IDC's Worldwide Performance-Intensive Computing (HPC, AI, and Analytics) Infrastructure and Services Taxonomy, 2025

Description

This IDC study lays the groundwork for sizing and forecasting the performance-intensive computing (PIC) competitive market. PIC combines infrastructure approaches across two major and fast-growing composite workloads: modeling and simulation (M&S aka HPC) and artificial intelligence (AI). It also combines infrastructure on premises and off premises and self-managed and as-a-service deployments. As a compute, storage, and connectivity paradigm, PIC offers the most powerful and efficient way to execute mathematically intensive, very complex instructions or to perform a relatively simple instruction on massive amounts of data."In the past few years, a common infrastructure paradigm has emerged, thanks to the importance of mathematically intensive computations in many use cases found in digital organizations," said Madhumitha Sathish, research manager at IDC's Infrastructure Systems, Platforms and Technologies Group. "IDC believes that performance-intensive computing offers an opportunity for vendors to approach these use cases in a more streamlined and targeted fashion."

Product Code: US53125225

IDC's Worldwide PIC (HPC, AI, and Analytics) Infrastructure and Services Taxonomy

PIC (HPC, AI, and Analytics) Infrastructure and Services Taxonomy Changes for 2025

Taxonomy Overview

Advice for the Technology Supplier
- Performance at All Costs
- Package Performance-Intensive Computing Software Stacks
- Low-Latency Interconnect and Network Support
- Hybrid Approaches

Definitions

What Is Performance-Intensive Computing?
- Performance-Intensive Computing Workloads
  - Composite Workloads
- Alignment with IDC's Enterprise Workloads
Performance-Intensive Computing Infrastructure Segments
Infrastructure Procurement and Deployment
- Vendor Type
- Buyer Type
- Control Plane
- Deployment Location
- Management Type
  - Procurement Type
Infrastructure Hardware
- Infrastructure Hardware: Compute
  - Coprocessors (Accelerated Computing)
- Infrastructure Hardware: Storage
  - Internal Storage (Server-Based Storage Platforms or Storage-Intensive Servers)
- Converged Infrastructure
  - Integrated Infrastructure
  - Hyperconverged Infrastructure
  - Composable Infrastructure
- Infrastructure Software
  - Physical and Virtual Computing Software
  - Storage Software
  - Systems Management Software
  - Other Infrastructure Software
  - PIC Dedicated and Public Cloud Services
  - PIC Infrastructure - Environmental Attributes
  - Workload Abstraction
    - Bare Metal
    - Virtualized
    - Containerized
  - Workload Profile
    - Compute Intensive
    - Memory Intensive
    - Data Intensive
    - Network Intensive
    - Hybrid
  - Parallelization (Cluster)
    - Traditional
    - Cloud Native
  - Scaling (Cluster)
    - Level 1: Intra-Node (Processor)
    - Level 2: Intra-Node (Accelerator)
    - Level 3: Inter-Node (Cluster)
    - Level 4: Inter-Node (Datacenter)
    - Level 5: Inter-Node (Geodispersed)
  - Size (Cluster)
  - Node-to-Node Communications Protocol (Cluster)
  - HPC and AI-Specific Attributes
    - HPC-Specific Attributes
    - AI-Specific Attributes

Related Markets

Learn More

Related Research
Synopsis