ChatGPT Enterprise on AWS Marketplace: How to Buy ChatGPT Through Your AWS Bill — and When It Pays Off

ChatGPT Enterprise on AWS Marketplace: How to Buy ChatGPT Through Your AWS Bill — and When It Pays Off

On 9 November 2026, OpenAI put ChatGPT Enterprise on AWS Marketplace, giving US companies a new way to pay for workplace AI: through the Amazon Web Services bill they already receive every month. The change is not a new model, a new feature set or a cheaper version of ChatGPT. It is a different route to the same product — and for the right buyer, that route is worth real money, because eligible spending can count toward an existing AWS commitment. For the wrong buyer, it changes almost nothing.

This guide explains what the ChatGPT Enterprise AWS Marketplace listing actually includes, how the billing and commitment rules work, what it costs (and why OpenAI will not publish a price), the residency catch that matters for regulated teams, and how it compares with buying direct from OpenAI and with Anthropic’s rival listing.

Table of Contents

ChatGPT Enterprise AWS Marketplace — business team reviewing ChatGPT Enterprise on a laptop in a modern office
A team weighing a ChatGPT Enterprise purchase — the AWS Marketplace route is a procurement decision first and a product decision second.

What Is ChatGPT Enterprise AWS Marketplace?

ChatGPT Enterprise AWS Marketplace is the name for the ChatGPT Enterprise AWS Marketplace listing that lets an organisation buy ChatGPT Enterprise — OpenAI’s workplace version of ChatGPT — through Amazon’s software storefront instead of through a direct contract with OpenAI. The listing went live on 9 November 2026, and OpenAI describes it as available immediately, not as a preview or a planned launch.

An organisation that already buys software through AWS can now discover, purchase and manage ChatGPT Enterprise inside the same purchasing relationship, with AWS handling the invoice and OpenAI continuing to run the product itself.

That last distinction is the single most important thing to understand about the ChatGPT Enterprise AWS Marketplace offer. Buying through AWS does not move ChatGPT onto Amazon’s infrastructure, and it does not turn ChatGPT Enterprise into an AWS service. OpenAI remains the seller and the data processor, and OpenAI manages the product experience — the models, the features, the admin controls and the release schedule. AWS provides the storefront, the billing and the procurement channel.

The listing itself says so plainly, and reporting on OpenAI’s documentation adds an explicit warning: buying through AWS Marketplace does not guarantee that all customer data stays on AWS infrastructure or inside the buyer’s own AWS environment. Anyone who assumes otherwise is buying a different product from the one on offer.

So why does the channel matter at all, if the product is identical? Because enterprise software is rarely bought on product merits alone. Large US companies typically hold a multi-year AWS spending commitment — an Enterprise Discount Program (EDP) agreement or a Private Pricing Agreement (PPA) — that obliges them to spend a set amount with AWS over the term. Money spent on eligible AWS Marketplace purchases can count toward that obligation, subject to the customer’s agreement with AWS.

For a company that already owes AWS a large sum, routing a ChatGPT Enterprise purchase through the ChatGPT Enterprise AWS Marketplace listing can turn a bill it was going to pay anyway into progress against a commitment it has to meet. That is the whole financial case, and it is a real one — but it is also narrower than the headlines suggest, as the billing section below explains in detail.

The listing also sits inside a much bigger shift. OpenAI now reaches more than 1.2 billion people using ChatGPT each week, according to its own announcement, and the enterprise version bundles the company’s full workplace stack into one seat. At the same time, Amazon’s cloud store has become a genuine channel for frontier AI vendors: Anthropic already sells Claude Enterprise there, and Cognition’s Devin is purchasable through the same store.

The ChatGPT Enterprise AWS Marketplace listing is therefore not an experiment. It is OpenAI joining a procurement pattern that its competitors have already adopted — and it is a signal that enterprise AI is now being bought the way companies have bought databases and security software for years: through the cloud bill.

