The POWER9 advantage

Built on open standards with accelerated bandwidth and compute options, the latest addition to the POWER® processor family delivers blazing speed with up to 1.5x the performance of its POWER8® predecessor. Built for data-intensive workloads, POWER9™ is the only processor with state of the art I/O subsystem technology, including next-generation NVIDIA NVLink, PCIe Gen4 and OpenCAPI.

What can POWER9 mean for you?

Enhanced core and chip architecture for next-gen workloads

  • Optimized for emerging algorithms to interpret and reason
  • Superior bandwidth, scale and capacity to ingest and analyze data

Processor family with optimized silicon for a range of platforms

  • Scale out for HPC and next-gen apps
  • Scale up to 16 sockets for the most demanding enterprise workloads

Premier platform for accelerated computing with maximized bandwidth

  • Ultra-high bandwidth CPU-GPU interconnect delivers ~ 4x faster training
  • OpenCAPI™ interface offers high bandwidth communication to NICs, FPGA accelerators, and storage controllers

Some compelling numbers

Power Systems™ are built for the most demanding, data-intensive, reliably secure computing on earth. With 2.6x more RAM and 1.8x more memory bandwidth than x86, these cloud-ready servers help you obtain greater insight from your data pipeline, so that you can manage mission-critical data as well as operational data stores and data lakes. From I/O and memory bandwidth to high-performance cores and RAM, see how the cloud-ready, scalable POWER9 servers match up to x86.


max I/O bandwidth vs. x86


high performance cores vs. x86

IBM’s POWER9 is literally the Swiss Army knife of ML acceleration as it supports an astronomical amount of IO and bandwidth, 10X of anything that’s out there today

Patrick Moorhead, principal analyst, Moor Insights & Strategy

Technical details

Emerging analytics, AI and cognitive

  • New core for stronger thread performance
  • Delivers 2x compute resource per socket
  • OpenPOWER solution enablement
  • Extreme processor performance (accelerator bandwidth and reduced latency)


  • Highest bandwidth GPU attach
  • Advanced GPU/CPU interaction and memory sharing
  • High bandwidth direct attach memory
  • Coherent memory and virtual addressing capability
  • Open and flexible solutions for a broad range of deployments and platforms


  • Power / packaging / cost optimizations for a range of platforms
  • Superior virtualization features: security, power management, QoS, interrupt
  • Industry-leading I/O subsystem with PCIeG4 to deliver 2x bandwidth
  • The ideal platform for hardware acceleration and innovative developers


  • Large, flat, scale-up systems up to 16 sockets
  • Buffered memory for maximum capacity (up to 4 TB / socket)
  • Leading RAS
  • Over 120 MB of last level cache (>3x Skylake LLC size)
  • State of the art I/O and acceleration attachment signaling
  • Promotes ecosystem enablement

CORAL supercomputers: fueled by POWER9

IBM, Nvidia, Mellanox and Red Hat have worked together to create Summit and Sierra, state of the art supercomputers built with everyday IBM POWER9 processors and IBM Storage that will run traditional HPC workloads and AI workloads. They each have the processing power to handle more than 200 quadrillion calculations per second, or roughly 500,000 times the number of stars in the Milky Way. They were not developed as highly customized, one-off systems seen in previous HPC solutions.


Join the Power Systems user community

Learn, share and compare notes with fellow Power Systems users, as well the latest Power Systems community news.

Breakthrough AI performance

IBM sets tera-scale machine learning benchmark record with POWER9 and NVIDIA® V100™. Available for customers to try as part of the PowerAI Tech Preview portfolio later this year.

Visit the IBM Portal for OpenPOWER

Help shape the future of OpenPOWER innovation – and see what’s new with POWER9, NVIDIA NVLink, CAPI and more.