62.5kW High-Efficiency Power Module Solutions for AI Data Center Upgrades

With the explosive growth of generative AI, data centers are facing unprecedented power pressures. The International Energy Agency (IEA) predicts that by 2030, global data center electricity consumption will double to 945 terawatt-hours (equivalent to Japan's current total electricity consumption), with AI applications being a key source of this demand.

✦ Surging Power Demand for AI

In 2024, data centers accounted for 1.5% (415 terawatt-hours) of global electricity consumption, but by 2030, this proportion is expected to jump to 4%, with demand from AI data centers projected to increase more than fourfold.

The United States accounts for 45% of global data center electricity consumption, and China accounts for 25%, driving global power demand growth.

Traditional server cabinets operate at a mere 10 kilowatts, while demand for AI server cabinets has soared to 60-120 kilowatts, with even 1 megawatt configurations already in the pipeline, posing extreme challenges to power supply systems.

✦ Power Bottlenecks Constraining Development

The surge in data center electricity consumption could delay 20% of planned projects worldwide due to power shortages. At the same time, electricity costs account for over 40% of AI data center operating costs, becoming the largest expense item.

Traditional power modules have an efficiency of only 94-96%. However, under high AI loads, every 1% improvement in efficiency can save millions in electricity costs and significantly reduce carbon emissions.

✦ Boosted by high-efficiency power modules

To address the above challenges, the MXR150062B-DC2 power module is designed specifically for AI data centers.

▸ 62.5kW High Power Output: A single module supports high-density cabinets, matching the power evolution of AI servers from 60 to 120kW.

▸>98% Peak Efficiency: Far exceeding traditional solutions, directly reducing electricity costs and alleviating heat dissipation pressures.

▸ 800V-1500V Wide Input Voltage: Adapts to the trend toward high-voltage direct current (HVDC) power supply, improving energy transmission efficiency.

▸Air-cooled design and 300-380V output: Balances maintenance costs and compatibility, precisely matching the voltage requirements of GPU servers.

With the explosive growth of generative AI, data centers are facing unprecedented power pressures. The International Energy Agency (IEA) predicts that by 2030, global data center electricity consumption will double to 945 terawatt-hours (equivalent to Japan's current total electricity consumption), with AI applications being a key source of this demand.