What Actually Changed on 9 November 2026

Before 9 November 2026, a US company that wanted ChatGPT Enterprise had one realistic route: negotiate a contract directly with OpenAI’s sales team, agree a per-seat price, and pay OpenAI. That route still exists, and OpenAI has not said it will go away.

What changed is that a second route opened alongside it. The ChatGPT Enterprise AWS Marketplace listing gives joint OpenAI–AWS customers a way to buy the same workplace product using the purchasing process, vendor approvals and billing systems they already have in place for AWS.

Three things are worth stressing about the announcement, because each one corrects a misunderstanding that has already spread. First, nothing about the product changed. OpenAI says in its announcement that Marketplace customers receive the same models, features and functionality as customers buying directly, with new features released at the same time. There is no delayed tier, no cut-down edition and no AWS-only variant. If a feature ships to direct customers on a Tuesday, it ships to Marketplace customers on the same Tuesday.

Second, the ChatGPT Enterprise AWS Marketplace listing covers the whole enterprise experience, not a single chatbot seat. The bundle is ChatGPT, ChatGPT Work, Codex and dots — the four parts of the workplace product, described in detail in the next section. Third, the ChatGPT Enterprise AWS Marketplace listing is available now. OpenAI framed it as an immediate availability, which matters for procurement teams planning a purchase this quarter rather than next year.

OpenAI also drew a deliberate line in its announcement between two very different things that both involve OpenAI and Amazon. OpenAI models in Amazon Bedrock are for developers who want to build applications and agents on top of the models through Amazon’s managed AI service. ChatGPT Enterprise AWS Marketplace, by contrast, is for teams across the whole business who want to use ChatGPT itself in their daily work — writing, research, analysis, coding and ongoing agent-assisted tasks. Bedrock is a building block; the ChatGPT Enterprise AWS Marketplace listing is a finished workplace product.

An organisation can, in principle, use both: build a custom application on Bedrock and give its staff ChatGPT Enterprise seats bought through Marketplace. The two do not compete with each other, and confusing them is the easiest way to buy the wrong thing.

The announcement also extended the ChatGPT Enterprise AWS Marketplace route to two regulated-industry editions. Eligible customers can access ChatGPT for Healthcare and ChatGPT for Financial Services through AWS Marketplace as well. OpenAI has not published the qualification criteria for either edition, and eligibility has to be confirmed through an OpenAI or AWS account team.

For a hospital system or a bank that already runs on AWS, that extension could matter more than the headline listing itself — but it is a conversation with an account team, not a self-service purchase, and it should be treated as one.

What’s Inside the Bundle: ChatGPT, ChatGPT Work, Codex and Dots

The ChatGPT Enterprise AWS Marketplace listing sells one subscription that contains four products, and understanding what each one does is essential to judging whether the seat is worth its (unpublished) price. OpenAI groups them on the ChatGPT Enterprise AWS Marketplace listing as Chat, ChatGPT Work, Codex and dots, and each covers a different kind of work.

A buyer who only needs the first one is almost certainly overbuying; a buyer who needs all four is getting a genuinely broad workplace stack.

Chat is the familiar conversational assistant: ask a question, draft a document, explore an idea, work through a problem in conversation. It is the part of the product most employees already know from the free or personal versions of ChatGPT, and in the enterprise edition it comes with the administrative controls, security and privacy terms that a company needs before it can let staff use it on real work. Business data is encrypted at rest and in transit, and it is not used to train OpenAI’s models by default — a point OpenAI repeats across its enterprise materials and that its enterprise privacy statement, updated on 8 January 2026, supports.

Administrators can manage user access, permissions and connected tools, while individual users stay in control of what agents can access and which actions need approval.

ChatGPT Work is the multi-step layer. Instead of a single question and answer, a user delegates a whole project — research, analysis or a report — and the system works through it in stages, drawing on context from connected apps and files to produce documents, presentations, spreadsheets and reports. The user reviews progress as it goes, steers the work and approves the important actions.

