Explore how Microsoft’s top-tier productivity, security, and AI offerings fit together, what licensing may require, and whether E7 makes sense for your organization.

On 1 May 2026 Microsoft shipped its first new enterprise tier in a decade. Microsoft 365 E7, branded the Frontier Suite, bundles E5, Copilot, the Entra Suite and Agent 365 for $99 per user per month — against $117 bought separately. The arithmetic is honest. The assumption underneath it is the part worth interrogating before you sign.

E7 is not E5 at a higher price. It is a statement about what Microsoft believes a mature enterprise now looks like: fully licensed for security, committed to Copilot, and beginning to operate AI agents under corporate governance. If that describes your estate, the bundle saves 15%. If it does not, E7 is an elegant way to pay for three products you have not yet justified. The sensible reading of this launch is therefore less about the bundle than about your own readiness — and the questions below are the ones a procurement team should settle first.

A decade in the making, a quarter in the negotiating

Microsoft announced E7 on 9 March 2026 and made it generally available on 1 May — the first new tier since E5 arrived in December 2015. It sells through the usual channels: Enterprise Agreement, EA Subscription, MCA-Enterprise and CSP, in variants with and without Teams for European compliance. Two pricing facts frame any negotiation. First, E7 escaped the 1 July 2026 increases that pushed E3 to $39 and E5 to $60 — the one large SKU in the portfolio whose list price is stable this year. Second, the CSP promotion that offered 10% off from ten seats and 15% from one hundred closed on 30 September 2026; write-ups still promising that discount are describing a door that has shut.

Context matters here. Microsoft removed EA volume discounts in November 2025, weakening the old lever of buying big to save big. The E7 promotion was the replacement lever — and its withdrawal is telling. Microsoft no longer needs to discount the bundle to sell the concept; it needs you to arrive already persuaded. That shifts negotiating weight back to architecture: how you mix tiers, how you re-harvest licences when roles change, and how well you can evidence usage. Partners report that estates with clean ITAM data negotiate meaningfully better terms than estates arguing from slide decks.

What the bundle actually contains

Four products, one invoice. Microsoft 365 E5 supplies the office applications, Exchange, SharePoint, Teams and the security stack — Defender XDR, Purview, Intune, Entra ID P2 and Power BI Pro. Microsoft 365 Copilot adds the assistant across Word, Excel, PowerPoint, Outlook and Teams, with Copilot Chat and the Work IQ context layer. The Entra Suite contributes identity governance, Private Access and Internet Access. And Agent 365 — itself GA on 1 May — supplies the control plane for AI agents: identities in Entra, Purview and Defender policies inherited, shadow agents discovered on Windows devices. That last component deserves emphasis, because it is the least understood: Agent 365 governs agents; it does not build them and it does not pay for their compute.

Fig. 1. The E7 bundle: E5, Copilot, Entra Suite and Agent 365 — $117 bought separately, $99 as a bundle

The four sums every CFO will ask for

The bundle’s value depends entirely on the starting point, and there are exactly four cases. An E5-only estate at $60 pays $39 more per user for E7 — indefensible without a funded Copilot plan. An E5-plus-Copilot estate at $90 pays $9 more and receives Entra Suite and Agent 365, effectively two products for the price of a sandwich; most committed Copilot shops pass through this point of near-parity. An estate already at E5, Copilot and Entra Suite pays $102 separately and $99 in the bundle — E7 is cheaper and adds Agent 365. And the full four-product stack at $117 saves $18 per user, or 15%; at a thousand seats that is roughly $216,000 a year at list prices. Note the direction of travel: E7 rewards those who have already bought nearly everything. It is a consolidation discount, not an adoption incentive.

Fig. 2. Four scenarios for E7 economics: from $39 more expensive for E5-only estates to an 18-dollar saving with the full stack 

The lines E7 does not cover

Three budget lines sit outside the $99. Agent compute is the first: Copilot Studio, Microsoft Foundry and Work IQ API calls are metered in Copilot Credits through Azure, so every agent you deploy spends money the bundle does not contain. The second is the frontier agent tier: Copilot Cowork remains a Frontier programme with usage billing, and Scout is experimental; neither is inside E7. The third is Security Copilot, which Microsoft is rolling into E5 and E7 tenants in a metered model — 400 SCUs a month per thousand users — and sustained use will outrun that allowance quickly. A defensible TCO model for E7 therefore reads like a cloud bill: a fixed platform line at $99, plus a measured consumption line for agents and security AI, plus a growth reserve. Estimating that reserve from a pilot cohort’s telemetry is far safer than extrapolating from Microsoft’s launch materials.

What Microsoft gains — and why that matters to you

It is worth stating the vendor’s interest plainly, because it explains the packaging. Microsoft has spent three years proving that Copilot sells; it has spent the last year proving that customers renew it. Bundling Copilot into a tier that also contains the security stack converts a discretionary AI experiment into infrastructure spend — the budget line that survives cost reviews. None of this makes E7 a poor deal; it makes it a deal to be examined on your numbers rather than Microsoft’s narrative. The healthy posture in a renewal conversation is symmetrical: Microsoft consolidates its revenue, you consolidate your discounts, and the audit trail of actual usage is the currency both sides trade in.

There is also a timing consideration. Organisations that restructured their Microsoft agreements after the November 2025 EA changes generally report less flexibility than those still mid-term, because the new commercial constructs reward commitment to suites rather than to volumes. If your EA matures within the next eighteen months, the E7 decision belongs in that negotiation as a structural question — bundle architecture, role mapping, renewal caps — rather than as a mid-term purchase, where the leverage is thinner and the margin for renegotiation smaller.

Mixed estates: the unglamorous answer that works

Consider an illustrative firm of 800 staff on E5, of whom 250 use Copilot intensively — executives, analysts, the sales floor. Licensing everyone into E7 costs $950,400 a year. Licensing the 250 into E7 and leaving 550 on E5 costs $693,000. The difference, some $257,400 annually, buys exactly nothing in capability terms except unused subscriptions. The research supports caution: Flexera’s State of ITAM 2026 reports that 59% of organisations see wasted AI software spend rising, while only 31% have full visibility into who uses what, and a mere 29% measure value. The same study’s discipline — usage telemetry, role-based triggers, six-monthly reviews — routinely saves more than any promotion ever did.

The practical rule is to license by trigger, not by seniority. E5 where a named security or compliance need exists; E7 where E5 already sits and Copilot plus agents appear in funded plans; E3 for the long tail whose work is documents, email and Teams. The failure mode to avoid is not under-licensing — it is the ceremonial upgrade of whole departments because the bundle is the flagship. A two-week telemetry baseline before any decision will almost always embarrass the org chart and vindicate the accountants.

Fig. 3. E3, E5 and E7: three tiers, three risk profiles — choosing by trigger rather than job title

Three questions, briefly

Question. Is E7 a replacement for E5?

Answer. No. E5 continues at $60 alongside it. E7 is a separate, higher tier; Microsoft positions them as adjacent levels of maturity rather than old and new.

Question. Can we license E7 for part of the workforce?

Answer. Yes — and it is the recommended pattern. Per-user licensing lets you assign E7 to Copilot-heavy roles while others remain on E5 or E3, which is the standard defence against paying for idle AI seats.

Question. Does E7 include agent compute?

Answer. No. Agent 365 governs agents; their consumption in Copilot Studio, Foundry and Work IQ APIs bills separately in Copilot Credits via Azure. The bundle buys control of agents, not their labour.

A strong fit for organizations that need enterprise productivity, security, and compliance without committing to every capability in the flagship bundle.