What changed in April 2026. Microsoft rewrote Copilot Studio licensing to introduce a four-mechanism buying model: pay-as-you-go consumption, Copilot Credit Capacity Packs (monthly subscriptions), Copilot Credit Pre-Purchase Plan with CCCUs (annual commit with 5–20% discount), and Agent Pre-Purchase Plan with ACUs (cross-platform agent commit). The change unified Microsoft’s agent commercial model across Copilot Studio and Microsoft Foundry.
What this means. Procurement teams now have four buying levers rather than one. The optimal combination varies by deployment maturity, workload predictability, and roadmap scope. Choosing the wrong mechanism overpays by 15–30% versus the optimal one for the same usage.
The recommended sequence. Start pay-as-you-go to learn actual costs. Move predictable baseline to Copilot Credit Pre-Purchase Plan after 3–6 months. Keep pay-as-you-go as spillover. Move to Agent Pre-Purchase Plan only if the roadmap spans the wider Microsoft agent stack.
Why Microsoft rewrote Copilot Studio licensing
The Copilot Studio pricing model that existed through early 2026 was structurally inadequate for enterprise adoption. The original model used per-message pricing with limited commitment options. As enterprise agent deployment matured, the per-message structure produced unpredictable bills, made budgeting difficult, and prevented procurement teams from negotiating meaningful commitments. Microsoft account teams routinely heard from CFOs that Copilot Studio costs were impossible to forecast with confidence.
The April 2026 rewrite addressed three problems simultaneously. First, predictability — the new pre-purchase plans give procurement teams fixed annual costs in exchange for usage commitments. Second, optionality — the four-mechanism structure lets organisations match buying approach to workload characteristics rather than forcing a single model. Third, unification — Microsoft’s broader agent strategy spans Copilot Studio, Microsoft Foundry, and eventually Fabric and GitHub agent services, and a unified commercial model across these platforms was needed.
The rewrite also reflected Microsoft’s commercial maturation around agent licensing. The Agent 365 governance product launched on 1 May 2026 (see the Agent 365 guide); the Copilot Studio rewrite happened the month before. Together they represent Microsoft’s first comprehensive enterprise agent commercial framework.
Mechanism 1: Pay-as-you-go
The simplest entry point. Organisations consume Copilot Credits against actual usage with no commitment, billed monthly at standard rates. Suitable for early-stage agent deployment where consumption patterns are unknown and forecast accuracy is poor.
The mechanic: each Copilot Studio operation consumes a metered amount of Copilot Credits. The credit consumption varies by operation type — AI model used (different models have different credit rates), tokens processed (longer prompts and responses cost more), tool calls made (each integration call consumes credits), and several other factors. Microsoft publishes a credit consumption reference that procurement teams should review during workload sizing.
When pay-as-you-go is the right answer:
- Pilot deployment with no production usage forecast yet
- Sporadic agent workloads where consumption is unpredictable
- Spillover capacity for organisations on pre-purchase plans whose baseline gets exceeded
- Early-stage organisations gathering data before committing
When pay-as-you-go is the wrong answer:
- Predictable baseline usage above modest levels — pay-as-you-go pricing is the worst per-credit rate available
- Organisations requiring budget certainty for the year
- Deployment volumes large enough to qualify for meaningful pre-purchase discount
Mechanism 2: Copilot Credit Capacity Packs
Monthly subscription with a fixed credit allotment, billed predictably. Bridges between pay-as-you-go variability and longer-term pre-purchase commitment. The default Dataverse allowance was tripled in the April 2026 update, making Capacity Packs more attractive for agent workloads using Dataverse-backed data.
The mechanic: subscribe to a Capacity Pack at one of several tier sizes, receive monthly credit allotment, consume against the allotment. Unused credits within a month do not roll over — the monthly use-it-or-lose-it structure is the most common procurement complaint about this mechanism. Overage above the allotment falls back to pay-as-you-go pricing.
When Capacity Packs are the right answer:
- Steady monthly workloads with low month-over-month variability
- Organisations not yet ready for annual commitment
- Specific workloads with predictable consumption
When Capacity Packs are the wrong answer:
- Highly variable workloads — the monthly use-it-or-lose-it penalises variability heavily
- Organisations with sufficient usage to justify annual pre-purchase — CCCU pricing is materially better
- Organisations using multiple Microsoft agent platforms — the broader ACU is more flexible
Mechanism 3: Copilot Credit Pre-Purchase Plan (CCCU)
The annual commitment mechanism. Organisations buy a one-year pool of Copilot Credit Commit Units (CCCUs) at discounted pricing, draw down against the pool through actual usage, and fall back to pay-as-you-go pricing if the pool exhausts before the year ends. The pool runs for a full year rather than expiring monthly — addressing the main Capacity Pack complaint.
