Is Microsoft Copilot catching up with Claude?

Is Microsoft Copilot finally catching up with Claude?

Microsoft Copilot is catching up with Claude in workplace AI, and the gap has narrowed considerably. Microsoft now offers much stronger AI agents, access to Anthropic's models, and a growing range of tools that can complete tasks across Microsoft 365. Claude still has important advantages in advanced coding and direct access to frontier AI capabilities.

The change became obvious in March 2026, when Microsoft introduced Copilot Cowork in collaboration with Anthropic. Instead of simply answering questions or drafting emails, Copilot could take on longer assignments involving documents, meetings, spreadsheets, and other workplace information. Copilot taking on longer assignments is one of the shifts of the past year, and we cover what changed in our agentic AI market report.

Microsoft had already begun adding Claude models to its products in September 2025. Within roughly six months, that relationship had expanded from selected research and agent-building tools into Microsoft's core productivity strategy. Microsoft also brought agentic capabilities into Word, Excel, and PowerPoint.

The commercial results followed. Microsoft's fiscal fourth-quarter earnings showed more than 30 million paid Microsoft 365 Copilot seats, compared with 15 million two quarters earlier. The company doubled its reported paid-seat base in approximately six months.

But Anthropic has been moving just as quickly. Claude now works directly in Microsoft Office applications, while Claude Code and Claude Cowork have become important parts of its business.

What would it mean for Microsoft Copilot to actually catch Claude?

Microsoft Copilot would catch Claude when people could rely on either assistant to complete the same demanding work with similar results, costs, and effort. Having access to the same AI model gets Microsoft closer, but the experience of using each product still matters.

The distinction starts with what the companies sell. Anthropic develops the Claude models and offers products including Claude, Claude Code, and Claude Cowork. Microsoft's Copilot family includes Microsoft 365 Copilot for office work and GitHub Copilot for software development, among other products.

Microsoft can combine models from Anthropic, OpenAI, and other developers with its own search, document access, security, and application tools. Anthropic controls the development of Claude and can build new capabilities directly around its models.

The useful tests are task quality, user preference, and commercial adoption. Can Copilot complete complicated assignments as successfully as Claude? When employees have access to both, which do they choose? And are companies replacing separate Claude subscriptions with Copilot, or keeping both?

Microsoft has already made enormous progress on commercial distribution. The evidence on task quality is more complicated, and publicly available data on direct user preference remains limited.

Has Microsoft Copilot closed the intelligence gap with Claude?

Microsoft Copilot can now offer Claude-level intelligence on supported tasks because Microsoft has integrated Anthropic's models directly into its products. That gives Copilot access to capabilities that previously required using Claude separately.

Microsoft's decision to offer multiple AI models has changed the comparison considerably.

In September 2025, Microsoft introduced Claude Sonnet 4 and Opus 4.1 into selected Microsoft 365 Copilot experiences. Its April 2026 product roadmap subsequently listed Claude Opus 4.7 for Copilot Cowork, Copilot Studio, and Excel.

Anthropic continued improving its models independently. Claude Opus 5, released in July 2026, demonstrated substantial advances in software engineering and knowledge-work evaluations, including Frontier-Bench and GDPval-AA. Anthropic also reported improvements over earlier Claude versions on demanding agentic coding tasks.

Those results are impressive, although they cannot tell us which finished assistant is better. Benchmark performance depends on the task, testing configuration, tool access, and the amount of computing used.

A Copilot user working with an Anthropic model can benefit from some of these advances. However, the result also depends on how Microsoft's application retrieves information, handles instructions, calls tools, and verifies changes.

There is still no comprehensive independent comparison showing that Microsoft 365 Copilot consistently equals native Claude across complex professional workflows.

We think Microsoft has substantially reduced the model-quality disadvantage that once made Copilot less attractive for demanding work. The unresolved gap now lies more in execution and product experience than raw language-model capability.

Why is Microsoft using Claude inside Copilot?

