OpenAI Opens ChatGPT Enterprise Sales Through AWS Marketplace

By Rebecca Stone

OpenAI announced on November 9, 2026 that ChatGPT Enterprise is available through AWS Marketplace, adding a purchasing and billing channel for organizations that already buy software through AWS. Eligible customers can apply the spend toward AWS commitments, placing OpenAI’s workplace AI suite inside an existing cloud procurement relationship.

The AWS Marketplace listing names OpenAI as the seller and classifies ChatGPT Enterprise as software as a service deployed on AWS. It does not publish a standard rate. Instead, the listing says pricing is customized through private offers based on customer requirements and eligibility. AWS handles purchasing and billing, while OpenAI continues to manage the product experience.

Under the listing, the commercial relationship and the service relationship remain separate. AWS handles purchasing and billing, while OpenAI remains responsible for the product experience. The marketplace route therefore changes the buying process, not the feature set that OpenAI says eligible customers receive.

Matt Yanchyshyn, vice president of AWS Marketplace and Partner Services, said customers can “buy it through their AWS account, govern it alongside the rest of their software spend, and move from evaluation to rollout faster.” His statement places the marketplace’s value in procurement speed and centralized oversight rather than in a separate edition of the product.

OpenAI said AWS customers receive the same models, features and functionality available directly from the company, with new features released at the same time. Eligible customers can also access ChatGPT for Healthcare and ChatGPT for Financial Services through AWS Marketplace. OpenAI or AWS account teams confirm eligibility for those industry offerings.

The announcement does not provide a universal qualification test for the healthcare and financial-services editions. It directs customers to OpenAI or AWS account teams instead. That makes eligibility part of the sales process for those editions, even though the announcement makes the general ChatGPT Enterprise bundle available through AWS Marketplace.

What does the Enterprise bundle contain?

The enterprise bundle combines four products: Chat, ChatGPT Work, Codex and dots. OpenAI describes each component this way:

• Chat supports conversations for exploring ideas, writing and working through problems.
• ChatGPT Work delegates multi-step research, analysis and projects. It uses connected apps and files to create documents, presentations, spreadsheets and reports.
• Codex helps developers understand codebases, build features, fix bugs, run tests and review changes.
• Always-on dots track updates, keep documents and analyses current, and bring forward results and decisions requiring attention between conversations.

ChatGPT Work also gives users controls to review progress, steer the work and approve important actions. OpenAI describes dots as always on, tracking updates and current documents between conversations. Together, those controls let an enterprise delegate multi-step work while retaining the user’s approval step for important actions.

The four components divide the bundle into distinct modes of use. Chat covers conversation and writing. ChatGPT Work handles delegated research and document production. Codex addresses codebases, features, bugs and tests. Dots provides persistent monitoring between conversations. OpenAI presents the bundle as one enterprise purchase that joins those functions without merging their separate controls.

What security controls does the listing describe?

OpenAI’s vendor-supplied listing says business data is encrypted at rest and in transit and is not used to train OpenAI models by default. Administrators can manage user access, permissions and connected tools. Users separately control what agents can access and which actions require approval.

The listing separates account administration from delegated-agent permissions. Enterprise administrators set user access, permissions and connected-tool policies; individual users decide which agents can reach information and which actions need approval. Encryption, residency and the default exclusion from model training address different parts of the data lifecycle, according to the vendor-supplied terms.

Does buying through the marketplace keep data in AWS?

No. The listing explicitly says the AWS deployment label does not guarantee that all customer data remains on AWS infrastructure or within the customer’s AWS environment. OpenAI is the data processor and engages subprocessors, including AWS, to support the service. Marketplace billing therefore does not, on its own, settle an enterprise’s data-location requirements.

The distinction limits what “deployed on AWS” can tell a procurement team. A customer can consolidate the purchase and its billing while still evaluating where content is stored, where GPUs perform inference and which subprocessors support the service. For organizations with workload-specific governance rules, those architecture questions remain separate from price and billing.

OpenAI distinguishes two residency controls. Data residency controls where supported customer content is stored at rest. Inference residency separately controls where model inference on that content runs on GPUs. The company says both options apply only when enabled, are not available in all regions and do not cover all processing or systems data.

The company separates the purchasing path from the processing path. In practical terms, the marketplace purchase is closer to adding another line item to an enterprise software order than to moving all customer processing into a customer’s AWS account. OpenAI remains the processor, AWS is one listed subprocessor, and the residency settings determine supported placements where enabled.

How does ChatGPT Enterprise differ from Amazon Bedrock?

OpenAI positions ChatGPT Enterprise as complementary to its models in Amazon Bedrock. Developers can build AI applications and agents with those models in Bedrock, while business teams use ChatGPT Enterprise in daily work. The company says organizations can select an approach according to their workloads and governance requirements, creating separate routes for custom development and packaged workplace AI.

That framing separates infrastructure-led development from packaged workplace adoption. Bedrock gives developers models for custom applications and agents. ChatGPT Enterprise gives teams an integrated set of workplace, delegation and coding tools. OpenAI says customers can use one route or combine them, rather than treating access to a model and access to a workplace application as the same procurement decision.

The marketplace launch follows an April 28, 2026 announcement in which OpenAI and AWS introduced limited previews of OpenAI models on AWS, Codex on AWS and Amazon Bedrock Managed Agents powered by OpenAI. Those earlier offerings focused on building with OpenAI capabilities in AWS environments. The new announcement adds a procurement channel for the ChatGPT Enterprise bundle.

The April releases addressed builders. The November 9, 2026 listing addresses buyers of packaged workplace software. OpenAI and AWS frame the two announcements as complementary. The marketplace adds a purchasing route around ChatGPT Enterprise, while the April 28 previews remain tied to model-building services and managed agents.

For AWS-centric enterprises, OpenAI and AWS frame the launch as a choice shaped by workload, governance and residency requirements. Organizations can select ChatGPT Enterprise, Amazon Bedrock or a combination. AWS can consolidate the ChatGPT Enterprise purchase with other software spending, while the product choice still depends on whether the organization needs packaged workplace tools, custom model-based development, or both.

Under the companies’ description, account teams can confirm eligibility and manage the purchase, while enterprise technology leaders still evaluate data residency, inference placement and model-training terms. Buyers can consolidate the commercial transaction, but they still face a workload-specific choice about where data is stored and where inference runs.