OpenAI made ChatGPT Enterprise available through AWS Marketplace on 9 November 2026, allowing organizations to purchase it through their existing AWS relationship and apply eligible spending toward AWS commitments. For universities and research offices using AWS, the development adds a procurement route, while the service’s data-processing arrangements and residency options remain separate questions.
A new purchasing channel for ChatGPT Enterprise
In its 9 November announcement, OpenAI said Marketplace customers receive the same models, features and functionality as customers buying directly, with new features released at the same time. The company described the listing as available immediately, rather than a preview or a planned launch.
Mission Critical Research reported on 10 November that OpenAI remains responsible for the product experience while purchasing and billing use the buyer’s AWS relationship. Its account identifies the seller as OpenAI and the delivery method as software as a service. That division matters for institutional purchasing: an AWS transaction still brings an OpenAI-managed service into the organization.
The commitment provision is conditional. The AWS Marketplace listing says eligible consumption counts toward an AWS EDP or PPA commitment, subject to the customer’s agreement. Availability in the catalogue therefore does not establish that every institution’s purchase will qualify. The listing provides a private-offer route, and OpenAI directs customers to their OpenAI or AWS account team to confirm eligibility.
What the offering includes
The seller’s Marketplace description groups four components into the enterprise offering. These are vendor descriptions of capabilities, rather than evidence of research performance:
| Component | Work described in the listing |
|---|---|
| Chat | Conversation, writing, questions and problem-solving. |
| ChatGPT Work | Multi-step research, analysis and projects using connected apps and files. |
| Codex | Understanding codebases, developing features, fixing bugs, running tests and reviewing changes. |
| Dots | Ongoing agent responsibilities, including tracking updates and keeping documents and analyses current. |
OpenAI places ChatGPT Enterprise alongside its models already available in Amazon Bedrock. Its announcement distinguishes developers building applications and agents in Bedrock from teams using ChatGPT Enterprise in their everyday work. For a research organization, those are different procurement objects: access to a finished workspace and model access for an application. The Marketplace announcement does not make the workspace a Bedrock deployment.
Storage location and inference location differ
The Marketplace listing names OpenAI as the data processor and identifies AWS among its subprocessors. It explicitly warns that buying through Marketplace does not guarantee that all customer data stays on AWS infrastructure or inside the customer’s AWS environment. Its “deployed on AWS” label therefore cannot be read as an assurance of confinement to an institution’s own cloud environment.
OpenAI’s residency documentation makes a further distinction. For ChatGPT Enterprise purchased through AWS Marketplace, supported customer content can be stored at rest in any of its listed data-residency regions, but inference residency is available only in the United States. A storage-region choice consequently does not establish that model execution occurs in the same region.
The documentation defines data residency around in-scope content and inference residency around GPU execution. It also identifies categories that may sit outside the chosen storage region, including external integrations, transient processing, workspace metadata, billing information and user logins. Those qualifications are material to research projects with geographic restrictions on data: “regional storage” describes a particular control, rather than every operation involving the workspace.
The launch reaction focuses on procurement
Aiholics’ 9 November coverage characterized the announcement’s executive statements as focused on buying rather than new capabilities. It reported that AWS Marketplace and Partner Services vice president Matt Yanchyshyn emphasized reducing additional procurement layers. OpenAI cloud-partnerships managing director Chris Grusz emphasized adoption through an existing cloud-provider relationship.
Aiholics also identified Sophos chief technology officer John Peterson’s statement as a supplied customer endorsement. Peterson linked the single purchasing channel and use of AWS commitments to less procurement friction. These comments express the launch participants’ expectations; they do not establish how much time a university will save or demonstrate an independently measured benefit.
Why the distinction matters for research offices
For institutions already buying through AWS, the immediate change concerns the commercial route available for an enterprise AI workspace. Its relevance to supplier contracting follows from the combination of AWS billing, agreement-dependent commitment eligibility and an OpenAI-managed product. The announcement establishes those arrangements without establishing the outcome of any particular university’s procurement process.
The data relationship also remains important. OpenAI’s enterprise privacy statement, updated 8 January 2026, says business data is not used for model training by default. It describes customer control over retention for Enterprise and Edu, connected internal sources, authentication and feature access. Those commitments concern the service and its administration, alongside the processor relationship addressed in a data processing agreement.
For research offices, these arrangements make the purchasing channel, permitted data uses and geographic controls distinct parts of the institutional decision. The development connects existing questions of AI governance and research data security to a new commercial route, especially when a workspace can draw context from institutional files and applications.
Availability is immediate; eligibility is customer-specific
With availability announced for 9 November, OpenAI’s stated route into the offering is confirmation of eligibility through an OpenAI or AWS account team. Research institutions now have an additional option within AWS purchasing arrangements, with commitment treatment and residency scope grounded in the relevant agreement and product documentation.
Sources
- OpenAI: ChatGPT Enterprise now available in AWS Marketplace. Published 9 November 2026; checked 2026-11-15.
- AWS Marketplace: ChatGPT Enterprise, sold by OpenAI. Live listing; checked 2026-11-15.
- Mission Critical Research: OpenAI offers ChatGPT Enterprise for purchase through AWS Marketplace. Published 10 November 2026; checked 2026-11-15.
- Aiholics: ChatGPT Enterprise can now be bought on AWS bills and counted against commitments. Published 9 November 2026; checked 2026-11-15.
- OpenAI Help Center: Data residency and inference residency for ChatGPT. Checked 2026-11-15.
- OpenAI: Enterprise privacy at OpenAI. Updated 8 January 2026; checked 2026-11-15.
