ChatGPT Enterprise comes to AWS Marketplace, with residency limits
ChatGPT Enterprise became available through AWS Marketplace on November 9, 2026. Organizations can use their existing AWS purchasing relationship, with eligible spending applied to cloud commitments. The practical benefit is a new procurement route. Buyers still need to check where their information is stored and processed before approving a rollout.
What changes for enterprise buyers
OpenAI says the offering has the same models, features and release timing as purchasing directly. It combines ChatGPT, ChatGPT Work, Codex and dots. Eligible customers can also obtain the Healthcare and Financial Services offerings through the marketplace, with eligibility confirmed through their OpenAI or AWS account team.
OpenAI distinguishes this workplace product from its models in Amazon Bedrock. The latter support developers building applications and agents. A team evaluating employee access should therefore identify the product it wants before comparing cloud purchasing options.
Pricing comes through a private offer
The AWS listing names OpenAI as the seller and offers a custom quote based on requirements and eligibility. It does not publish a universal seat price. Eligible consumption can count toward an AWS Enterprise Discount Program or Private Pricing Agreement commitment, subject to the customer’s agreement.
Procurement teams should request an explicit breakdown of covered products, seats, usage charges, commitment eligibility and renewal terms. Compare that offer with the existing contract before assuming a marketplace purchase is cheaper. Consolidating an invoice and reducing total cost are separate outcomes.
An AWS bill does not establish AWS-only hosting
The listing says OpenAI manages the product experience and acts as the data processor, using subprocessors that include AWS. It expressly warns that buying through AWS Marketplace does not guarantee every piece of customer data stays on AWS infrastructure or inside the buyer’s AWS environment.
That distinction should be part of the approval conversation, especially when a purchasing team and a security team use the same cloud provider name to mean different things. ByteForward’s guide to Claude’s managed Google Cloud routes examines a related separation between the billing relationship and the service operator.
Inference residency is available only in the US for AWS purchases
OpenAI’s current residency documentation allows AWS Marketplace workspaces to store supported customer content in its listed data residency regions. For these workspaces, however, inference residency is available only in the United States.
Storage residency covers supported content at rest. Inference residency concerns GPU execution on that content. The broader documentation lists Europe and the UAE as additional inference regions, but the AWS purchasing restriction narrows that list for these customers.
Neither setting confines every operation to one location. External integrations, some non-GPU processing and system information have separate treatment. Feature coverage also varies. Administrators can inspect their configured storage and inference regions in Workspace settings under General, but those displayed fields do not change the configuration.
Check the agreement and the workspace together
Before purchasing, put the commercial offer beside the deployment requirements. Confirm the required storage location, whether in-region GPU processing is required, which features employees need and which external connections they will use. Ask the account team to resolve any mismatch before expanding access.
For an organization already buying software through AWS, this route may simplify the purchasing process. Its suitability still depends on the exact agreement and configured service boundaries. This report reviews published documentation, not an independently tested enterprise deployment.
