The global skills and competency framework for the digital world

#1614 Embed AI Infrastructure, Accelerator Utilisation, and High-Density Compute Operations into ITOP, IFDN, and CPMG Skills [GreenOps] change request pending

Amend ITOP, IFDN, and CPMG to include specific competency dimensions for managing high-density AI infrastructure, accelerator (GPU/TPU) utilisation, workload placement trade-offs between training and inference, and dynamic capacity planning.

Problem Statement

CR Raised in consultation with global GreenOps industry groups.

SFIA 9 Infrastructure Operations (ITOP), Infrastructure Design (IFDN), and Capacity Management (CPMG) cover traditional physical, virtual, and cloud infrastructure, but lack explicit competencies for the unique operational demands of Artificial Intelligence (AI) and machine learning workloads. AI training and inference introduce extreme power density, specialised accelerator hardware (GPUs, TPUs, NPUs), complex liquid/air cooling constraints, and severe carbon implications. Without explicit level descriptors, infrastructure managers and architects lack a standardised competency benchmark for managing AI workload placement and power-efficient accelerator operations.

 

Proposed Change

Amend skill descriptors at Levels 3–6 across ITOP, IFDN, and CPMG:

  • IFDN (Levels 4–6): Incorporate workload placement and architecture trade-offs for AI training vs. inference, evaluating accelerator selection, rack density limits, power availability constraints, and specialised cooling requirements.
  • ITOP (Levels 3–5): Add operational monitoring and optimisation of accelerator (GPU/TPU) utilisation, removing idle compute capacity in AI clusters, and managing dynamic IT load against thermal/cooling headroom.
  • CPMG (Levels 3–5): Incorporate capacity planning based on actual workload demand rather than provisioned headroom for high-density AI workloads, balancing training queue scheduling with energy availability.

Rationale

  • Directly addresses emerging technology demands and power/cooling constraints flagged in the GreenOps framework.
  • Fills the competency gap between general cloud infrastructure management and high-performance AI compute operations.
  • Focuses on durable SFIA 10 principles: workload placement, accelerator efficiency, and actual vs. provisioned capacity. 

Note: More than 36 jurisdictions globally are actively adopting or aligning their corporate disclosure regulations with the IFRS Sustainability Disclosure Standards (ISSB) to create a global baseline for climate and sustainability reporting. Beyond the UK, the EU, and Australia, numerous other countries have formally finalised their approach to adopt or closely align with these rules, including Brazil, Canada, Chile, China, Hong Kong, Japan, Malaysia, Mexico, Nigeria, Qatar, and Turkey. Additionally, nations such as Singapore, South Korea, India, and Indonesia are in advanced stages of implementing ISSB-aligned frameworks. View official implementation maps and country-by-country data on the IFRS Foundation Jurisdictional Profiles page.

Proposed change applies to Infrastructure operations

Current status of this request: pending