Deloitte's 2026 forecast for inference share of AI compute (½ in 2025, ⅓ in 2023)
~2/3forecast
| Value kind | forecast — Forecasts are time-bound outlooks, not specifications; test decisions across plausible scenarios. |
|---|---|
| Scope | Deloitte’s November 2025 prediction for inference as a share of AI compute in 2026. Not an observed fleet share, installed capacity, energy or instantaneous electrical draw. The forecast does not allocate an individual fleet. |
| Caveat | A forecast for calendar 2026, not an observed 2026 outcome. |
| As of | 2025-11-18 |
| Source | Deloitte TMT Predictions 2026, “More compute for AI, not less,” published 2025-11-18. · Opening section: inference accounts for roughly two-thirds of 2026 AI compute, up from one-third in 2023 and half in 2025 |
| Review | checking…review by 2027-02-01 · standard cadence |
| Recorded changes | last 2026-09-16 · 3 revisions tracked |
| Claim id | inference-share-of-ai-compute-in-2026-in-2025-3 |
Where the guide uses it
- 0.1 Orientation: The AI Data Center as a Single Co-Designed Machine
- 0.2 How to Read This Guide: Decisions, Consequences & Reference Data
- 1.1 The Archetype Decision Framework: Workload Is the Master Variable
- 1.3 Inference Data Centers: Bursty, Distributed, Always-On
- 1.5 Edge Inference & Distributed Micro-Datacenters
- 10.3 Multi-Tenancy, Isolation & Resource Sharing
- 10.11 Inference Serving Engineering: SLOs, Batching, Disaggregation & Goodput-Optimal Scheduling
- 16.3 Software, Orchestration & Efficiency at the Frontier
← Full numbers register — every date-stamped figure in the guide, with revision history.