Jul 13, 2026

Microsoft Copilot enterprise pricing in 2026: full cost guide

Microsoft Copilot enterprise pricing now combines Microsoft 365 licenses, a $30 Copilot seat, Copilot Cowork usage, GitHub Copilot AI Credits, and optional agent capacity. This guide explains the real 2026 cost.

GUIDE10 min readThe Currai team / Engineering

TL;DR: Microsoft Copilot enterprise pricing is no longer one price. Microsoft 365 Copilot costs $30 per user per month with an annual commitment, on top of a qualifying Microsoft 365 license. Copilot Cowork adds usage-based Copilot Credits for long-running agent tasks. GitHub Copilot Business costs $19 per user per month and GitHub Copilot Enterprise costs $39, with additional AI Credits available after the included allowance.

For a 100-person company on Microsoft 365 E3, Microsoft 365 Copilot brings the published license total to $6,900 per month before Cowork usage. Add GitHub Copilot Enterprise for the same 100 people and the fixed license floor becomes $10,800 per month. Copilot Studio, extra AI Credits, taxes, and negotiated contract terms can move the total further.

The right way to budget is to separate fixed seats from metered agent work, apply hard limits before enabling usage, and measure cost per successful task rather than cost per prompt.

Microsoft Copilot enterprise pricing at a glance

“Microsoft Copilot” can refer to several products with different licenses and billing systems. Enterprise buyers should separate them before comparing a quote or modeling a rollout.

ProductPublished priceWhat the base price coversVariable cost
Microsoft 365 Copilot$30/user/month, paid yearlyCopilot in Microsoft 365 apps, Copilot Chat, Work IQ, and included agentsCowork and other eligible usage-based services use Copilot Credits
Microsoft 365 E3$39/user/monthQualifying Microsoft 365 suiteAdd-ons and metered services
Microsoft 365 E5$60/user/monthHigher-tier Microsoft 365 suiteAdd-ons and metered services
GitHub Copilot Business$19/user/monthOrganization coding assistance and 1,900 AI Credits per userAdditional credits at $0.01 each when enabled
GitHub Copilot Enterprise$39/user/monthEnterprise coding assistance and 3,900 AI Credits per userAdditional credits at $0.01 each when enabled
Copilot CoworkRequires Microsoft 365 CopilotAccess to multi-step Microsoft 365 agent tasksCopilot Credits based on model, context, tools, and runtime
Microsoft Copilot StudioSeparate capacity or pay-as-you-go licensingBuilding and managing custom agentsCopilot Credit consumption varies by agent behavior

These are US list prices as of July 13, 2026. They exclude taxes, regional pricing, promotions, and enterprise contract discounts.

Microsoft 365 Copilot costs $30 plus a qualifying license

Microsoft lists Microsoft 365 Copilot at $30 per user per month with an annual subscription. The add-on includes Copilot in Word, Excel, PowerPoint, Outlook, Teams, and the Microsoft 365 Copilot app, along with Copilot Chat, Work IQ, and Microsoft's included agent experiences.

The $30 price is not the full seat cost. Each user also needs a qualifying Microsoft 365 license.

Microsoft's July 2026 pricing update sets the US list price for Microsoft 365 E3 with Teams at $39 per user per month and Microsoft 365 E5 with Teams at $60. Existing customers remain on their prior price until renewal, so the effective date depends on the contract.

The fixed monthly license floor is therefore:

ConfigurationBase licenseCopilot add-onPublished total per user/month
Microsoft 365 E3 + Copilot$39$30$69
Microsoft 365 E5 + Copilot$60$30$90

That total buys access. It does not guarantee that every employee will use the product enough to justify a seat, and it does not include all usage-based agent work.

Copilot Cowork adds usage-based pricing

Copilot Cowork is Microsoft's agentic Microsoft 365 experience for longer, multi-step tasks. It can research information, work across organizational context, create and edit documents, use tools, and take approved actions in Microsoft 365.

Cowork requires a Microsoft 365 Copilot User Subscription License. Its task usage is then billed in Copilot Credits on top of that seat.

Microsoft calculates task consumption from four inputs:

  • The AI models used during the task.
  • The organizational or external context retrieved.
  • Tool and skill calls performed.
  • The total runtime of the task.

This makes the bill workload-dependent. A short drafting task and a long-running research workflow do not consume the same number of credits, even when the same user starts both.

