ChatGPT data residency: what stays in Europe

A new customer contract arrives with a clause saying customer data must stay in the European Economic Area (EEA), and the leadership team that signs it drafts half its documents in ChatGPT. ChatGPT data residency, meaning customer content stored at rest in a region you pick, is OpenAI's answer to that clause, and on some plans it keeps processing in the same region.¹

The guide covers the location question only, for ChatGPT Enterprise, ChatGPT Business and the API, while the contract and the transfer rules sit in our guide to whether ChatGPT is GDPR compliant. Our ChatGPT Enterprise guide weighs Enterprise against Business, and nothing here is legal advice.

---

What is ChatGPT data residency, and which plans offer it in Europe?

OpenAI splits residency into two layers, and the difference decides what you can promise a customer.¹ Data residency stores in-scope customer content, meaning prompts, files and responses, at rest in the chosen region, with backups and replicas kept there too.¹ Inference residency adds that the GPUs, meaning the chips that run the model, process that content in the same region, and it requires data residency in that region first.¹ Europe means the EEA plus Switzerland on every OpenAI page.

OpenAI launched European residency on 5 February 2025 and added inference in Europe on 16 January 2026, and its help page changed again in the week we wrote this.²¹

ChatGPT data residency in Enterprise: what stays in Europe

A new Enterprise or Edu workspace provisioned with Europe as its region stores six kinds of content in the EEA and Switzerland, together with their backups and replicas:¹

• Conversations in text, image and voice
• ChatGPT Memory, the context it keeps between chats
• Uploaded files such as images and documents
• Custom GPTs with their prompts and outputs
• Code Interpreter and data analysis files
• Image generation inputs and outputs

With inference residency switched on, GPU work on that same content runs in Europe, including the embeddings, meaning number versions of your documents that ChatGPT uses to search them.¹ The switch affects only conversations started after the change, so a thread opened on Monday and continued on Friday keeps its old setting.¹ OpenAI also states that any new feature stays outside residency until its help page names it as supported.¹

ChatGPT Enterprise: what can still leave the region

Residency covers customer content and leaves system data, meaning the data OpenAI needs to run and secure the service, free to sit or run anywhere.¹ Four categories stay outside the promise even with both layers switched on:¹

• Account data: workspace name, metadata, billing details, logins and usage statistics
• CPU steps: extracting text from PDF and Word files, routing and sign-in
• Connected apps and web search: data sent to an app, an MCP server (a connector to another system) or a search provider follows that provider's terms
• Unsupported features: third-party GPTs, beta features and Codex on the web, while the Codex app and command line are covered

Shared GPTs change too, because a workspace with residency cannot set GPT sharing to Anyone, and only colleagues in the same residency region can open a shared GPT.¹ A group with staff in London and Lyon should check that first, since OpenAI lists the United Kingdom as a separate region.¹

How does ChatGPT Business storage differ from Enterprise?

ChatGPT Business lets the buyer pick a storage region during checkout, which decides where prompts and responses rest, although OpenAI says the option is still reaching customers gradually.³ The region does not cover inference, and OpenAI also keeps a copy of every prompt and response in the US for a limited time to monitor abuse.³ OpenAI's pricing table still marks data residency as No for Business, so check the checkout page before you plan around it.⁴

Access differs as well, since OpenAI's privacy page lets authorised employees and specialised third-party contractors review Business conversations for abuse.⁵ On Enterprise, authorised OpenAI employees open conversations only to resolve incidents, to recover a chat with your explicit permission or when the law requires it.⁵ The European Data Protection Board counts remote access from outside the EEA as a transfer, even for data stored in Europe.⁶

How does OpenAI API data residency work in Europe?

Developers on the same team meet a third set of rules, because API residency is set per project, meaning a container for API keys and usage inside your OpenAI organisation.⁷ A project created with Europe as its region stores and processes supported requests in the EEA and Switzerland through the address eu.api.openai.com.⁷ Global projects can also route single requests there, which suits a team that only needs Europe for one workflow.⁷

Outside the US region, OpenAI must approve you for Modified Abuse Monitoring or Zero Data Retention, meaning prompts stay out of its 30-day abuse logs, and you sign a Modified Retention amendment.⁷ Models released on or after 5 March 2026 cost 10% more on residency endpoints, so $1,000 of monthly usage becomes $1,100, about €965 instead of €877.⁷

Can you move an existing ChatGPT workspace to EU residency?

