Delta Electronics unveils energy-efficient AI hardware at NVIDIA GTC

The rapid rise of generative AI is resulting in an unprecedented increase in data center energy consumption.

Today’s data centers already use over 1% of the world’s electricity, but that share could triple by 2030 to power the huge AI models behind today’s chatbots, image generators and more.

Microsoft, Google, Amazon, NVIDIA and others plan to expand their data infrastructure plans in the approaching years, with power consumption for hyperscale and exascale data centers rising into the gigawatt range.

Traditional nuclear fission power, nuclear mergerGeothermal and solar energy will all play their part in sustaining this – but increasing efficiency in AI infrastructure is prime. This requires a coordinated response from your entire technology industry.

The recent NVIDIA GTC conference, some of the influential AI gatherings of the 12 months, featured a Taiwanese electronics company Delta Electronics discussed several cutting-edge advances to optimize energy consumption within the gigawatt data centers required to coach massive AI models.

Ralf Pieper, R&D director of Delta’s Custom Design Business Unit, said: “The introduction of gigawatt data centers for AI training and inference applications, likely including 250kW racks, is inevitable. To ensure highly efficient, reliable and seamless power conversion and delivery to AI computing chips, Delta continues to develop revolutionary solutions for various layers inside the grid-to-chip ecosystem.”

Delta’s latest data center hardware includes modular racks, high-efficiency power racks and breakthrough voltage regulators mounted vertically on GPU boards to attenuate power loss. Special:

• ORV3 racks and 50V DC bus bars efficiently distribute power in AI systems. They have shelves that support 8,000W of power and have as much as 97.5% efficiency.
• Delta’s DC-DC converters significantly improve GPU power delivery by converting power from 48V to 0.8V with high efficiency. A novel voltage regulator module minimizes power loss by integrating the facility module directly behind the GPU chips, achieving power savings of as much as 15%.
• Modular solid-state transformers convert mains power on to 800 VDC with high efficiency and, because of their modular structure, are designed to be adaptable to varied international power levels.
• Uninterruptible power supply (UPS) systems ensure continuous operations in AI data centers, with a formidable 97.5% efficiency to keep up efficiency even when operating above normal capability.
• Designed for AI data centers, the 80 kW power shelf features two 40 kW rails and uses isolated DC/DC converters for efficient power conversion from 800 VDC to 50 VDC, increasing overall efficiency to 98%.

Founded in 1971, Delta is a worldwide leader in power supplies and thermal management. The company has a powerful portfolio of energy-saving solutions for data centers, renewable energy, electric vehicle charging and more.

With a mission to “Provide revolutionary, clean and energy efficient solutions for a greater future,” Delta leverages its expertise in high-efficiency power electronics and ESG-driven business model to deal with key sustainability challenges.

Throughout its history, Delta has received quite a few global awards for its business achievements, revolutionary technologies and commitment to ESG.

This incorporates Listed and represented within the Dow Jones Sustainability World Index for 12 years in a row the Carbon Disclosure Project (CDP)s A-List in 2020, 2022 and 2023 for contributions to addressing climate change and water security issues.

Pieper explained Delta’s deal with energy-efficient AI hardware: “Our unique expertise in high-efficiency server performance and DC/DC converters, in addition to ICT and energy infrastructure, allows us to advance the event of breakthrough solutions that may support the AI ​​megatrend” through optimization the grid-to-chip power conversion cycle.”

Delta estimates that its high-efficiency products have already helped customers save nearly 40 billion kilowatt hours (kWh) since 2010, enough to ppower roughly 3.73 million average American homes for a 12 months.