Microsoft offers two payment paths. Pay-as-you-go, or PayGo, charges $0.01 per Copilot Credit. The prepaid P3 model lets an enterprise commit to a volume of credits for a discount. The economically useful comparison is not simply PayGo versus P3; it is actual task volume versus the committed capacity that would go unused.

Do not budget Cowork using an assumed universal price per task. Microsoft exposes a cost estimator and usage reports because the task price changes with the model, context, tools, and runtime. Start with a production-shaped pilot, then build the forecast from observed task distributions.

GitHub Copilot Enterprise pricing is a separate bill

GitHub Copilot is owned by Microsoft, but it is not included in Microsoft 365 Copilot. Engineering organizations license and manage it separately.

According to GitHub's organization and enterprise billing documentation:

  • GitHub Copilot Business costs $19 per user per month and includes 1,900 AI Credits per user.
  • GitHub Copilot Enterprise costs $39 per user per month and includes 3,900 AI Credits per user. It requires GitHub Enterprise Cloud.

One GitHub AI Credit equals $0.01. Chat, agent mode, code review, cloud agents, Copilot CLI, and other eligible agentic features consume credits. Paid-plan code completions and next-edit suggestions do not consume them.

Existing Business and Enterprise customers receive higher promotional credit allowances during the June through August 2026 transition. A long-term forecast should use the standard allowance rather than treating the promotional pool as permanent capacity.

If the included pool runs out, administrators decide whether additional paid usage is allowed. GitHub provides user, cost-center, organization, and enterprise budget controls. To turn a budget into an actual cap, administrators must enable the option that stops usage when the limit is reached.

The coding-agent side deserves its own evaluation. Read the best GitHub Copilot alternatives in 2026 for comparisons with Cursor, Claude Code, Codex, Cline, Continue, Aider, and OpenCode.

Microsoft Copilot pricing for 100 users

A worked example makes the fixed costs easier to see. Assume 100 users, US list prices, and the same employees receiving each selected license.

DeploymentMonthly fixed costAnnual fixed cost
Microsoft 365 E3 + Microsoft 365 Copilot$6,900$82,800
Microsoft 365 E5 + Microsoft 365 Copilot$9,000$108,000
GitHub Copilot Business$1,900$22,800
GitHub Copilot Enterprise$3,900$46,800
E3 + Microsoft 365 Copilot + GitHub Copilot Enterprise$10,800$129,600

The combined example is a license floor, not a final invoice. Add:

  • Copilot Cowork credit consumption.
  • GitHub AI Credits beyond the included pool.
  • Copilot Studio capacity or pay-as-you-go use.
  • GitHub Actions minutes used by eligible code-review workflows.
  • Security, compliance, implementation, and support products not included in the selected plans.
  • Taxes, currency differences, and contract terms.

The example also assumes a 100% seat assignment. In practice, the largest early cost lever is often licensing fewer users and expanding after the pilot proves which roles use Copilot enough to justify the fixed seat.

Microsoft Copilot Studio pricing is another layer

Microsoft Copilot Studio is the low-code and pro-code environment for building custom agents. It has its own licensing and billing model based on Copilot Credit capacity, pay-as-you-go consumption, or prepaid packs.

Some agent actions available to licensed Microsoft 365 Copilot users are zero-rated in Microsoft 365. Other agent behavior, connectors, autonomous actions, or deployment surfaces can consume paid capacity. That distinction is why a Microsoft 365 Copilot seat should not be treated as unlimited custom-agent inference.

Before deployment, document:

  • Which agents are included with the Microsoft 365 Copilot license.
  • Which actions consume Copilot Studio capacity.
  • Whether prepaid packs or pay-as-you-go billing apply.
  • Where excess usage is charged.
  • Which administrators own the budgets and access policies.

The enterprise cost-control playbook

Enterprise Copilot cost control now has two separate jobs: manage fixed seat utilization and govern variable agent consumption.

1. Separate licenses from usage

Report the fixed monthly seat floor separately from Cowork, GitHub AI Credits, Copilot Studio, and other metered services. A blended monthly number hides whether cost increased because more employees received licenses or because existing users ran more agent tasks.

2. Configure hard limits before enabling agents

Microsoft's current Copilot Credit cost-management controls support access policies, allocations, budgets, alerts, and hard caps for Cowork and the Work IQ API. GitHub also supports stop-usage behavior on configured budgets.

