Preview|MSFT 2Q26: Copilot Sales Likely to Beat Expectations; User Experience Continues to Improve, but Competition Remains Intense; Azure Capacity and Pricing Trends are Improving

Ahead of the quarterly results, we conducted focused interviews with five experts and updated our views on several key topics:
Quarterly view

1. M365 Copilot and E7 sales progress, and the FY27 sales outlook
2. Copilot user experience, product improvements, and the competitive landscape
3. Copilot Studio progress
4. Azure AI Foundry and enterprise multi-model adoption
5. GPU leasing, data center capacity, supply chain, and Azure traditional cloud trends
6. Updates to our Azure AI and MSFT models.

Copilot and E7 Sales Progress and Product Updates

We spoke with four experts with direct experience in M365 Copilot. Key takeaways below.

1. Copilot FY26Q4 sales progress

After each discussion with sales partner experts A and B, we update our Copilot Seat Model using the QoQ growth data they provide. Our quarterly estimates are as follows:

• End Dec-2024: 6.3mn seats, including 5.5mn Enterprise seats and 0.8mn SMB seats.
• End Mar-2025: 9.6mn seats, including 8.5mn Enterprise seats and 1.1mn SMB seats.
• End Jun-2025: 12.4mn seats, including 11.0mn Enterprise seats and 1.3mn SMB seats.
• End Sep-2025: 15.1mn seats, including 13.4mn Enterprise seats and 1.7mn SMB seats.
• End Dec-2025: 18.1mn seats, including 15.1mn Enterprise seats and 3.0mn SMB seats.
• End Mar-2026: 22.0mn seats, including 19.0mn Enterprise seats and 3.0mn SMB seats.
• End Jun-2026: 37.0mn seats, including 34.0mn Enterprise seats and 3.0mn SMB seats.

2. Copilot sales materially exceeded the experts’ prior sales targets this quarter. Most of the increase came from larger purchases by enterprises that had already completed limited deployments.

a. Some customers had previously deployed only a few thousand seats within a single department or region. This quarter, they began expanding to companywide deployments involving hundreds of thousands of seats.

b. Based on expert feedback, we estimate that approximately 3.0mn incremental seats came from the conversion of free users to paid subscriptions.

c. The number of new customers also increased, but expansion among existing large customers remained the largest source of incremental seats.

d. The two new sales partner experts we added this quarter, experts C and D, observed the following changes in Copilot adoption:

• Expert C saw the share of E5 customers purchasing Copilot increase from 50% to 55% at the end of March to approximately 65% at the end of June.
• Expert D saw the same measure increase from 20% to 25% at end March to 30% to 40% at end June.

3. E7 sales progress

a. E7 was formally launched this quarter and remains at an early stage of sales and customer migration. We estimate approximately 1.5mn E7 users, which are included in the 37.0mn Copilot seats discussed above.

b. The experts cited the following attractions of E7:

• Large customers already using E5, Copilot and other Microsoft products can reduce the complexity of license and vendor management.
• Large customers generally have more room to negotiate discounts on the overall E7 package. The effective cost per seat may be below $90.
• E7 enables Agent 365 and billing based on AI Credits usage, providing a common entry point for customers to adopt CoWork and other Agent products.

c. The experts also identified several near term barriers to E7 sales:

• Customers that have already purchased E5 and Copilot separately need to address existing contracts, license reallocation and potential duplicate billing.
• Some customers are still assessing whether E7 offers a meaningful cost advantage over their current product mix, which is slowing contract conversions.
• E7 spans identity, security, data governance and Agent management. Some customers still need to complete system migration and internal approvals. Expert D estimated a migration period of two to six months.
• E7 includes a relatively complete identity, security and access management stack, but some customers already use third party products such as Okta under contracts that remain in force. The near term product overlap makes it difficult to realize immediate savings through vendor consolidation, which reduces the incentive to migrate to E7.
• Microsoft is using license discounts and migration funding to offset part of the migration cost for selected large customers.

4. Sales outlook

a. Based on expert feedback, we expect FY27Q1 Copilot sales to be materially weaker than FY26Q4, mainly due to Microsoft’s normal sales seasonality.

• Most of Microsoft’s large contracts are signed in FYQ4, while FYQ1 is usually the weakest sales quarter of the year.
• At the start of a new fiscal year, sales teams need to reset annual plans and sales quotas, leaving less time for customer expansion activity.
• Customers also use this period to reassess Copilot, Claude and other AI products, which may delay some expansions and contract signings.

b. For E7, we estimate 1.0mn to 1.5mn incremental seats next quarter, which could bring the user base to approximately 3.0mn by the end of September. Based on feedback from several channels, we believe Microsoft may aim to convert approximately 10% of E3 and E5 customers to E7 in FY27, implying a target of roughly 20mn seats.

c. Based on feedback across our channel checks, we believe Microsoft’s FY27 Copilot sales plan is aggressive and may target more than a doubling of total seats from the current level by fiscal year end..

5. Follow up on Copilot organizational changes

a. Last quarter, we learned that Microsoft had further separated Copilot sales responsibilities from technical development. The sales team became more focused on customer adoption and commercialization, while the technical team focused on LLMs, product architecture and collaboration with model providers such as OpenAI and Anthropic. This quarter, Microsoft took further steps toward a One Copilot product organization.

b. Microsoft previously operated Consumer Copilot and Commercial Copilot separately, with relatively independent teams and product entry points across Copilot offerings. These teams were further consolidated this quarter and are gradually forming a unified Copilot product team and entry point.

• The One Copilot strategy has been discussed for some time, but this quarter brought more substantive organizational changes.
• The changes were concentrated in product and engineering. Sales and customer support channels were also adjusted, although the changes there were more limited.

c. Potential benefits of the reorganization include:

• Copilot product and engineering resources that were previously spread across product lines are being consolidated, which should reduce duplicated work across teams.
• The user experience across Copilot, CoWork and other Copilot products could become more consistent, reducing fragmentation across entry points and applications.
• Microsoft intends to build a unified Copilot entry point similar to a “Super App,” through which users can access a broader set of Copilot products and capabilities.

d. The reorganization has also affected near-term organizational efficiency:

• The former product teams each had relatively complete front end, back end and product functions. Following the consolidation, there is meaningful overlap in personnel and responsibilities.
• The larger organization and team consolidation have increased management complexity. Some responsibilities, project ownership and product priorities are still being reassigned.
• Several engineering projects were affected by the reorganization this quarter. The changes are likely to continue for some time and may weigh on product development and project execution in the near term.
• Over the longer term, the reorganization should help Microsoft concentrate Copilot resources and deliver a more consistent product experience. At present, however, personnel, teams and projects are still being realigned, which may pressure near-term efficiency..

6. Copilot product improvements and new features

1. Copilot product feedback continues to improve:
   1. NPS was measured using a new complex task methodology this quarter and is not directly comparable with last quarter’s channel check.
   2. Product feedback…