Claude Enterprise in the EU: Pricing vs Bedrock 2026
Claude Enterprise pricing, seat minimums, SSO, SCIM and retention compared with Claude on Bedrock, Google Cloud and Foundry for European companies in 2026.
TL;DR
Claude Enterprise costs US$20 per seat per month, billed annually, plus all usage at API token rates, with a minimum of 20 seats (50 if bought through sales). It adds SCIM, audit logs and custom retention over Team, but has no EU data residency. For EU-only processing, run Claude through Bedrock or Google Cloud instead.
What does Claude Enterprise cost in 2026?
Claude Enterprise now has a published price: US$20 per seat per month, billed annually, and every token used in Chat, Claude Code or Cowork is billed on top at standard API rates. The minimum is 20 seats self-serve or 50 seats sales-assisted, so the entry ticket is US$4,800 a year before usage.
The Claude pricing page describes the plan as “Seat price + usage at API rates”. The billing help article spells out what that means: the seat fee buys access to Claude on web, desktop and mobile, Claude Code and Cowork, but includes no usage allowance at all. If you want Cowork without the seat fee, our Claude Cowork enterprise guide shows the 3P setup on Bedrock or Google Cloud.
| Item | Claude Team | Claude Enterprise |
|---|---|---|
| Seat price | Standard US$20/month annual (US$25 monthly); Premium US$100/month annual (US$125 monthly) | US$20/seat/month, billed annually |
| Usage | Included, with limits (Premium = 5x standard) | Billed separately at API rates, no seat-level usage limits |
| Seats | 2 to 150 | Minimum 20 (self-serve) or 50 (sales-assisted) |
| Billing | Monthly or annual | Self-serve: card or ACH, USD only. Sales: invoice, multi-currency |
| Tax | Prices exclude applicable tax | Prices exclude applicable tax |
Sources: Claude pricing, What is the Enterprise plan?.
Three details matter for budgeting in Europe:
- USD, not EUR. No EUR list price is published, and self-serve Enterprise is USD only. Multi-currency invoicing exists only on sales-assisted contracts.
- Usage is open-ended. Self-serve plans prepay credits and usage stops when they run out. Sales-assisted plans are billed monthly in arrears. You can cap spend per organisation and per user.
- Seats only go up. Self-serve Enterprise cannot reduce seats during the annual term, according to the seat management article. Added seats are prorated.
What does usage cost in practice? Enterprise usage uses the same token prices as the API: Sonnet 5 at US$2 input and US$10 output per million tokens, Opus 5 at US$5 and US$25. As an illustration with our own assumptions: a knowledge worker sending 2 million input and 0.4 million output tokens a month on Sonnet 5 costs US$8 in usage. A heavy Claude Code user at 20 million input and 1 million output tokens on Opus 5 costs US$125, before any prompt caching discount. The seat fee is the small part of the bill for developers and the large part for occasional users.
What do Team and Enterprise include?
Both plans give you SSO, admin controls, central billing and no training on your data by default. Enterprise adds the controls a security or compliance team usually asks for: SCIM provisioning, audit logs, a Compliance API, custom retention, IP allowlisting, customer-managed encryption keys and role-based access with fine-grained permissions.
| Control | Team | Enterprise |
|---|---|---|
| SSO (SAML) and domain verification | Yes | Yes |
| JIT provisioning | Yes | Yes |
| SCIM provisioning | No | Yes |
| Audit logs | No | Yes, plus OpenTelemetry monitoring |
| Compliance API and Analytics API | No | Yes |
| Custom data retention | No | Yes, minimum 30 days |
| Customer-managed encryption keys (CMEK) | No | Yes |
| IP allowlisting, network access control | No | Yes |
| Spend limits per user and organisation | No | Yes |
| HIPAA-ready option | No | Yes |
| US-only inference | No | Yes, at 1.1x API rates |
| EU-only inference or storage | No | No |
Sources: Claude pricing, Claude Enterprise, JIT and SCIM provisioning.
Our view: for a 30-person team without an identity governance mandate, Team is enough. SSO works on Team, and JIT provisioning creates accounts on first login. The moment your IT policy requires automatic deprovisioning via SCIM, or your works council or DPO wants audit logs and a defined deletion period, you need Enterprise. Those two requirements, not the model quality, are what push most European companies up a tier. All seat prices from Free to Enterprise are listed in our Claude pricing plans comparison.
Does Claude Enterprise keep data in the EU?
No. Claude Enterprise offers two inference locations: global (the default) and US-only. There is no EU inference option and no EU storage option documented on any plan page or help article we read. Anthropic acts as your processor for Claude for Work, so the DPA with Standard Contractual Clauses carries the transfer.
