltage DC system with a megawatt-scale liquid cooling CDU. The company says the approach is designed to reduce power losses, support high-density computing and improve the efficiency of converting electricity into AI output.

Key points
  • ZTE presented an 800V high-voltage DC system for high-density AI infrastructure.
  • Its liquid cooling CDU provides 1.5 MW per unit and can expand to 2.4 MW.
  • The company says the combined systems can keep PUE within 1.15.
  • A Nanjing data center upgrade reached power density of up to 60 kW.
  • Karakoram-01 uses liquid cooling and modular power technologies in Pakistan.

Rising rack power is changing data center design

ZTE says the benchmark for AI data centers, or AIDCs, is shifting from computing scale to computing efficiency as GPU deployments increase power demand. Traditional data center racks typically consume 10, 20 kW, while high-density AI deployments can exceed 100 kW, reach 600 kW or rise to 1 MW.

The company says conventional power architectures lose energy through multiple conversion stages between the transformer and server. Cooling is also becoming a constraint. According to ZTE, air cooling has an engineering limit of approximately 20, 30 kW per rack, making liquid cooling a mainstream solution for denser GPU clusters.

800V DC system targets power losses

ZTE's first infrastructure component is an 800V high-voltage DC system. The company says its topology design, DC/DC intelligent battery management, application modes and scalability are intended to provide high-density, efficient and reliable DC power for computing infrastructure.

The system supports full-link management and system visibility, including fault recording, temperature measurement at key locations and selective protection simulation. ZTE says predictive maintenance covers module faults, system faults and battery lifespan.

With dual-standard compatibility, the system can switch between ±400V and 800V. Its modular design supports capacity expansion as required, which ZTE says can reduce initial investment pressure. The company presents it as suitable for centralized computing bases and distributed edge upgrades, serving operators, internet companies and professional computing power investors.

Liquid cooling reaches megawatt capacity

ZTE has also launched a cold plate liquid cooling CDU for equipment rooms. The company says one unit provides 1.5 MW of cooling capacity and can be expanded to 2.4 MW, with configurations using two to five CDU cabinets.

The 1500 kW CDU combines sensors for flow, pressure and temperature. According to ZTE, it provides temperature control of ±0.5°C and pressure control of ±0.05 bar, alongside health indicators and status visualization for operations and maintenance. N+X redundant hot backup and dual power feeds are designed to support 7×24-hour operation.

ZTE says combining the 800V system with the 1500 kW CDU can keep power usage effectiveness, or PUE, within 1.15. At the Nanjing Binjiang Data Center, the company reports that a liquid-cooling integrated upgrade reached power density of up to 60 kW and achieved a PUE of 1.15.

A broader infrastructure approach

ZTE says AIDC infrastructure must coordinate computing, power, cooling and operations rather than develop each area separately. Its approach combines source-grid-load-storage integration, high-voltage DC power, liquid cooling, battery energy storage, and iDCIM and iEMS management systems.

The company also describes orthogonal SuperPod AI computing servers developed with industry partners, as well as software for training and inference optimization. Its green energy offering includes smart PV-storage solutions for PV power stations, grid-side energy storage, AIDC energy storage and PV-storage microgrids.

ZTE recently launched Karakoram-01 with partners, which it describes as Pakistan's first and largest purpose-built, AI-ready data center campus. The site uses cold plate liquid cooling, prefabricated power modules and smart busway to support elastic cooling and power supply. For users and data center operators, the announcement points to an infrastructure focus on handling higher AI rack densities while controlling energy use, cooling requirements and expansion costs.

Related news

ZTE Targets AI Data Center Efficiency With 800V Power and Liquid Cooling
Chips and infrastructureNovember 16, 2026, 11:33

ZTE Targets AI Data Center Efficiency With 800V Power and Liquid Cooling

ZTE is positioning an 800V high-voltage DC system and a megawatt-level liquid cooling CDU as the foundation for more efficient AI data centers. The company says the…

SambaNova and Intel demonstrate air-cooled AI inference system
Chips and infrastructureNovember 16, 2026, 05:47

SambaNova and Intel demonstrate air-cooled AI inference system

SambaNova and Intel are demonstrating an air-cooled inference system that pairs Intel GPU D-Series processors for prefill with SambaNova SN50 RDUs for decode. The…

Intel and RadixArk Optimize SGLang for Open AI Inference
For developersNovember 12, 2026, 21:30

Intel and RadixArk Optimize SGLang for Open AI Inference

Intel and RadixArk are collaborating to optimize SGLang, an open-source AI inference stack, for Intel platforms. The companies say the work aims to help developers and…