Claude Office usage trails Copilot in reported Microsoft data | Magica
Claude trails Copilot in reported Office usage, with limits to the comparison
Editorial Team
••📖5 min read
Reported internal Microsoft data puts Claude for Microsoft 365 usage below one-tenth of Copilot’s. Undisclosed measurement details and Claude’s availability inside Copilot limit what the gap says about Anthropic’s workplace business.
Reported internal Microsoft data puts Claude for Microsoft 365 usage below one-tenth of Copilot’s, with the gap widening.
Anthropic offers Office add-ins and cloud deployment options; its models are also available within Microsoft’s competing assistant.
Eligible European customers face a December 9 change to Claude access settings in Copilot.
Usage of Claude for Microsoft 365, AI developer Anthropic’s tools for working inside Microsoft’s Office applications, is less than one-tenth of Microsoft’s Copilot usage and falling further behind, according to internal Microsoft data reported by The Information. A November 16 Reuters dispatch carrying that report gives the ratio but no underlying counts or measurement period.
The comparison concerns competing assistants within Microsoft’s workplace software. Microsoft owns Word, Excel, PowerPoint and Outlook and sells its own integrated AI assistant, Microsoft 365 Copilot. Anthropic supplies Claude through separate add-ins, but also supplies models available inside Copilot. That makes the contest over the customer interface and deployment route different from the contest over whose AI models get used.
Microsoft's Copilot in Excel model picker includes Claude from Anthropic, a different access route from Anthropic's separate Office add-ins. Source: Microsoft. Source: Microsoft.
The reported gap has an undisclosed measurement basis
The accounts trace the usage claim to the same original reporting; they do not provide independent measurements confirming it. A market commentary on the report cautions that the figures are unverifiable and reflect one vendor’s vantage point. The underlying counts, definition of usage and comparison period are absent from these accounts.
A widening relative gap does not establish that Claude usage is declining in absolute terms. Nor does usage alone establish which assistant produces better work, saves more time or earns more revenue per customer.
Records detail Chinese data-center investments linked to Hexi founder Kang Wei, husband of former Megaspeed chief Alice Huang. Filings document financing and Huawei business, but do not establish a route for restricted Nvidia chips.
Editorial Team
A separate account citing Bloomberg puts paid Copilot licenses above 30 million at the end of June 2026. That is a commercial scale indicator, not an active-user count or the missing denominator for the reported usage ratio.
That account also says Anthropic’s November 10 introduction of Claude Meetings, Sheets and Whiteboards unsettled Microsoft investors. Those newer workplace tools are distinct from the Office add-ins already available since May. The reported Office comparison therefore cannot, on its own, settle the prospects of that broader expansion.
A head start, but more than one deployment route
Microsoft’s enterprise launch announcement dates Copilot’s general availability to November 1, 2023. It described integration with Microsoft 365 applications, access to organizational information through Microsoft Graph and use of existing security and compliance policies. Copilot’s position inside the suite predates this competitive snapshot by three years.
Anthropic’s Office offering has a shorter history. The add-ins for Excel, PowerPoint and Word became generally available in May 2026, while Outlook entered public beta for paid plans. In its product announcement, Anthropic says a conversation retains context across applications and can carry changes between open spreadsheets, presentations and documents. It says Outlook can triage messages and prepare replies and invitations, with users reviewing them before sending.
Anthropic also describes administrative controls: deployment through Microsoft’s admin center, activity reporting by user, application and day, and enterprise telemetry covering prompts, tool calls and document references. These are company descriptions of functionality, rather than independent evidence of productivity or adoption.
The announcement also carries customer testimonials from Citadel, Bain & Company and ServiceNow describing spreadsheet work and, in ServiceNow’s case, adoption. Those company-selected endorsements provide examples of use, not a representative comparison with Copilot.
The deployment choices complicate a simple incumbent-versus-outsider explanation. Anthropic offers access on paid Claude plans, but says organizations can also route the four add-ins through an existing language-model gateway to Claude on Amazon Bedrock, Google Cloud’s Vertex AI or Microsoft Foundry. It separately says Microsoft 365 Copilot customers can use Claude models directly in Excel and PowerPoint.
Microsoft’s existing licensing relationships may help it distribute Copilot. The separate account characterizes additional contracts and privacy reviews as friction for Claude, but supplies no measured effect. The gateway options and shared Microsoft admin center also narrow any claim that every Claude deployment requires an entirely separate procurement and administration system. The accounts supply no like-for-like comparison of total deployment costs, so the usage gap cannot be treated as proof of a price or cost advantage.
Microsoft's notice schedules default Claude access for eligible EU and EFTA tenants without an Anthropic selection on December 9. Administrators can opt out or limit access; other existing selections are respected. Source: Microsoft 365 Message Center Archive.
December puts an access decision before European administrators
Microsoft’s next scheduled change could broaden access to Anthropic’s models through Copilot. An archived Microsoft customer notice published November 10 says processing for Claude models in Copilot will move into the EU Data Boundary, Microsoft’s regional data-processing commitment, in early December 2026. Until then, the notice says, those models are outside that processing commitment.