The US-only inference article is explicit: the setting is available on usage-based Enterprise plans, bills usage at 1.1x for Opus 4.6, Sonnet 4.6 and later models, and “applies to inference only”. Where your organisation’s data is stored is a separate matter. The Anthropic data residency page for the API lists the same limits: inference_geo accepts only global and us.
Retention on Enterprise works like this:
- Default is indefinite. Chats and projects are kept until users delete them, unless an owner sets a custom period. The retention article states that the minimum custom period is 30 days.
- Exclusions. Custom retention does not cover Claude Design, Claude Tag, Claude Managed Agents or features built on Claude Code on the web. Check whether your teams use them before you promise a deletion period to your DPO.
- Deleted chats leave Anthropic’s backend within 30 days, per the Privacy Center. Content flagged for usage policy violations can be kept for up to 2 years.
Anthropic acts as processor for Claude for Work, and the Consumer Terms do not apply, per the Privacy Center. That is a sound legal basis for internal documents and code. For HR files, health data or customer tickets with personal data, the missing EU residency is a real gap, as we explain in our Claude GDPR provider comparison. If ChatGPT is still on your shortlist, our Claude vs ChatGPT comparison for Europe shows that ChatGPT Enterprise documents EU residency.
How does Claude Enterprise compare with Bedrock, Google Cloud and Foundry?
Claude Enterprise is a finished workplace product with apps, connectors and admin console. Bedrock, Google Cloud and Foundry sell model access by the token, with no seats and no chat interface. Our Amazon Bedrock guide explains the AWS platform beyond Claude. Only Bedrock and Google Cloud keep Opus 5 and Sonnet 5 inference inside the EU, at a 10% premium over their global price. Our Claude API pricing EU comparison lists the per-token rates for each platform.
| Criterion | Claude Enterprise | AWS Bedrock | Google Cloud (Vertex AI) | Microsoft Foundry |
|---|---|---|---|---|
| What you buy | Seats + usage | Tokens | Tokens | Tokens |
| Fixed cost | US$20/seat/month, min. 20 seats | None | None | None |
| Sonnet 5 per 1M tokens (in/out) | US$2 / US$10 | Global US$2 / US$10; EU geo US$2.20 / US$11 | Global US$2 / US$10; EU US$2.20 / US$11 | US$2 / US$10 |
| EU processing | No (global or US-only) | Yes, EU geo profile | Yes, eu multi-region |
No EU data zone |
| Chat app, connectors, Projects | Yes | No, build or buy your own | No, build or buy your own | No |
| Claude Code | Included | Yes, CLAUDE_CODE_USE_BEDROCK=1 |
Yes, CLAUDE_CODE_USE_VERTEX=1, CLOUD_ML_REGION=eu |
Yes, CLAUDE_CODE_USE_FOUNDRY=1 |
| Identity | SSO, SCIM on Claude side | AWS IAM | Google Cloud IAM | Entra ID / Azure RBAC |
| Audit trail | Audit logs, Compliance API | CloudTrail | Cloud Audit Logs | Azure Monitor |
| Processor | Anthropic | AWS | Anthropic for Microsoft |
Sources: Claude pricing, Amazon Bedrock pricing, Google Cloud generative AI pricing, Foundry region availability.
The practical consequence: via the clouds you get EU residency but lose the Claude apps. There is no Claude.ai interface running on Bedrock. You either build an internal chat front end, buy one, or limit the use case to developers. For developers, the gap is small. Claude Code runs directly against Bedrock, and the Claude Code Bedrock docs show how to prefer EU inference profiles:
export CLAUDE_CODE_USE_BEDROCK=1
export AWS_REGION=eu-central-1
export ANTHROPIC_BEDROCK_REGION_PREFIX=eu
# opus alias now resolves to eu.anthropic.claude-opus-5
claudeOne limitation from the same page: the WebSearch tool is not available in Claude Code on Bedrock. For the full AWS setup with IAM and SCPs, see our Claude on AWS Bedrock EU guide. On Google Cloud, the equivalent is covered in our Claude on Vertex AI Europe guide.
Which option fits which company?
