Business AI Compute Cost Checklist

Planning a substantial AI project or reviewing cloud AI costs? Use this checklist before committing to hardware, reserved capacity or a long-term provider contract.

Workload

• □ Define the business purpose of the AI workload
• □ Identify whether the workload is training, inference or both
• □ Estimate usage volume and how variable demand may be
• □ Identify minimum performance requirements

Hardware

• □ Confirm suitable GPU types for the workload
• □ Check memory requirements
• □ Compare alternative hardware types
• □ Consider hardware obsolescence and next-generation timing

Provider

• □ Compare several providers
• □ Check available region and data-centre location
• □ Check uptime guarantees and historical availability
• □ Check technical support
• □ Check security, compliance and data-residency support

Price

• □ Compare on-demand pricing
• □ Compare reserved or committed pricing
• □ Include storage costs
• □ Include networking costs
• □ Include support costs

Flexibility

• □ Check minimum commitment period
• □ Check scaling rules and increments
• □ Check exit terms and early-termination penalties
• □ Check workload portability between providers

Energy and Location

• □ Check energy assumptions and electricity cost
• □ Check data-centre location
• □ Check data-residency requirements
• □ Check environmental reporting availability

Financial Products

• □ Understand the benchmark before relying on any financial product
• □ Confirm regulatory status of any financial contract
• □ Confirm whether settlement is cash or physical
• □ Obtain specialist financial and legal advice