Microsoft is using Claude because Anthropic has developed technology that can strengthen Copilot faster than Microsoft could necessarily reproduce it internally. The partnership also allows Microsoft to offer customers a choice of leading AI models without depending on one provider.

The relationship has become much deeper over time.

Microsoft initially introduced Claude models into Researcher and Copilot Studio in September 2025. In November, Microsoft, NVIDIA, and Anthropic announced an expanded partnership involving Azure capacity and potential investments of up to $5 billion from Microsoft and $10 billion from NVIDIA.

Then came Copilot Cowork. Microsoft's March 2026 announcement explicitly credited Anthropic with the underlying technology behind Claude Cowork that it was bringing into Microsoft 365.

Across three major stages, Microsoft moved from offering Anthropic models as alternatives to incorporating Anthropic's agent technology into one of its most important workplace AI products.

Microsoft's advantage is its ability to package these capabilities with enterprise data, security, and existing software. Anthropic gains another way to distribute its technology and generate model demand.

The relationship makes a simple technological winner increasingly difficult to identify: Microsoft can improve Copilot through Claude's advances, while Anthropic benefits from more Microsoft customers using its models.

Is Microsoft Copilot growing faster than Claude?

Microsoft 365 Copilot is growing extremely fast in paid seats, but Anthropic's revenue has also exploded. We cannot reliably identify the faster-growing product from these figures alone because the companies disclose different measures.

Microsoft reported 15 million paid Microsoft 365 Copilot seats in its fiscal second quarter, more than 20 million in its third quarter, and more than 30 million in its fourth quarter. Copilot seats doubling in two quarters is a big move, and we look at who is ahead and who is falling behind in our report on who's ahead in agentic AI.

The reported seat base therefore doubled over two quarters. The progression also indicates accelerating additions: Microsoft added at least 5 million seats between the first two milestones and reached more than 30 million in the following quarter. Microsoft itself reported that net paid-seat additions more than doubled sequentially during the final quarter.

Anthropic's trajectory is equally unusual. According to Reuters reporting in August 2026, the company's annualized revenue exceeded $65 billion at the end of July, compared with $47 billion in May and approximately $9 billion at the end of 2025.

That represents more than sevenfold growth from the end of 2025 in roughly seven months. The $18 billion increase between May and July alone was twice Anthropic's entire annualized revenue at the previous year-end.

Annualized revenue is calculated from the current pace of business, so those figures should not be mistaken for revenue already earned over a full year. They also include Anthropic's API business, coding tools, and other products.

Reporting period	Microsoft 365 Copilot	Anthropic
End of 2025	No directly comparable reported figure	Approximately $9B annualized revenue
Microsoft FY2026 Q2	15M paid seats	Different reporting period
February 2026	No corresponding seat update	More than $14B annualized revenue
Microsoft FY2026 Q3	More than 20M paid seats	Different reporting period
April 2026	No corresponding seat update	More than $30B annualized revenue
May 2026	No corresponding seat update	More than $47B annualized revenue
Microsoft FY2026 Q4	More than 30M paid seats	Different reporting period
End of July 2026	No corresponding seat update	More than $65B annualized revenue
Measured increase	Approximately 2× in two quarters	More than 7× since end-2025

Both companies are clearly attracting substantial demand. Microsoft is rapidly expanding Copilot distribution, while Anthropic is converting frontier AI capabilities into one of the fastest-growing enterprise software businesses.

Does Microsoft Copilot already have more users than Claude?

Microsoft Copilot has much broader built-in distribution, but we cannot establish that more people actively use Copilot than Claude from the available public figures.

Microsoft reported approximately 150 million monthly users across its consumer and commercial Copilot experiences in early 2026, according to figures clarified to TechCrunch. That followed an earlier disclosure of more than 100 million monthly users.

These totals cover a wide set of Copilot experiences. They should not be confused with the paid Microsoft 365 Copilot product, whose disclosed figure is more than 30 million seats.

A paid seat also says little about usage frequency. An employee may have access to Copilot but rarely open it, while another may use Claude for several hours every day.