This is the part of the bundle aimed at analysts, marketers, operations staff and anyone whose job is to turn scattered information into a finished deliverable. It is also the part that most clearly separates an enterprise seat from a personal subscription: the value is not a smarter answer, it is a supervised workflow that ends in a document a colleague can use.

ChatGPT Enterprise AWS Marketplace — developer using Codex coding agent on a laptop with code on screen
Codex, the coding agent in the bundle, works directly with a codebase — understanding it, building features, fixing bugs and running tests.

Codex is OpenAI’s coding agent, and for engineering-led organisations it may be the most valuable component of the ChatGPT Enterprise AWS Marketplace package. Codex understands a codebase, develops features, fixes bugs, runs tests and reviews changes — not as a code-completion suggestion inside an editor, but as an agent that can carry a task through several steps.

For a team that already pays separately for AI coding tools, the inclusion of Codex in the same enterprise seat changes the maths: one subscription may replace two or three vendor relationships, which is exactly the kind of consolidation that procurement teams like to see on a single AWS invoice.

Dots is the newest and least familiar part of the bundle. OpenAI describes dots as agents that are always on: they learn a user’s preferences, follow through on ongoing responsibilities, track updates, keep documents and analyses current, and move projects forward between conversations, surfacing the results and decisions that need human attention. In other words, where ChatGPT Work handles a project you hand it, dots handle the work that never really ends — monitoring, updating, chasing.

One practical detail matters here: reporting indicates that OpenAI starts billing for dots on 18 November 2026, just nine days after the Marketplace launch. A buyer negotiating a private offer should ask explicitly how dots usage is priced inside the Marketplace subscription and how that usage will appear on the AWS invoice, because the announcement itself does not explain the mapping between credit use and the bill.

Taken together, the four components explain why OpenAI calls the ChatGPT Enterprise AWS Marketplace listing a single enterprise offering rather than four products sold side by side.

The ChatGPT Enterprise AWS Marketplace seat is meant to cover conversation, project work, software development and ongoing agent responsibilities in one place. That breadth is also the reason the ChatGPT Enterprise AWS Marketplace listing resists a simple price comparison: there is no single per-seat figure to compare with a rival, because the value depends entirely on how many of the four components a given team will actually use.

How Billing Works: Private Offers and AWS Commitments

The billing mechanics of the ChatGPT Enterprise AWS Marketplace listing are where the story gets genuinely interesting — and where a buyer can most easily make an expensive mistake. There are two separate questions, and they must not be confused. The first is what you pay. The second is what your payment counts toward. The listing answers the second question in principle and leaves the first entirely to negotiation.

The private offer: no public price, no published terms

The ChatGPT Enterprise AWS Marketplace listing has no public price. Reporting on the ChatGPT Enterprise AWS Marketplace listing confirms that buyers must request a private offer — a price and licence that OpenAI creates for one AWS account alone, on terms negotiated for that organisation. AWS’s own buyer guide defines a private offer as custom pricing and licence terms offered to a specific buyer, and that is exactly what is on sale here.

The listing publishes no per-seat price, no seat minimum, no term length and no usage rates. There is no self-service checkout and no universal rate card to compare against.

That opacity is a deliberate commercial choice, and it is not unusual for enterprise software. It does, however, change how a buyer should approach the purchase. Without a published price, the only way to know whether the ChatGPT Enterprise AWS Marketplace route is good value is to obtain equivalent offers through both channels — the Marketplace private offer and a direct OpenAI quote — matched on the same covered products, access levels, usage allowances and contract period, and then compare them side by side.

A buyer who requests only one of the two has no basis for judging the other. The announcement publishes neither fixed discounts nor the exact rules for commitment drawdown, so any claim that the ChatGPT Enterprise AWS Marketplace route is automatically cheaper is simply unsupported by the published facts.

How EDP and PPA commitment drawdown works

The financial benefit that the ChatGPT Enterprise AWS Marketplace listing does offer is commitment drawdown. According to the ChatGPT Enterprise AWS Marketplace listing, eligible consumption counts toward an AWS Enterprise Discount Program (EDP) commitment or a Private Pricing Agreement (PPA), subject to the customer’s agreement with AWS. In plain terms: if your company has promised AWS a certain level of spending over a number of years, the money you spend on ChatGPT Enterprise through Marketplace may be recognised as part of that promised spending.