Discount tiers scale with commitment size. Small commitments earn approximately 5% discount; the largest commitments earn approximately 20%. The exact tier breakpoints are negotiated commercially; Microsoft account teams have flexibility on tier placement for enterprise customers.
| Commitment size | Typical discount | Effective rate vs pay-as-you-go |
|---|---|---|
| Small (entry-level) | ~5% | 95% of pay-as-you-go rate |
| Medium (mid-market) | ~10–12% | 88–90% of pay-as-you-go rate |
| Large (enterprise) | ~15–18% | 82–85% of pay-as-you-go rate |
| Strategic (largest commits) | ~20% | 80% of pay-as-you-go rate |
When CCCU pre-purchase is the right answer:
- Organisations with 3–6 months of pay-as-you-go data establishing predictable baseline
- Annual budget cycle organisations preferring fixed annual cost
- Deployment volume large enough to qualify for meaningful discount tier
- Workloads concentrated in Copilot Studio specifically (rather than spanning the broader agent stack)
Mechanism 4: Agent Pre-Purchase Plan (ACU)
Added in February 2026, this is the cross-platform commitment mechanism. ACUs work like CCCUs but draw down against eligible consumption across Copilot Studio, Microsoft Foundry, and eventually Fabric and GitHub agentic services. The roadmap is that ACUs will be the unified commitment unit for the Microsoft agent ecosystem; CCCUs are Copilot-Studio-specific.
The discount structure parallels CCCUs but the underlying value proposition is different. CCCUs lock organisations into Copilot Studio specifically. ACUs preserve flexibility to move consumption between Microsoft agent platforms as roadmaps evolve. For organisations whose 2026–2028 plan spans Copilot Studio AND Foundry AND Fabric agents AND GitHub agents, ACUs are the correct shape.
When ACU pre-purchase is the right answer:
- Multi-platform agent roadmaps spanning Copilot Studio, Foundry, and beyond
- Organisations actively using or planning Microsoft Foundry for agent development
- Strategic Microsoft customers consolidating multiple agent platforms
When ACU pre-purchase is the wrong answer:
- Copilot-Studio-only deployments — CCCUs provide same discount with simpler commercial structure
- Organisations not yet using Microsoft Foundry or other agent platforms
- Smaller commitments where the cross-platform flexibility is not actionable
The optimal combination strategy
Most enterprise organisations end up using two or three of the four mechanisms rather than one. The optimal combination depends on deployment maturity and workload characteristics.
Stage 1: Discovery (0–3 months)
Pure pay-as-you-go. No commitment. Goal: learn actual consumption patterns. Output: baseline data for sizing the next stage.
Stage 2: Baseline commitment (3–9 months)
CCCU pre-purchase sized to cover 70–80% of expected annual consumption. Pay-as-you-go retained as spillover for the remaining 20–30%. The 70–80% sizing prevents the pool from exhausting early (which would force expensive pay-as-you-go rates for the rest of the year) while capturing the discount on the predictable baseline.
Stage 3: Mature deployment (9+ months)
Two paths depending on platform scope. Copilot-Studio-only organisations continue with CCCU + pay-as-you-go spillover. Multi-platform organisations migrate the CCCU commitment to ACU and add Foundry, Fabric, or GitHub agent consumption to the same commit pool.
Stage 4: Strategic commitment (12+ months)
Large enterprise deployments capable of 20% discount tier consolidate to ACU at strategic scale, with multi-year EA-level commitments that lock the discount tier across renewal cycles.
Negotiation considerations
Three negotiation levers apply specifically to Copilot Studio licensing in 2026.
Tier placement flexibility. Microsoft account teams have discretion on the discount tier applied to specific CCCU and ACU commits. Aggressive procurement teams move borderline commits up one tier through negotiation — the difference between 10% and 12% discount across a large commit is material.
Multi-year commitments. CCCU and ACU pre-purchases default to one-year terms. Multi-year commitments (2-3 years) are negotiable for strategic customers and capture additional discount on top of the volume tier.
EA bundling. Copilot Studio consumption commitments can be bundled into Enterprise Agreement renewals, providing a single procurement vehicle and additional negotiation leverage. The bundling also helps the EA account team meet their attach quotas, which can deliver flow-back discount on other EA line items.
Common mistakes procurement teams make
Mistake 1: Sizing the commit too large. The biggest commercial error in Copilot Studio pre-purchase is overcommitting. Unused CCCUs at year end are lost — they do not roll forward. Sizing the commit to 70–80% of expected usage rather than 100% protects against this. The retained spillover capacity through pay-as-you-go is more expensive per credit but cheaper than overcommitted capacity that goes unused.
Mistake 2: Sizing the commit too small. Undersized commits exhaust early and force pay-as-you-go rates for the rest of the year. The economics flip: discount achieved on the committed portion is overwhelmed by premium paid on the uncommitted portion.
Mistake 3: Choosing ACU when CCCU is sufficient. ACU exists for cross-platform flexibility. Organisations using only Copilot Studio receive no benefit from ACU but accept the additional commercial complexity. Default to CCCU unless the cross-platform need is real.
Mistake 4: Treating Capacity Packs as a default. Microsoft account teams sometimes push Capacity Packs as the entry-level commitment for organisations not ready for CCCU. The monthly use-it-or-lose-it structure makes Capacity Packs economically inferior to pay-as-you-go in most scenarios. Skip Capacity Packs for most enterprise customers.
Action plan
- Identify your current stage. Discovery, baseline, mature, or strategic. The right mechanism mix flows from the stage.
- Pull consumption data if available. Organisations already on pay-as-you-go should have 3+ months of data to size pre-purchase commits.
- Model the optimal mix. Pay-as-you-go + CCCU at 70–80% sizing is the default for most enterprises in their first year of pre-purchase.
- Negotiate tier placement. Push for the next discount tier on borderline commits. The 2–3 percentage point delta is material.
- Engage independent advisory. The four-mechanism model is genuinely new and the optimal answer is not always obvious. Book a scoping call.