Pick Claude Enterprise when you want the apps for many non-technical employees and your prompts contain little or no personal data. Pick Bedrock or Google Cloud when EU residency is a hard requirement, when you build Claude into your own product, or when only developers need access. Team fits small groups without SCIM or audit requirements.
| Your situation | Our recommendation | Why |
|---|---|---|
| Under 20 users, SSO is enough | Claude Team | No 20-seat minimum, usage included |
| 20+ users, need SCIM, audit logs, retention | Claude Enterprise | Only plan with these controls on the Claude apps |
| Personal or special-category data must stay in the EU | Bedrock EU geo or Google Cloud eu |
Claude Enterprise has no EU inference |
| Developers only, AWS landing zone exists | Claude Code on Bedrock with eu. profiles |
Pay per token, no seats, EU processing |
| Building a customer-facing product | Bedrock or Google Cloud | Token pricing, cloud IAM, EU endpoints |
| Azure-only procurement, low-risk data | Foundry | Billing through Azure, but no EU data zone |
| Mixed workforce with some sensitive workloads | Enterprise for general staff + Bedrock or Google Cloud for sensitive apps | Split by data class, not by department |
How we would run the decision:
- Classify prompt content. List the top five use cases and mark which contain personal data under GDPR.
- Count active users, not employees. The 20-seat minimum and annual term make overbuying expensive, and self-serve seats cannot be reduced mid-term.
- Estimate usage per seat. Multiply expected tokens by API rates. For developer-heavy teams, usage will exceed the seat fee.
- Check the DPO questions. EU residency, deletion period, audit trail. If residency is non-negotiable, Claude Enterprise drops out for those workloads.
- Pilot both paths for 30 days. Enterprise self-serve for general staff, Claude Code on Bedrock or Google Cloud for developers, then compare cost and adoption.
FAQ
Each answer reflects the pricing page and help articles read on 19 September 2026. Anthropic changed the Enterprise seat model in 2026, and older contracts with Standard and Premium Enterprise seats move to the usage-based model at their next renewal, so confirm the current terms before you sign.
How much does Claude Enterprise cost?
US$20 per seat per month, billed annually, plus all usage at standard API token rates. The minimum is 20 seats self-serve or 50 sales-assisted, which means at least US$4,800 or US$12,000 a year in seat fees before usage. Prices exclude tax and are quoted in USD.
Claude Enterprise vs Claude Team: what is the difference?
Team costs US$20 per standard seat annually with usage included, for 2 to 150 users, and offers SSO and JIT provisioning. Enterprise charges the same seat price but bills usage separately and adds SCIM, audit logs, a Compliance API, custom retention, CMEK, IP allowlisting and US-only inference.
Does Claude Enterprise offer EU data residency?
No. As of September 2026 Claude Enterprise supports global inference or US-only inference at 1.1x API rates. No EU inference or EU storage option is documented. If EU processing is mandatory, use Claude through Amazon Bedrock’s EU geo profile or Google Cloud’s eu multi-region endpoint instead.
Claude Enterprise vs AWS Bedrock: which is cheaper?
For occasional users, Bedrock is usually cheaper because you pay only tokens and no US$20 seat fee. For heavy users, costs converge, since Enterprise usage is billed at the same API rates. Bedrock’s EU geo profile adds 10%, for example Sonnet 5 at US$2.20 input and US$11 output per million tokens.
Is data from Claude Enterprise used for training?
No, not by default. Anthropic states on its Enterprise page that prompts, data and results are not used to train its models by default, and it acts as data processor for Claude for Work under your organisation’s agreement. The Consumer Terms do not apply to Team or Enterprise accounts.
Can I use Claude Code under Claude Enterprise and on Bedrock at the same time?
Yes. Claude Code is included in the Enterprise seat, and it can also authenticate against Amazon Bedrock with CLAUDE_CODE_USE_BEDROCK=1. Many teams keep Enterprise for general staff and route sensitive repositories through Bedrock’s EU inference profiles, which are billed through AWS instead of Anthropic.
Sources
- Claude: Plans and pricing (19 September 2026)
- Claude: Enterprise plan (19 September 2026)
- Claude Help Center: What is the Enterprise plan? (19 September 2026)
- Claude Help Center: How am I billed for my Enterprise plan? (19 September 2026)
- Claude Help Center: Enable US-only inference (19 September 2026)
- Claude Help Center: Custom data retention controls (19 September 2026)
- Claude Help Center: Set up JIT or SCIM provisioning (19 September 2026)
- Anthropic Privacy Center: How long do you store my organization's data? (19 September 2026)
- Claude Code docs: Claude Code on Amazon Bedrock (19 September 2026)
- Anthropic docs: Data residency (18 September 2026)
- AWS: Amazon Bedrock pricing (18 September 2026)
- Google Cloud: Generative AI pricing (18 September 2026)
- Microsoft Learn: Foundry Models from partners and community (18 September 2026)