OpenAI publishes no route for moving an existing workspace into a residency region, because its help page describes content stored in the region when a new workspace is provisioned.¹ Self-service export is also unavailable in Business and Enterprise workspaces, so members cannot simply download their history and carry it across.⁸

Take a made-up leadership team of 8 people with 14 months of chats and 6 custom GPTs in ChatGPT Business, moving to an Enterprise workspace in Europe. Before the move, every prompt rests wherever checkout put it while the model runs anywhere, and after it, new prompts rest and run in Europe from the first day.¹³ Plan as if the history stays behind, so someone copies the instructions and files of all 6 custom GPTs across by hand before the old workspace closes.

What does EU residency cost on ChatGPT Enterprise?

Storage in Europe costs nothing extra on Enterprise and Edu, according to OpenAI's residency page, while Enterprise itself has no public price.¹⁴ Eligible new Enterprise agreements can use token-based billing, meaning usage is metered in tokens, the small chunks of text a model reads and writes.⁹ Usage across chat, ChatGPT Work and Codex is charged in US dollars on top of any seat fees and billed after each month.⁹

The agreement can also hold a spending commitment across ChatGPT Enterprise and the API, and any shortfall is billed under the contract, so the commitment works as a floor.⁹

Put the commitment next to what the same people cost on Business, where 8 Standard seats billed monthly come to $2,400 a year, about €2,105.⁴ The gap to the Enterprise quote buys residency and the other Enterprise controls, and only OpenAI's sales team knows its size.

Is EU data residency enough for the GDPR and your contracts?

Residency answers one line of a data protection assessment, namely where content rests and where the model runs, and the rest still needs paperwork.⁶ Customers in the EEA or Switzerland sign OpenAI's Data Processing Addendum, meaning the processor contract that Article 28 of the GDPR requires, with OpenAI Ireland Ltd.¹⁰¹¹ OpenAI Ireland passes data to affiliates outside Europe under Standard Contractual Clauses, meaning the Commission's model contract for transfers.¹⁰

The Data Privacy Framework, meaning the scheme under which US companies certify for EU transfers, listed no OpenAI entry when we searched it on 6 November 2026.¹² Transfers to the US therefore rest on those clauses plus a transfer impact assessment, meaning a written check that US law does not undercut them.⁶

A tender or a customer audit usually asks three things, namely which data goes into the tool, where it is stored and where it is processed. With EU storage and inference residency in place, the answer fits in one paragraph:

"Prompts, files and responses in our ChatGPT Enterprise workspace are stored at rest in the EEA and Switzerland, and model inference on them runs on GPUs in the same region. Logins, billing data and workspace metadata may be processed outside that region, and connected apps follow their own providers' terms."¹

The first sentence of that answer does not hold on ChatGPT Business, because Business fixes where content rests but not where the model runs.³ Its US copy for abuse monitoring would also need its own line.

Which route fits your leadership team?

Three routes cover what we see in leadership teams, and the wording of the customer contract picks between them.

1. Business with Europe as the storage region. It fits a contract that names storage only, at $25 per Standard seat billed monthly, about €22.³⁴
2. Business with a redaction filter. A redaction filter, meaning software that swaps names and customer details for placeholders before text reaches ChatGPT, works with the workspace you already have, but anything staff type straight into the chat window bypasses it.
3. Enterprise with EU storage and inference residency. It fits a contract that names processing in the EEA, in a new workspace with a spending commitment.¹⁹

Pick ChatGPT Enterprise with EU storage and inference residency when a customer contract requires processing inside the EEA. Keep ChatGPT Business at €22 per Standard seat, with Europe as the storage region, when the contract names storage alone and your assessment accepts the US copy. A redaction filter adds a guard to either route and never replaces a written rule on what staff may paste in.

When does Claude fit better?

Claude for Work, meaning Anthropic's Team and Enterprise plans, has no EU setting, and Anthropic says data from its commercial products is stored in the US.¹³ EU processing for Claude runs through EU cross-region inference on Amazon Bedrock or the EU multi-region endpoint on Google Cloud, which are routes for developers rather than a chat window for staff.¹⁴¹⁵ A leadership team that wants a chat app with inference in Europe today gets it from ChatGPT Enterprise, and that point goes to OpenAI.

Claude fits better when the EU work is one workflow rather than every chat, such as a contract review tool on Bedrock with the EU profile.¹⁴ Our Claude EU data residency guide covers each route model by model.

