Claude Enterprise: a UK buyer's guide to pricing, data residency and deployment

Claude Enterprise is Anthropic's plan for organisations that need identity management, audit and compliance controls, and data-residency options that standard Claude plans do not provide. It is priced per user per month, billed annually, with that seat fee covering access only; usage across Claude, Claude Code, and the desktop tools is billed separately at standard API rates. For UK buyers, the two questions that decide the evaluation are usually data residency and total cost, and both have clearer answers in 2026 than they did a year ago. This guide sets out what the plan includes, how it is priced, where UK and EU data can be processed, and how to run a sensible evaluation.

It is written for UK enterprise buyers: IT, security, procurement, and the budget holder sponsoring a Claude rollout.

What Claude Enterprise includes that standard plans do not

Claude Enterprise builds on the Team plan and adds the controls that enterprise IT, security, and procurement functions require before they will approve a tool for regulated or sensitive work. The security and administration set is the substantive difference, not the model. The same Claude models are available on lower plans; what Enterprise adds is the governance wrapper around them.

The headline controls are SCIM for automated user provisioning and de-provisioning, and audit logs that capture user actions, system events, and data access; single sign-on ships on Team; Anthropic's Team help article also lists domain capture and role-based permissions, while its pricing comparison marks those Enterprise-only, so confirm them for your tenant. For data handling, Enterprise adds custom data retention controls, a Compliance API for programmatically retrieving usage data including activity logs and chat content, an Analytics API for adoption metrics, and customer-managed encryption keys, where you provision a key in your own cloud and Anthropic uses it to protect your team's chats, projects, and files. There is also an option to keep inference within the United States for organisations that want that specific control.

Enterprise covers Claude Code and the Cowork desktop tooling under a single agreement, which matters for buyers who do not want engineering and business adoption fragmented across separate contracts and security reviews. Managed connectors let Claude retrieve context from Google Drive, Gmail, Google Calendar, Microsoft 365, Slack, and GitHub without manual uploads, subject to the access you grant.

How Claude Enterprise is priced

Claude Enterprise uses a single seat type, priced per user per month and billed annually. The important structural point for budgeting is that the seat fee covers access only. It includes no token allowance and imposes no per-seat usage cap. Every token your team consumes, in chat, in Claude Code, or in the desktop tools, is billed separately at standard API rates on top of the seat cost. Administrators can set spend limits at both the organisation and individual user level to keep consumption predictable.

Anthropic now publishes a headline figure of $20 per seat per month, billed annually, with pricing stated to be subject to change at Anthropic's discretion. Also published, and useful for planning, are the purchase routes and their minimums. The plan is available two ways. Self-serve, purchased online, carries a minimum of 20 seats, is billed in US dollars, and takes usage credits purchased upfront. Sales-assisted, arranged through Anthropic's sales team, carries a minimum of 50 seats, supports multiple currencies including pounds sterling, bills usage monthly in arrears, and is the route that offers invoicing and dedicated support; a HIPAA-ready configuration with a Business Associate Agreement is available to eligible organisations on either route. Enterprise is also available through the AWS Marketplace, which some organisations prefer for procurement and committed-spend reasons.

The realistic way to model cost is therefore three numbers, not one: the annual per-seat fee multiplied by your seat count, plus expected consumption at API rates, plus the internal cost of the rollout itself. The third number is the one most evaluations underestimate, and it is where the maturity of your existing governance makes the biggest difference. Because Enterprise usage is unmetered at the seat level and billed on consumption, the spend-limit controls are not optional housekeeping; set them before go-live.

The data residency question for UK and EU buyers

For UK and EU buyers, data residency is usually the first question, and the answer has two layers worth separating.

The first layer is the first-party Claude Enterprise plan. Here the relevant control is the option to keep inference within the United States, alongside customer-managed encryption keys and the contractual commitment that customer data from commercial deployments is not used to train Anthropic's models. That US-only inference option is a control some buyers want, but it is a United States control, not a UK or EU one, so it does not by itself satisfy a requirement to keep data and processing inside the UK or EEA.

A partner-cloud API deployment is a separate product and architecture from a Claude Enterprise subscription. As at 14 October 2026, AWS lists London (eu-west-2) in-region availability for Claude Sonnet 4.6 on bedrock-runtime, but the same model card gives its in-region endpoint as N/A. Confirm the endpoint and account access with AWS before relying on UK inference. AWS separately lists Claude Haiku 4.5 on bedrock-mantle for in-region inference in Ireland (eu-west-1) and Stockholm (eu-north-1). Storage, retention and transfers need separate assessment. Our dated Claude residency matrix lists the models and endpoints individually, including documentation conflicts.