An alert and a hard limit are different controls. Confirm the behavior at 100% of budget in a test group before enabling a broad rollout.

3. Allocate by role and workload

Do not assign every product to every employee. Knowledge workers may need Microsoft 365 Copilot, developers may need GitHub Copilot, and a smaller group may need both. Cowork access and credit limits can be assigned to the roles with repeatable multi-step workflows.

4. Right-size the model

Cowork model selection affects task cost, speed, and output. Use efficient models for routine drafting and extraction, then reserve deeper reasoning or review modes for tasks where the extra quality changes the business outcome.

5. Measure cost per useful outcome

License utilization is not value. Track completed research tasks, accepted documents, resolved support work, merged pull requests, developer review time, and human corrections. Divide total fixed and variable cost by the outcomes the deployment was meant to improve.

For an agent workload:

cost per successful task =
  (allocated seat cost + usage cost + review cost) / accepted tasks

This catches two common errors: celebrating cheap failed tasks and ignoring the human cost of correcting plausible-looking output.

Where Currai fits in the cost model

Currai does not directly trace closed Microsoft 365 Copilot, Cowork, or GitHub Copilot sessions. Use Microsoft's and GitHub's administration, billing, and usage exports for those products.

Currai applies the same cost-per-outcome discipline to the custom AI apps and agents your organization builds around them. A trace can connect model calls, retrieval, tools, latency, token use, errors, and final outcomes for a custom support agent, RAG workflow, or internal assistant.

That gives an AI platform team a consistent evaluation model across buy and build decisions:

  • Microsoft and GitHub reports show licensed-product adoption and credit spend.
  • Currai traces show how custom applications use models and tools.
  • Shared outcome definitions make the two categories comparable at the business level.

Read track token cost across models and traces, test agent cost-efficiency, and set budgets and alerts for runaway LLM cost for the custom-application side of the framework.

A practical Microsoft Copilot rollout plan

Start with a controlled deployment instead of purchasing every seat and agent capacity at once:

  1. Identify three to five repeatable workflows with measurable outcomes.
  2. Assign Microsoft 365 Copilot or GitHub Copilot only to the relevant pilot group.
  3. Configure Cowork, AI Credit, and Copilot Studio limits before enabling paid usage.
  4. Record the fixed license cost, metered consumption, completion rate, and human review effort.
  5. Compare the pilot against the previous workflow for at least one full billing cycle.
  6. Expand the workloads that show a defensible cost per outcome.
  7. Remove inactive seats and revise credit allocations before renewal.

This turns procurement from a one-time seat decision into an operating model. The goal is not the highest Copilot adoption percentage. It is useful work at a cost the organization can explain and control.

Frequently asked questions

How much does Microsoft 365 Copilot cost for enterprise customers?

Microsoft 365 Copilot is listed at $30 per user per month with an annual commitment. It requires a qualifying Microsoft 365 license. At July 2026 US list prices, Microsoft 365 E3 plus Copilot totals $69 per user per month, while E5 plus Copilot totals $90.

Is Copilot Cowork included in the $30 Microsoft 365 Copilot price?

The Microsoft 365 Copilot license enables access to Cowork, but Cowork tasks are billed separately through usage-based Copilot Credits. PayGo credits cost $0.01 each, and prepaid commitment options are available.

How much does GitHub Copilot Enterprise cost?

GitHub Copilot Enterprise costs $39 per user per month and includes 3,900 AI Credits per user under the standard allowance. GitHub Enterprise Cloud is required. Additional credits cost $0.01 each when paid usage is enabled.

Can Microsoft Copilot enterprise spending be capped?

Yes. Microsoft's current Copilot Credit management supports policies, budgets, alerts, limits, and hard caps for Cowork and the Work IQ API. GitHub budgets can also stop additional Copilot usage when configured to do so. Administrators should test the stop behavior before a broad deployment.

Is Microsoft Copilot Studio included with Microsoft 365 Copilot?

Microsoft 365 Copilot includes some agent-building and agent-use capabilities, but Copilot Studio has separate capacity and pay-as-you-go licensing for many custom-agent scenarios. Review the licensing guide for the exact agent actions and deployment surfaces in scope.

Related: GitHub Copilot alternatives, LLM observability, and LLM evals.

03

Keep going with nearby topics from the Currai blog.