Anthropic does not publish a directly comparable series covering active Claude users across its consumer applications, enterprise accounts, and developer tools.

Microsoft's distribution advantage is clear: Copilot can reach employees through software they already use. Claude has to earn much of that attention through product preference, subscriptions, developer adoption, or integrations. But without comparable active-user figures, we cannot say which assistant people use more.

Are employees really using Microsoft Copilot at work?

Microsoft Copilot has demonstrated genuine workplace usage, although adoption varies sharply depending on what employees are trying to do.

The UK government's Microsoft 365 Copilot experiment provides a useful independent example. Conducted across participating public-sector organizations in 2024, the study examined how people used the assistant in actual working conditions.

Among survey respondents, 39% reported using Copilot multiple times daily and another 43% at least multiple times weekly.

But adoption was far from uniform across Microsoft's applications. Teams reached a maximum adoption rate of 71%, compared with 23% for Excel and 24% for PowerPoint.

Meeting summaries and communication-related work attracted much more participation than spreadsheet or presentation tasks during the experiment.

Accenture published another useful dataset from an earlier Copilot rollout. The company reported approximately 101,500 assigned licenses, a 79% active-user rate, and 2.8 million actions over four weeks.

Using those figures, we calculate roughly 80,000 active users and approximately 35 actions per active user during the period.

An average of about nine actions per active user per week suggests recurring usage, although it also allows for a substantial difference between frequent and occasional users.

These studies predate Microsoft's latest agent capabilities, so they cannot establish current global behavior. They do, however, show regular employee usage, particularly in communication-heavy workflows.

Is Claude still ahead of Microsoft Copilot for coding?

Claude remains one of the strongest choices for demanding AI-assisted coding, while Microsoft's GitHub Copilot leads on several reported distribution measures.

Anthropic disclosed in February 2026 that Claude Code had exceeded $2.5 billion in annualized revenue. The figure had more than doubled since the beginning of the year.

During the same period, weekly active Claude Code users doubled, business subscriptions quadrupled, and enterprise customers came to represent more than half of Claude Code revenue.

Those numbers are unusually strong for a product that became generally available in May 2025. Claude Code developed into a multibillion-dollar annualized business in less than a year.

Microsoft's GitHub Copilot has a different commercial profile. Its fiscal second-quarter results reported more than 4.7 million paid subscribers, up 75% year over year. By the fourth quarter, Microsoft reported approximately 50 million GitHub Copilot users across its broader offering.

GitHub's reach gives Microsoft a strong position in existing developer workflows. Claude Code's growth suggests that many developers are willing to adopt a separate tool for more autonomous coding tasks.

Technically, both products now support agentic work, and GitHub Copilot can use models from several developers, including Anthropic.

We would give Claude Code a particularly strong position in advanced agent-driven development, while recognizing GitHub Copilot's much wider reported reach. Without comparable task-success and retention data, the evidence does not justify declaring a universal coding winner.

Can Claude now compete directly with Copilot in Word, Excel, and PowerPoint?

Claude can now work directly inside Microsoft's main Office applications, which has removed one of Copilot's clearest earlier advantages.

For a long time, Microsoft's strongest selling point was straightforward: Copilot was built into the software people already used. Claude users often had to move between applications, upload files, and transfer generated content. Agents working inside apps people already use are catching on, and we follow which technologies are gaining ground in our report covering agentic AI.

Anthropic has spent the past year closing that gap.

Claude for Excel appeared in beta in October 2025. PowerPoint support followed in February 2026, cross-application context expanded in March, and Word became available in April.

On May 7, 2026, Anthropic made Claude for Excel, PowerPoint, and Word generally available. Claude for Outlook entered public beta at the same time.

The company also introduced a shared conversation that can carry information across open Office files.

Consider an analyst preparing an investment presentation. Claude can examine source documents in Word, work on calculations in Excel, and update a PowerPoint presentation using the same context. Anthropic says changes to underlying spreadsheet figures can flow through to related documents and slides.

