At the product and solution launch event during MWC Barcelona 2026, Bob He, vice president of Huawei Digital Power, unveiled the GW-level AIDC solutions. These solutions help operators build a solid foundation for AI computing in the era of the agentic internet, driving intelligent and low-carbon transformation.

GW-level AIDC solution

The intelligent era has witnessed a surge in demand for computing power, driving the construction of AIDC facilities. However, this also brings challenges in reliability, energy consumption, delivery, and smooth evolution.

A single GPU now costs over 20000 USD, meaning the value of AI devices have increased 40 times, making any downtime extremely costly. A gigawatt-level AI data center consumes nearly ten billion kilowatt-hours of electricity annually, equivalent to the total annual power consumption of a city with one million residents, leaving energy efficiency as a key challenge.

Meanwhile, customers require faster time to market (TTM), shortening the delivery cycle from 18 to three months. For smooth evolution, AI computing facilities must be built to support continuous computing power upgrades over the next three-five years to keep pace with fast-evolving AI technology.

To address these challenges, the Huawei data center facility reconstructs the architecture of the four core subsystems of power supply, cooling, energy storage, and operations, as well as the construction mode to provide a highly reliable, energy-efficient, fast-delivered, and fully compatible AIDC solution, maximizing tokens per watt and powering the AI era forward.

• Power supply subsystem for AIDC: Diversified computing power drives diversified power supply system architectures for data centers. Low-density computing still relies on alternating current (AC) power, while direct current (DC) solutions will gradually take over in high-density data center scenarios with cabinet power density exceeding 300kW. Through innovations in power supply from the grid to components and by productizing power supply links, Huawei has developed a high-density, high-efficiency, and reliable integrated PowerPOD solution to facilitate rapid AIDC delivery.
• Energy storage subsystem for AIDC: As data centers are increasingly confronted with energy shortage challenges, grid-forming energy storage systems (ESS) emerge as a vital solution to ensure reliable power supply. Through innovations in energy storage from campuses to chips, Huawei has built a full‑lifecycle multi‑level storage architecture that has effectively addressed long‑term energy storage needs in AIDC and short‑term AI workload fluctuations.
• Cooling subsystem for AIDC: Liquid cooling systems have become a must-have for AI data centers. Through innovations in heat dissipation from components to the outdoor environment and by productizing cooling links, Huawei has built an efficient, reliable, and intelligent integrated IT POD solution to help AIDCs achieve efficient and highly reliable cooling.
• AI for DC intelligent operations: AI is leveraged to enhance data centers, improving AIDC safety and energy efficiency throughout the lifecycle. AI for safety: Fault prediction and identification facilitate the transition from passive response to predictive maintenance. AI for energy efficiency: Intelligent coordination and optimization of cooling and power supply reduce energy consumption.

Bob He stated that operators deal with soaring energy consumption, insufficient power capacity, and increasing safety risks in ICT facilities in the AI era.

Huawei will fully leverage its advantages in converged technologies and innovations across the energy, digital and power electronic fields to help operators build more reliable, agile and sustainable AI computing facilities, accelerate their intelligent transformation, and achieve continuous business success together.