Claude models are currently disabled by default for tenants in the European Union and European Free Trade Association. On December 9, Microsoft plans to enable them for all users in eligible tenants that have made no Anthropic access selection. It will also retire the option allowing Claude only in Copilot’s Microsoft 365 applications; tenants with that selection will move to access for all users.
Other existing selections will be respected. Administrators can select no users or specific users and groups before the change, and can disable or limit access afterward. Anthropic continues to act as a Microsoft subprocessor. A separate setting for Anthropic models with data retention is outside this change.
For affected organizations, the immediate decision is which employees should have access under the new processing arrangements. For assessing the rivalry, the next useful evidence is a consistently defined comparison of active use, paid seats and model consumption that distinguishes Claude add-ins from Claude running inside Copilot. Default availability alone will not show whether employees choose to use either route.
ElevenLabs plans to acquire Kotoba, the startup behind Japanese speech technology used in Salesforce’s Agentforce Voice. The deal adds a 15-person team and deployed products to its existing Japanese operation, but financial terms and integration plans remain undisclosed.
Tesla plans €3 billion of expansion at its profitable Grünheide factory to build Cybercabs from 2029, eventually targeting up to one million annually. The driverless taxi still lacks German and EU approval, and its capacity target differs from an older plan for the whole site.
Our World in Data’s comparison finds negative views of AI exceed positive ones across five probability-based surveys. The primary results also reveal substantial mixed views, acceptance of some applications and a sharper divide over who controls development and bears its local costs.
NVIDIA's Nemotron for Science pairs an existing open model with nearly 50,000 chemistry questions and scientific post-training recipes. Partner results suggest scope for specialized agents, but the reported accuracy and inference savings do not establish the full cost of adapting them.
AMD says the Lux AI supercomputer is open to scientific users at Oak Ridge. Half its annual capacity supports the Genesis Mission and half is reserved for paid access, including AMD development, while workload limits and undisclosed rates shape its practical value.
Apple is adding app-by-app and folder-specific permissions, repeat approval requests and clearer warnings in macOS 27.3 beta 2. Existing permissions remain intact, limiting the immediate reach of the privacy changes.
OpenAI reportedly discussed buying Goodfire for more than $5 billion, over four times its February financing valuation. No deal resulted. A disclosed paid assessment agreement offers cooperation without ownership, while the technology’s safety benefits remain to be demonstrated.
Gemini’s regional voice rollout includes more than ten UK options and applies to live conversations and spoken chat responses. Earlier accent consistency problems remain untested in the new release, while pricing and rollout eligibility are unspecified.
Hugging Face is moving language-model requests to OpenRouter’s provider router, adding automatic failover while ending custom provider keys on that route. Provider rates carry no Hugging Face markup, but the selected service—and the resulting bill—can change.
Nasiko’s OpenRuntime combines agent spending reports, model routing and gateway budget limits. Earlier releases and company benchmarks show what it can do, but metering disputes, workload trade-offs and runtime charges complicate the savings case.
A report citing business records identifies Anthropic cybersecurity researcher Paul Crowley as a manager of Red Means No. The disclosure adds to scrutiny of consent safeguards and professional networking at the Bay Area sex parties.
South Korea is strengthening device checks and considering public Wi-Fi shutdowns before its November 19 college entrance exam, after a tax-accountant exam top scorer admitted using AI to cheat. The response still depends heavily on human screening, with no electronic device detectors planned for this year's test.
Russia’s objections to observer participation stalled the opening of a Geneva weapons conference, where governments must decide whether to authorize negotiations on autonomous weapons and whether any resulting rules should be binding.
Apple’s iOS 27.3 beta ties higher allowances for server-based AI to existing iCloud+ plans. Basic Siri commands remain free, but undisclosed daily limits leave the value of paid access uncertain.
The partners plan to combine IBM quantum processors, NVIDIA computing and Fermilab physics expertise to study modular quantum systems. A separate decoder release promises faster error-correction research, with important limits on its performance comparisons.
Federal Reserve Chairman Kevin Warsh outlined AI scenarios spanning shared prosperity, job displacement and an investment downturn. The exercise offers no forecast or rate guidance as colleagues worry that the infrastructure buildout is adding to inflation.
OpenAI is offering approved organizations a free classifier for suspected child sexual abuse images, including AI-generated material. Early results suggest less review work, but existing detection alternatives and the need for human review qualify the launch.
Nvidia plans scientific computing racks for early 2027, extending Vera beyond its initial AI deployments. Its 2.4× performance-per-watt claim comes from one airflow simulation against AMD Turin; performance against newer rivals and delivered costs remain unresolved.
OpenAI’s CEO expressed empathy for Nvidia founder Jensen Huang but withheld judgment on reinstatement. His confidence in continued AI financing comes with a direct commercial stake and unresolved deployment risks.