Microsoft offers comparable ambitions through Copilot and its own agentic Office capabilities. Its advantage comes from deeper ownership of the surrounding productivity platform, including organization-wide search, permissions, and administration.

Office task	Microsoft 365 Copilot	Claude
Draft a Word document	Native Copilot integration	Claude for Word add-in
Analyze an Excel workbook	Copilot and Excel agents	Claude for Excel add-in
Build PowerPoint slides	Copilot in PowerPoint	Claude for PowerPoint add-in
Triage Outlook emails	Copilot in Outlook	Claude for Outlook, public beta
Work across applications	Microsoft 365 and Work IQ context	Shared conversation across supported Office add-ins
Create standalone Office files	Supported through Copilot workflows	File generation through Claude
Use organizational documents	Microsoft Graph and connected enterprise data	Microsoft 365 connector and supported integrations
Run longer assignments	Copilot Cowork and Autopilots	Claude Cowork and agent workflows
Manage organization-wide access	Microsoft 365 administration and identity	Claude enterprise controls and connector permissions

Does Microsoft Copilot understand company information better than Claude?

Microsoft Copilot currently has a structural advantage in understanding information spread across Microsoft 365, although Claude can now access many of the same corporate data sources.

Microsoft has developed Work IQ to connect documents, communication, people, relationships, and work activity.

During its fiscal third-quarter earnings call, Microsoft described an underlying work-data system spanning more than 17 exabytes, growing 35% year over year.

An exabyte equals one billion gigabytes. The figure therefore describes an enormous pool of workplace information, although it does not mean that Copilot can read all of it for an individual user.

Microsoft's advantage is that Outlook, Teams, SharePoint, OneDrive, and other services already operate within a shared enterprise environment. Identity management and permission systems are part of the same infrastructure.

Claude is catching up through connectors.

Anthropic's Microsoft 365 integration allows users to access relevant information from services including SharePoint, OneDrive, Outlook, and Teams. Authorized configurations can also support actions such as modifying files or working with calendar and communication tools.

Microsoft still has better control over the underlying workplace infrastructure. But Claude can now reach much of the same information, leaving the quality of search, document retrieval, and cross-application execution as the more useful tests.

Has Microsoft Copilot caught Claude in autonomous AI agents?

Microsoft Copilot has made rapid progress in autonomous work, but Claude remains highly influential in the technology powering those capabilities.

Both companies are now developing assistants that can handle assignments involving several steps, rather than requiring a fresh prompt after every action.

Claude Cowork allows users to delegate work involving files, research, and connected applications. Anthropic has expanded that system through plugins and specialized workflows, including financial-analysis tools capable of helping prepare pitchbooks, review compliance files, and produce financial reports.

Microsoft has followed a similar direction through Copilot Cowork and Autopilots.

Copilot Cowork can use workplace information to plan and carry out longer tasks, while Microsoft's Autopilots are designed for autonomous activity over extended periods.

Microsoft's Agent 365 adds another capability: centralized governance for agents operating inside companies.

The company reported nearly 40 million agents registered through Agent 365 only two months after launch, across tens of thousands of organizations.

That sounds enormous, but registrations do not establish how many agents are completing useful assignments every day. The disclosure tells us more about deployment interest and administrative reach than actual productivity.

The difference between the companies is becoming clearer. Anthropic is pushing agent capabilities forward through its models and direct products. Microsoft is concentrating heavily on making agents accessible, manageable, and secure inside large organizations.

Copilot has narrowed the practical gap considerably. We still lack reliable independent measurements proving that its autonomous workflows are as successful as Claude's on demanding, comparable assignments.

Why are big companies choosing Microsoft Copilot when Claude is growing so fast?

Large companies are buying Microsoft Copilot partly because they can deploy it across existing Microsoft 365 environments without introducing an entirely separate productivity platform. But Anthropic's explosive enterprise growth shows that Microsoft's distribution advantage is far from unbeatable.

Microsoft's fiscal third-quarter earnings provide unusually concrete evidence of its position.