The practical guidance for a UK buyer is to confirm the exact region and model availability for your requirement at the point of purchase rather than assuming a specific UK region is live. On the compliance side, Anthropic publishes support for GDPR with regional processing within the EU and EEA, a Data Processing Addendum, HIPAA support for protected health information, and the security certifications a UK procurement team will look for, including SOC 2 Type 2 and ISO/IEC 27001, 27017, and 27018. As one published example of in-region public-sector use, Anthropic cites the European Parliament running Claude via AWS Bedrock.

Deployment options and what they change

The self-serve and sales-assisted routes deliver the same features and the same seat pricing; the differences are commercial. Self-serve suits a smaller, US-dollar, credit-card purchase at 20 seats or more and gets you running quickly. Sales-assisted is the route for UK organisations that need invoicing in sterling, multi-currency billing, or dedicated support, and it starts at 50 seats. If your evaluation involves regulated personal data, the sales-assisted route plus a cloud-platform regional deployment is usually the combination that clears a UK security review.

A point worth making to budget holders: the connectors and Claude Code coverage mean the plan is not only a chat licence. The same agreement can support knowledge retrieval across your document estate, engineering work, and desktop task automation, which changes the comparison from "another chatbot subscription" to "a governed AI platform across several teams". That is the framing that makes the per-seat cost defensible, or not, depending on how many of those uses you will genuinely adopt.

As a firm Certified in the Claude Partner Network, The AI Consultancy runs these evaluations for UK organisations, and the recurring lesson is that the plan choice is downstream of two decisions: where your regulated data must be processed, and how many of the platform's uses you will actually deploy in year one. Buyers reasonably want the people advising on that to have examined Claude expertise, which is the practical reason to ask which of them hold a current Claude Certified Architect certification.

How Claude Enterprise differs from Team, and from ChatGPT Enterprise

Against the Team plan, Enterprise is distinguished by the security and compliance layer: SCIM, audit logs, custom retention, the Compliance and Analytics APIs, customer-managed keys, the US-only inference option, and contractual structures such as a Business Associate Agreement for HIPAA-relevant work; single sign-on is on Team, while Anthropic’s help and pricing pages disagree on Team domain capture and role permissions, so confirm those controls in writing. Team is the right plan for collaboration without those controls; Enterprise is the plan once IT, security, and procurement need evidence and governance.

Against ChatGPT Enterprise, the comparison turns less on raw capability and more on governance posture, data-residency routes, and how each plan meters and bills usage. We cover that decision in detail in our Claude vs ChatGPT enterprise decision guide; the short version is that both are credible enterprise platforms and the right answer is the one that fits your existing cloud, your data-residency requirement, and your adoption plan, not a generic ranking.

How to run the evaluation

A sensible UK evaluation tends to run in this order. Define the data-residency requirement first, because it determines whether you can use the first-party plan or need an in-region cloud deployment. Size the seat count honestly against who will use the platform in the first year, not who might eventually. Model usage cost at API rates for your expected consumption, and set organisation and per-user spend limits before go-live. Choose the purchase route, self-serve or sales-assisted, against your need for sterling invoicing, a BAA, and support. Then pilot on a low-sensitivity, high-frequency use case to build the evidence base before expanding to regulated work.

For the underlying delivery, see our Claude implementation service, and for help running the evaluation itself, our Claude consulting service.

Vendor due diligence now has a primary source. Anthropic filed a preliminary S-1 on 26 October 2026, and our Anthropic IPO tracker sets out the filed revenue, losses, cash, customer metrics, compute commitments and governance with the page each figure came from, together with what the filing changes for a UK buyer's contract.

Sources

• Claude Help Center, "What is the Enterprise plan?", pricing and minimums rechecked 14 October 2026 (per-seat annual pricing, usage at API rates, 20-seat self-serve and 50-seat sales-assisted minimums, SCIM, audit logs, customer-managed encryption keys, US-only inference, HIPAA BAA, AWS Marketplace).
• Claude, "Regional Compliance", accessed June 2026 (regional deployment options requiring a separate model and endpoint check; global vs regional endpoints; GDPR, SOC 2 Type 2, ISO/IEC 27001, 27017, 27018; European Parliament via AWS Bedrock).
• Anthropic Privacy Center, "How does Anthropic protect the personal data of Claude users?", accessed June 2026 (international transfers via adequacy decisions and Standard Contractual Clauses).
• Anthropic, "Claude is now available in Europe" (EU availability of Claude.ai, iOS, and Team plan).