Checklist: four steps before the workspace opens

• Step 1: Copy the residency clause from your strictest customer contract and mark whether it names storage, processing or both.
• Step 2: Ask OpenAI's sales team in writing whether you qualify for inference residency in Europe and what can move from your current workspace.
• Step 3: Decide which apps, MCP servers and web search stay on.¹
• Step 4: After provisioning, open Workspace settings > General and read the Data Residency Region and Inference Residency Region fields.¹

What to do next

satori-chatgpt runs Business workspaces at €490 setup, then €14 per user per month plus the licence at €22 Standard or €110 Premium, all excl. VAT. Talk to us on a free 30-minute call for a second reader on an Enterprise quote or a residency clause.

Open the customer contract with the strictest data clause and copy its wording into a document next to the four exclusions listed under what can still leave the region. Fifteen minutes later you hold a two-column table, and every exclusion the clause does not tolerate becomes a question for OpenAI.

---

Sources

Footnotes

1. OpenAI Help Center (2026). Data residency and inference residency for ChatGPT, checked 6 and 9 November 2026. https://help.openai.com/en/articles/9903489-data-residency-and-inference-residency-for-chatgpt ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³ ↩¹⁴ ↩¹⁵ ↩¹⁶ ↩¹⁷ ↩¹⁸ ↩¹⁹ ↩²⁰

2. OpenAI (2025). Introducing data residency in Europe, 5 February 2025, updated 16 January 2026. https://openai.com/index/introducing-data-residency-in-europe/ ↩

3. OpenAI Help Center (2026). Where your ChatGPT Business content is stored, checked 6 and 9 November 2026. https://help.openai.com/en/articles/20001418-where-your-chatgpt-business-content-is-stored ↩ ↩² ↩³ ↩⁴ ↩⁵

4. OpenAI (2026). Business Pricing: compare features across plans, checked 6 November 2026. https://openai.com/business/pricing/ ↩ ↩² ↩³ ↩⁴

5. OpenAI (2026). Enterprise privacy at OpenAI, updated 8 January 2026. https://openai.com/enterprise-privacy/ ↩ ↩²

6. European Data Protection Board (2021). Recommendations 01/2020 on measures that supplement transfer tools, version 2.0. https://www.edpb.europa.eu/system/files/2021-06/edpb_recommendations_202001vo.2.0_supplementarymeasurestransferstools_en.pdf ↩ ↩² ↩³

7. OpenAI (2026). Data controls in the OpenAI platform, API documentation, checked 6 November 2026. https://developers.openai.com/api/docs/guides/your-data ↩ ↩² ↩³ ↩⁴ ↩⁵

8. OpenAI Help Center (2026). Exporting your ChatGPT history and data, checked 6 November 2026. https://help.openai.com/en/articles/7260999-exporting-your-chatgpt-history-and-data ↩

9. OpenAI Help Center (2026). Token-based billing for ChatGPT Enterprise, checked 6 November 2026. https://help.openai.com/en/articles/20001520-token-based-billing-for-chatgpt-enterprise ↩ ↩² ↩³ ↩⁴

10. OpenAI (2025). OpenAI Data Processing Addendum, updated 1 December 2025, effective 1 January 2026. https://openai.com/policies/data-processing-addendum/ ↩ ↩²

11. European Union (2016). Regulation (EU) 2016/679 (General Data Protection Regulation), Articles 28 and 44 to 49. EUR-Lex. https://eur-lex.europa.eu/eli/reg/2016/679/oj ↩

12. International Trade Administration, US Department of Commerce (2026). Data Privacy Framework List, searched for OpenAI and Microsoft on 6 November 2026. https://www.dataprivacyframework.gov/list ↩

13. Anthropic (2026). Where are your servers located? Do you host your models on EU servers? Anthropic Privacy Center, updated 15 June 2026. https://privacy.claude.com/en/articles/7996890-where-are-your-servers-located-do-you-host-your-models-on-eu-servers ↩

14. Amazon Web Services (2026). Route model inference requests across AWS Regions with cross-Region inference. Amazon Bedrock User Guide. https://docs.aws.amazon.com/bedrock/latest/userguide/cross-region-inference.html ↩ ↩²

15. Google Cloud (2026). Data residency. Gemini Enterprise Agent Platform documentation. https://docs.cloud.google.com/gemini-enterprise-agent-platform/resources/data-residency ↩