Accenture had more than 740,000 Copilot seats, making it Microsoft's largest reported deployment at the time. Bayer, Johnson & Johnson, Mercedes, and Roche had each committed to at least 90,000 seats.

Together, those five organizations represented more than 1.1 million seats or commitments.

Microsoft also reported that the number of customers with over 50,000 seats had quadrupled year over year.

These are large deployments by any enterprise software standard. They show how Microsoft's procurement relationships can turn an AI product into a company-wide service.

Existing identity management, security policies, compliance requirements, and billing arrangements all help Microsoft. A company already paying for Microsoft 365 may find it easier to expand an existing agreement than approve another AI platform for thousands of employees.

Anthropic's enterprise growth provides an interesting contrast.

In February 2026, Anthropic reported more than 500 customers spending over $1 million annually. By April, that figure had exceeded 1,000 customers, doubling in less than two months.

Two years earlier, only about a dozen customers had reached that spending level. The increase from roughly 12 to more than 1,000 represents over eightyfold growth in high-spending business customers.

Anthropic also reported that eight of the Fortune 10 had become Claude customers.

These companies may be buying API access, Claude Code, enterprise subscriptions, or a combination. We cannot assume every dollar represents spending that would otherwise have gone to Copilot.

Nevertheless, Anthropic has demonstrated that major corporations are willing to allocate substantial budgets to Claude even when they already use Microsoft software.

Corporate purchasing and employee preference can also differ. An employee may have Copilot included in a company deployment while choosing Claude for coding, financial analysis, or complex research.

Microsoft has the stronger position for broad company-wide deployment, while Anthropic is proving that specialized AI products can still win substantial enterprise spending.

Is Microsoft Copilot cheaper than Claude?

Microsoft Copilot can be cheaper to operate across an existing Microsoft environment, but Claude's entry-level subscription prices are generally lower. Actual enterprise costs depend heavily on usage and the products employees need.

Microsoft lists Microsoft 365 Copilot at $30 per user per month, typically alongside qualifying Microsoft 365 licensing. Eligible commercial customers also receive access to Copilot Chat, although the free and paid experiences differ. A $30 per-user seat is one way to charge, and we compare which business models are winning in our report on agentic AI.

Anthropic lists Claude Pro at $20 per month, or approximately $17 per month with annual billing.

Claude Team standard seats cost $25 per user per month with annual billing, while premium seats cost $100. Claude Enterprise starts at $20 per seat monthly with annual billing, but usage is charged separately at API rates.

That last distinction is especially important. A $20 Claude Enterprise seat can become considerably more expensive when an employee runs many advanced models and autonomous tasks.

Microsoft also charges separately for certain consumption-based capabilities, including Copilot Cowork, so the $30 subscription does not necessarily cover every advanced workflow.

For a company with 1,000 employees, Microsoft 365 Copilot at its standard $30 monthly price would cost $360,000 annually before other license requirements or additional usage charges.

Claude Team standard seats at $25 monthly would cost $300,000 annually for the same number of users. But the comparison changes if those users need premium seats, heavier usage, or Enterprise API billing.

Product	Published US price	Main limitation or extra cost
Microsoft Copilot Chat	Included for eligible Microsoft 365 customers	Advanced capabilities may require paid licensing or metered usage
Microsoft 365 Copilot	$30/user/month	Qualifying Microsoft 365 licensing
Copilot Cowork	Consumption-based charges	Usage depends on tasks performed
Claude Free	$0	Restricted usage
Claude Pro	$20/month	Usage limits
Claude Pro annual plan	Approximately $17/month	Annual commitment
Claude Team Standard	$25/user/month annually	Seat and usage limits
Claude Team Premium	$100/user/month annually	Higher usage allowance
Claude Enterprise	$20/seat/month annually plus usage	API-based usage charges

Enterprise discounts, taxes, and specific contract terms can change the actual amounts paid.

For a small team choosing a standalone AI assistant, Claude may be the more affordable option. For a large business already committed to Microsoft, Copilot's integration and administration could outweigh a higher subscription price.

