Power modules

Reducing the TCO of generative AI datacenters

anchor

Overview

OptiMOS™ dual- and quad-phase power modules integrate OptiMOS™ power stages with proprietary inductors, capacitors and optimized packaging on a single substrate. This advanced integration allows multiple phases of a buck regulator to be deployed in 40 percent less area with 50 percent lower power loss compared to the equivalent discrete solutions. Infineon’s dual-phase modules utilize a patented inductor-on-top design for improved thermal and electrical performance yielding superior energy efficiency.

Key Features

  • Proprietary inductor-on-top design
  • Power density up to 2A/mm2
  • Up to 6% more efficiency than competitors
  • Up to 11C cooler junction temperature
  • Switch-nodes from motherboard
  • Common footprint
  • +2% more efficient at full load
  • -5°C cooler at full load
  • Minimized PDN losses
  • 80 A per phase thermally managed

Products

About

Lowering the total cost of ownership (TOC) of GenAI datacenters

Artificial intelligence is currently driving an exponential increase in global data generation and consequently increasing the energy demand of the chips that support this enormous data growth, now approaching a thermal design power of >1000 W.

As AI servers require three times more energy than traditional servers and as datacenters are already consuming more than 2 percent of the global energy supply, it is essential to find innovative power solutions and architecture designs that can lead to a measurable impact on global energy savings and minimizing total cost of ownership for datacenters.

Multi-phase power modules are critical to meet the power demands of high-power GPU systems by providing high power density and best-in-class efficiency without compromising power integrity. By locating modules close to the processor and reducing PDN losses, achieving >2000 A becomes feasible. Improving power efficiency at the core yields significant energy savings at scale. Since typical deployments consist of ~100,000 processors, saving 130W/processor translates into megawatts for a datacenter. This means millions of dollars saved over the system’s lifetime.

Why choose Infineon for computing solutions
  • Power semiconductor market leader in computing (Si and WBG technologies, BCD processes)
  • Cutting-edge technologies and advanced packing (multi-die embedding, modules, integrated magnetics)
  • System architecture (novel 48 V and 12 V topologies, magnetics, vertical power delivery, scalable architecture)
  • Smart control and software (hyper-transient digital controllers, hot-swap/e-fuse, point of loads)
Go to interactive block diagram
Use case: Infineon's solutions for green computing

Infineon components enable exceptional power usage effectiveness for datacenters. Our power stages provide the best power efficiency in the industry, while our power IC's high-temperature tolerance and excellent reliability enable operations at high ambient temperature, reducing the need for the energy-intensive external cooling. Supermicro collaborates with Infineon on green computing. Supermicro MicroBlade servers contain 28 digital multiphase controllers, 112 power stages, and 28 point-of-load controllers. In one use case, the end-customer of Supermicto's MicroBlade server saved 56 percent in datacenter space utilization, 45 percent in capex, and $13m/year in electricity. This led to the customer’s datacenter power usage effectiveness (PUE) increasing to 1.061. An ideal PUE value is 1.0, which means that all the power required for a datacenter is in the actual computing devices, not in overhead costs such as cooling or power conversion.

Read more about green energy in data centers here

Lowering the total cost of ownership (TOC) of GenAI datacenters

Artificial intelligence is currently driving an exponential increase in global data generation and consequently increasing the energy demand of the chips that support this enormous data growth, now approaching a thermal design power of >1000 W.

As AI servers require three times more energy than traditional servers and as datacenters are already consuming more than 2 percent of the global energy supply, it is essential to find innovative power solutions and architecture designs that can lead to a measurable impact on global energy savings and minimizing total cost of ownership for datacenters.

Multi-phase power modules are critical to meet the power demands of high-power GPU systems by providing high power density and best-in-class efficiency without compromising power integrity. By locating modules close to the processor and reducing PDN losses, achieving >2000 A becomes feasible. Improving power efficiency at the core yields significant energy savings at scale. Since typical deployments consist of ~100,000 processors, saving 130W/processor translates into megawatts for a datacenter. This means millions of dollars saved over the system’s lifetime.

Why choose Infineon for computing solutions
  • Power semiconductor market leader in computing (Si and WBG technologies, BCD processes)
  • Cutting-edge technologies and advanced packing (multi-die embedding, modules, integrated magnetics)
  • System architecture (novel 48 V and 12 V topologies, magnetics, vertical power delivery, scalable architecture)
  • Smart control and software (hyper-transient digital controllers, hot-swap/e-fuse, point of loads)
Go to interactive block diagram

Use case: Infineon's solutions for green computing

Infineon components enable exceptional power usage effectiveness for datacenters. Our power stages provide the best power efficiency in the industry, while our power IC's high-temperature tolerance and excellent reliability enable operations at high ambient temperature, reducing the need for the energy-intensive external cooling. Supermicro collaborates with Infineon on green computing. Supermicro MicroBlade servers contain 28 digital multiphase controllers, 112 power stages, and 28 point-of-load controllers. In one use case, the end-customer of Supermicto's MicroBlade server saved 56 percent in datacenter space utilization, 45 percent in capex, and $13m/year in electricity. This led to the customer’s datacenter power usage effectiveness (PUE) increasing to 1.061. An ideal PUE value is 1.0, which means that all the power required for a datacenter is in the actual computing devices, not in overhead costs such as cooling or power conversion.

Read more about green energy in data centers here

Design Resources Component

Developer community

{ "ctalist":[ { "link" : "https://community.infineon.com/t5/forums/postpage/choose-node/true", "label" : "Ask the community", "labelEn" : "Ask the community" }, { "link" : "https://community.infineon.com/t5/Forums/ct-p/products", "label" : "View all discussions", "labelEn" : "View all discussions" } ] }