The exam blueprint
The exam has 30 scenario questions, 5 from each module, and you have 45 minutes. Every learning outcome below is tested by at least one question; the fifth question in each module is drawn at random. The pass mark is 75%, so you can miss up to 7 questions. Each sitting draws a new set from a bank of 150. Your result shows your score for each module, not the answers to individual questions.
| Module | Questions | Weight | Capstone use |
|---|---|---|---|
| 1. Why AI economics is different | 5 | 17% | Vocabulary for the memo |
| 2. What AI really costs | 5 | 17% | From rate card to fully loaded cost, then the self-hosting test |
| 3. Allocating and reporting AI cost | 5 | 17% | Allocate a month of shared GPU cost |
| 4. Measuring AI value | 5 | 17% | Compute attribution coverage for a portfolio |
| 5. Building the business case | 5 | 17% | Stress-test a case against price and demand |
| 6. Governing the portfolio | 5 | 17% | Redesign before you scale: cost per successful outcome |
Module by module
Use the outcomes as a checklist. If you can do each one without notes, and explain the worked example in your own words, you are ready for that module's questions.
Module 1
Why AI economics is different
You can
- Explain the AI value gap and the two-meter asymmetry behind it.
- Distinguish the token as a unit of consumption from the outcomes that carry value.
- Trace a workload through production, consumption, value and decision.
- Place an organisation on the five levels of AI economics maturity and name its next move.
Terms to know
AI Value Gap · Token · Token Productivity · Jevons Paradox · AI Value Management
Worked example
One workload through the whole chain: Without the full chain, this team would have reported 3,300 hours saved and asked to scale.
Readings
The AI Value Gap · The Token Is the Meter, Not the Value · The AI Economics Map · The Five Levels of AI Economics
Module 2
What AI really costs
You can
- Name the seven layers of AI TCO and the costs most often left out of a business case.
- Count each cost once across SaaS, APIs, agents and owned infrastructure.
- Build a fully loaded inference cost from a rate card: caching, output ratio, uplifts and fees.
- Judge when self-hosting is likely to pay, and which inputs decide it.
Terms to know
AI TCO · Total cost of AI (TCA) · Inference Cost · Reasoning Tokens · Model Hosting · Model Routing
Worked example
When does a dedicated endpoint beat serverless?: Self-hosting pays when volume is high, steady and measured. Price the idle hours and the people before the hardware.
Readings
AI TCO Framework · The Components of AI TCO · What Cloud Taught Us About the Real Cost of AI Inference · How Inference Choices Change AI Cost
Module 3
Allocating and reporting AI cost
You can
- Build a monthly cost pool for shared AI capacity and allocate it with a fair driver.
- Split idle capacity by cause and assign each cause an owner.
- Sequence the buying levers (caching, batch, commitment) and test a commitment by break-even utilisation.
- Map AI charges to FOCUS 1.4 columns and explain where FinOps hands over to value management.
Terms to know
Cost Pools · Cost Allocation · Idle Capacity · Showback · Chargeback · Shared AI Platform · FOCUS · FinOps Scope
Worked example
Commit for the floor, not the peak: Test every commitment against its break-even utilisation, and apply caching and batch before you commit.
Readings
Allocating Shared AI Capacity · Buying AI Capacity and Reporting It in FOCUS · FinOps and AI · FinOps for AI Is Necessary, Not Sufficient
Module 4
Measuring AI value
You can
- Build a value claim with a baseline, a quality floor, netting and a stated conversion.
- Grade evidence from E0 to E4 and apply the rule that no monetised claim sits below E2.
- Specify the data and the workload identifier that join cost to outcome.
- Compute attribution coverage and state which test and denominator it uses.
Terms to know
AI Baseline · Quality floor · Observe, Net, Convert, Book · Counterfactual evidence level · Value claim · AI value ledger · Cost per Outcome
Worked example
A customer-service agent, netted to the quality floor: Counting every contact the agent touched, instead of those that stayed resolved, would have overstated the volume by more than 8,000 a month.
Readings
How to Measure AI Value · AI Value Attribution · What You Actually Need to Measure AI Value · The AI Value Ledger
Module 5
Building the business case
You can
- Test a business case for a measured baseline, a conversion mechanism and a full cost denominator.
- Apply NPV, IRR and payback to an AI case and explain why the horizon should be short.
- Classify return dimensions, capture status and portfolio categories without blending them into one number.
- Stress-test a case against price and demand and state where it fails.
Terms to know
AI ROI · Cost Avoidance vs. Cost Reduction (AI) · Realised Productivity Gain · Value Capture Failure · Net AI value
Worked example
The same case, as finance sees it: Discount, keep the horizon short, state the accounting treatment and stress-test before the case goes in.
Readings
What CFOs Should Ask of AI ROI Claims · The AI Business Case: Finance Treatment · AI ROI Models · What to Bring to an AI Investment Review
Module 6
Governing the portfolio
You can
- Apply scale, hold, redesign or stop using return on scarce capacity and evidence level.
- Match the evidence required to the size and reversibility of a commitment.
- Write exit criteria and tell rational persistence from escalation of commitment.
- Assign accountability for AI economics across finance, business owners, platform, FinOps and line managers.
Terms to know
Return on scarce capacity · Exit Criteria · Stage Gate (AI Investment) · Governance Theater · Value Realisation
Worked example
A five-initiative portfolio review: Realised value set beside claimed value is often the most important number on the page.
Readings
Scale, Hold, Redesign, Stop · When to Stop an AI Initiative · The AI Portfolio Review Pack · Who Owns AI Economics?
How the capstone memo is marked
Five criteria, each scored from 1 to 5. You need 3 or more on every one, and your figures must be within tolerance.
- Cost boundary
- Full cost on the same workload boundary as the benefit, with exclusions stated.
- Value claim
- Baseline, quality floor, netting and a named conversion, with capture status.
- Evidence
- An evidence level that fits the commitment requested, and how it will be raised.
- Decision
- Scale, hold, redesign or stop, consistent with the figures, with an owner and a date.
- Risk and stop rule
- The assumption most likely to break the case, and the condition that would stop it.
Before you sit it
- Do every module exercise with the tool open. The capstone figures use the same calculations.
- Take the five practice questions at the end of each module. The exam uses different questions in the same style.
- Read the questions for the decision being asked, not just the figure. Most test whether you would act on the number.
- You may use the site and your notes during the exam, but 45 minutes for 30 questions leaves little time to look things up.