ChatGPT Enterprise is now on AWS Marketplace. It changes who sends the invoice, not where your data goes
If your company has an AWS spending commitment, ChatGPT Enterprise can now count toward it. Finance will like that. Read the data-processing line on the AWS listing before anyone tells the security team it's "on AWS now": OpenAI is the processor, and buying through the marketplace doesn't put the product inside your AWS environment.

Since November 9, 2026, ChatGPT Enterprise can be bought through AWS Marketplace, billed by AWS and, where your agreement allows, counted toward AWS EDP or PPA commitments, with the same product and features as buying from OpenAI. Pricing is by private offer. OpenAI remains the data processor, and AWS is one of the subprocessors it uses, so a marketplace purchase does not put the product inside your AWS environment. Treat it as a procurement change and run the usual data and admin review separately.
The announcement
On November 9, OpenAI made ChatGPT Enterprise available through AWS Marketplace. It's sold by OpenAI and billed through AWS. OpenAI says AWS customers get "the same models, features, and functionality available directly from OpenAI, with new features released at the same time." The package includes ChatGPT, ChatGPT Work, Codex and dots, OpenAI's always-on agents.
The selling point is procurement. A company that already buys through AWS can add ChatGPT Enterprise without setting up a new vendor, and eligible spend can count toward its AWS commitments. Sophos's CTO, quoted in the launch, put it plainly: buying through one channel and applying existing AWS commitments "removes friction that often slows enterprise AI adoption."
For many mid-size companies, that's genuinely useful. Vendor onboarding can take longer than the decision itself.
It also changes who tends to drive the purchase. When buying through AWS is easy, the request often starts with whoever manages the AWS account, usually engineering or IT, rather than with the teams who'll use the assistant every day. That's not a problem in itself, but it means the people who know how staff will actually use it, and which data they'll paste into it, should be in the room before the private offer is signed.
What it changes
Who bills you. Charges come through AWS. If your company has an Enterprise Discount Program or Private Pricing Agreement with AWS, the listing says eligible spend counts toward it, "subject to your agreement." That last phrase matters. Ask your AWS account team, in writing, whether this purchase counts under your particular terms.
How you get a price. There's no public price on the listing. Pricing is "based on your specific requirements and eligibility," and you request a private offer for a custom quote. Get a direct quote from OpenAI as well, so you know what the convenience costs, if anything.
Who you talk to for purchasing. AWS handles the marketplace side. Technical support still comes from OpenAI, through a marketplace support address on the listing.
Decision guide
What moves to AWS, and what stays with OpenAI
| Consideration | Moves to AWS | Stays with OpenAI |
|---|---|---|
| Billing | Charges appear on your AWS bill | The product, features and release timing are OpenAI's |
| Commitments | Eligible spend can count toward AWS EDP or PPA, subject to your agreement | Price set by a private offer, not a public rate |
| Data | AWS is one of OpenAI's subprocessors | OpenAI is the data processor; no guarantee data stays on AWS |
| Support | AWS handles marketplace purchasing | Technical support from OpenAI's marketplace support address |
Buying through AWS changes the paperwork. It doesn't change the product, the processor or the settings.
What it doesn't change
Buying through AWS changes the paperwork. It doesn't change the product, the processor or the settings.
Where your data goes. This is the part to read twice. The listing names OpenAI as the data processor for ChatGPT Enterprise, and AWS as one of the subprocessors it uses. A marketplace purchase therefore does not put the product inside your AWS environment, or under the data controls you already apply there. If someone in your organization hears "we bought it on AWS" and assumes those controls now cover ChatGPT, correct that early.
Residency. OpenAI offers data residency, which controls where supported content is stored at rest, and inference residency, which controls where the model runs on that content. The listing notes both apply only when enabled and aren't available in all regions. That's the same as buying direct. Check them against your requirements.
Admin decisions. Administrators still manage user access, permissions and connected tools, and users control what agents can reach and which actions need approval. With dots and ChatGPT Work in the bundle, those settings matter more than they did for a chat tool. Our OpenAI Dots review covers what always-on agents can do once connected.
The contract basics. Exit terms, data export and what happens at renewal are as important through a marketplace as they are direct. Our note on what belongs in an ongoing software partnership contract applies here too.
ChatGPT Enterprise, Bedrock, or both?
OpenAI positions the two as complementary. ChatGPT Enterprise is the product your staff use every day. OpenAI models in Amazon Bedrock are for developers building your own applications and agents inside AWS. If your goal is to put an AI assistant in front of employees, ChatGPT Enterprise is the relevant purchase. If your goal is an AI feature inside your own product or internal system, Bedrock, or a direct API, is where to look, and that's a build decision with its own data questions.
You'll also see other assistants on the same marketplace. The AWS listing page shows Anthropic's Claude Enterprise alongside, with spend that likewise draws down AWS commitments. If you're choosing between assistants, the marketplace makes buying either easier. It doesn't make the choice for you.
Sequence of work
A sensible order for the purchase
- Week 1
Finance confirms whether this spend counts under your specific AWS agreement.
- Week 1
Security reviews OpenAI's data processing terms and residency options, as for a direct purchase.
- Week 2
IT decides admin settings: who gets access, which connected tools, which agent actions need approval.
- Week 3
Request the private offer and compare it with a direct quote before signing.
Questions to put in the private offer request
Because the price comes from a private offer, the request is your chance to get answers in writing. Five worth asking:
- Does this purchase count toward our specific AWS commitment, and at what rate?
- Is the price the same as a direct purchase from OpenAI for the same seats and term? If not, what's the difference for?
- Which data residency and inference residency options apply to our region, and are they included?
- What happens at renewal: does the term, price or billing route change, and can we move to direct billing later?
- Who do we call for an outage, and who for a billing dispute?
Ask for the answers in the offer itself or in an email from the account team, not just on a call. Keep the answers with the contract. A year from now, nobody will remember what was said on the sales call, and the person renewing may not be the person who bought.
A short order of work
Finance checks the commitment question with AWS. Security reviews OpenAI's data processing terms and residency options as it would for a direct purchase. IT decides the admin settings, especially for agents. Then request the private offer and compare it with a direct quote. None of these steps is new, which is the point: the marketplace removes a vendor onboarding step, not the review.
One last practical note. Decide who owns the assistant after purchase: who approves new connectors, who reviews agent permissions every quarter, and who answers staff questions about what's allowed. A tool bought in a week can sit unmanaged for a year if nobody's name is on it.
If you're deciding where an AI assistant ends and a custom AI application should begin, that's our AI applications and agents work. Talk to us.
Related work: Sutton|Place
We build alongside the people doing the work.
Bring us the problem you’re thinking about.