Great Wall Power Integrates InnoGaN Technology to Reconstruct AI Data Center Energy Systems

Great Wall Power has adopted Innoscience's Gallium Nitride (InnoGaN) technology in its titanium-level power supplies for AI data centers, achieving an ultra-high power conversion efficiency exceeding 96%, surpassing the global top-tier 80PLUS Titanium energy efficiency standard.

With the advancement of applications like artificial intelligence, cloud computing, and data mining, the demand for computing power in data centers has increased significantly. According to data from the International Energy Agency (IEA), global electricity consumption by data centers is projected to reach 3,000 TWh by 2030, accounting for 10% of the world's total electricity consumption. In comparison, data center power consumption is expected to be around 4% globally in 2025.

A key factor limiting the substantial growth of data center power consumption is power conversion efficiency. Traditional silicon-based power supplies have lower conversion efficiency, leading to power losses of up to 10%. For a standard 10-megawatt data center, this translates to an annual waste of 9 million kilowatt-hours.

Every 1% improvement in power conversion efficiency can save hundreds of millions of RMB in electricity costs. The higher efficiency of GaN-based power supplies helps eliminate this waste. Additionally, GaN enables higher system power density, reducing Bill of Materials (BOM) costs and minimizing waste of BOM components ("dematerialization"). This also contributes to lower greenhouse gas emissions.

Addressing the energy consumption challenges in server power supplies, Innoscience pioneered the launch of the E-GaN power IC series (ISG612xTD SolidGaN) in To-247-4 packages, which integrates gate drivers and short-circuit protection. With a withstand voltage of 700V and Rdson ranging from 22~59mΩ, this series features a precision Vgs gate driver, fast short-circuit protection, and excellent thermal performance. It meets the high-frequency switching requirements for Titanium Plus efficiency, doubling power density compared to traditional solutions.

According to the OCTC "White Paper on High-Power-Density Server Power Supply Modular Design (2024)", during typical load conditions of 20%-50%, which account for 80% of server runtime, GaN-based titanium-level power supplies can maintain stable conversion efficiency above 95.5%-96%, effectively avoiding hidden energy losses due to "efficiency cliffs". Great Wall's server power supplies are the first to adopt Innoscience's packaged chips ISG6122TD and ISG6123TD. Compared to traditional power supplies, they reduce light-to-medium load power losses by at least 30%, achieving an improvement of up to 4 percentage points in the 20%-50% typical load range, with conversion efficiency exceeding 96%. It is estimated that adopting the GaN-based titanium-level power supply solution can save over 2 million RMB annually in electricity costs per ten thousand servers, reduce heat generation by 50%, and lower air conditioning energy consumption by 18%. This significantly promotes the breakthrough of intelligent computing center PUE below 1.2, achieving dual benefits of "energy saving + heat reduction".

Through collaborative efforts, Great Wall Power and Innoscience are rapidly advancing the industrialization and widespread adoption of GaN-based titanium-level power supplies, realizing the leap from concept to application of GaN technology in the AI data center field. This helps transform the power supply system of intelligent computing centers from component innovation to energy system reconstruction. Currently, Innoscience has partnered with numerous data center industry players, developing comprehensive products and solutions. It is believed that in the near future, fully GaN-based data center systems across the entire chain will be achieved, contributing to energy conservation and carbon reduction in data centers.