Could Microsoft Copilot eventually replace Claude at work?

Microsoft Copilot could replace separate Claude subscriptions for many employees, especially if Copilot continues improving its ability to execute complicated tasks across Microsoft 365.

Microsoft already has large corporate deployments, a familiar interface, enterprise data access, and models from several leading AI developers.

For an employee mainly using AI to summarize Teams meetings, draft Outlook emails, search SharePoint documents, or prepare standard presentations, Copilot may eventually cover almost everything required.

Specialist users present a harder challenge.

A software engineer relying on Claude Code for extensive development work, or an analyst using Claude Cowork for complex research, may have stronger reasons to keep direct Anthropic access.

Claude's latest integrations also make the decision less obvious. It now operates directly in Word, Excel, PowerPoint, and Outlook, reducing the inconvenience of switching between products.

We think Copilot has a realistic path to becoming the default AI assistant for broad groups of Microsoft 365 employees. Replacing Claude among demanding professional users will depend on consistently matching its results and flexibility.

What could stop Microsoft Copilot from catching Claude?

Microsoft Copilot's biggest remaining weakness is the lack of convincing public evidence that its wider distribution translates into equal or better results on difficult professional tasks.

Microsoft has demonstrated that companies are purchasing licenses at scale. It has also shown progress in integrating advanced AI models and building autonomous workflows.

But the commercial numbers tell us little about how well Copilot completes complicated assignments compared with Claude.

A 30-million-seat product could still leave its heaviest users dissatisfied if they regularly need to correct financial models, repair generated slides, or move to another assistant for deeper research.

The earlier UK government experiment illustrated how adoption can vary by application. Microsoft's newer agent capabilities could change that pattern, but comparable current usage data would be needed to demonstrate the improvement.

There is also a product-development risk. Microsoft benefits from Anthropic's model advances, while Anthropic can release new features in its own products without first adapting them to Microsoft's interface and enterprise requirements.

Claude faces its own difficulties. Its connectors must work reliably, its security controls must satisfy large organizations, and its consumption-based pricing must remain acceptable as usage grows.

For now, Microsoft's clearest advantage is the scale of its enterprise platform. Claude's strength lies in the capabilities and direct experience of its AI products.

The crucial missing evidence is a sustained comparison of what employees can actually accomplish with each assistant.

Is Microsoft Copilot catching up with Claude, or has it already caught up?

Microsoft Copilot is genuinely catching up with Claude, especially for everyday enterprise productivity. In some large-company deployments, Microsoft already has the stronger commercial position. But claiming that Copilot has fully caught Claude would go beyond the available evidence.

Microsoft has doubled its paid Microsoft 365 Copilot seat base across two quarters, introduced more capable autonomous workflows, and expanded its use of Anthropic models. Its enterprise integration is now a serious competitive advantage, particularly for organizations deeply invested in Microsoft 365.

Claude remains a powerful independent competitor. Anthropic's annualized revenue grew from approximately $9 billion at the end of 2025 to more than $65 billion by July 2026. Claude Code has developed into a major coding business, and Claude's direct integration with Microsoft Office means Anthropic can now compete inside many of the applications that once made Copilot uniquely convenient.

One of our clearest findings is that Microsoft's progress increasingly depends on technology supplied by the company it is trying to catch. This works commercially for Microsoft, but it limits any claim that Copilot has overtaken Claude technologically.

We would currently give Microsoft the advantage in enterprise distribution and integration, while Claude remains particularly strong in frontier model capabilities, sophisticated coding, and specialized autonomous work.

The final judgment is clear enough: Microsoft Copilot is catching up quickly, and it may become the default AI assistant for many Microsoft 365 customers. Yet Claude's rapid growth and improving workplace integrations show that Anthropic is strengthening its own position at the same time.

For businesses, the decision increasingly comes down to whether Copilot's integration with existing software is good enough to replace a separate Claude subscription. That's where the next stage of the competition will be decided.