The key word is may. Whether a given purchase qualifies, and how much of it qualifies, is decided by the customer’s own agreement with AWS — not by OpenAI, and not by the ChatGPT Enterprise AWS Marketplace listing page.

That conditionality matters because the rules are not uniform. Third-party guides on AWS Marketplace drawdown commonly quote a cap of around 25% of the annual commitment for eligible Marketplace purchases, according to a CloudForecast guide from August 2026 cited in reporting on this listing — but that is a general guide figure, not a promise about this listing.

There is also a narrower rule worth checking: since 1 May 2025, only software-as-a-service products hosted entirely on AWS and carrying a specific badge have qualified for PPA drawdown, and the ChatGPT Enterprise listing’s status under that rule is not something the announcement clarifies. A procurement team should therefore ask its AWS account team two questions in writing before signing anything: how much of this specific purchase will count toward our EDP or PPA, and under which clause of our agreement? The answer, not the headline, is the financial case.

It is equally important to understand what drawdown does not do. Commitment drawdown and a lower price are different outcomes. Suppose a company has $100,000 remaining on an AWS commitment and AWS confirms that a $40,000 ChatGPT Enterprise purchase qualifies in full. The company would then have $60,000 left to fulfil, assuming nothing else changes. That is useful accounting — it retires an obligation.

It does not make the ChatGPT purchase free, and it does not prove the company saved $40,000. The money still leaves the company; it simply counts twice, once as a software purchase and once as commitment progress. The strongest financial case for the ChatGPT Enterprise AWS Marketplace route is therefore a company that genuinely wants ChatGPT Enterprise and genuinely risks leaving part of its AWS commitment unused. A company that already expects to meet its commitment has a weaker case: invoice consolidation may still help, but the quoted price deserves its own comparison against a direct OpenAI offer.

ChatGPT Enterprise AWS Marketplace — rows of servers in a cloud data center
Buying through AWS Marketplace is a billing route, not a hosting move — the service is not automatically deployed inside your own AWS environment.

The Price Question: Why There Is No Public Price

The absence of a public price on the ChatGPT Enterprise AWS Marketplace listing is the detail buyers ask about most, and it deserves a straight answer. OpenAI has not published a per-seat rate for the ChatGPT Enterprise AWS Marketplace listing, has not published a seat minimum, and has not published the term lengths available. The listing tells buyers to request a private offer for a quote, and the price will be, in the ChatGPT Enterprise AWS Marketplace listing’s own words, based on the buyer’s specific requirements and eligibility. That is the entire public pricing picture.

There are several plausible reasons for the opacity, and a buyer should understand them before entering the negotiation. First, ChatGPT Enterprise has always been sold as a negotiated enterprise product, and OpenAI has never published a standard per-seat price for it even through the direct channel. The Marketplace listing simply continues that practice in a new storefront.

Second, the bundle’s value varies enormously by customer: a 50-person team that uses only Chat and a 5,000-person company that runs Codex and dots across engineering and operations are buying very different amounts of value, and a single published price would mislead one of them. Third, the commitment-drawdown angle makes the effective cost depend on the buyer’s AWS agreement, which OpenAI cannot know in advance. A price that looks expensive on its face may be attractive to a company with a large unmet commitment, and vice versa.

None of that means a buyer should accept the first number offered. The practical advice is straightforward. Ask for the private offer in writing, with the seat count, the included components, the usage allowances for Codex and dots, the contract term and the renewal terms all stated explicitly.

Ask for the equivalent direct quote from OpenAI on the same terms. Ask the AWS account team, also in writing, exactly how much of the Marketplace purchase will count toward the EDP or PPA and whether any cap applies. And ask how dots usage will be billed once OpenAI’s usage billing for dots begins, since that is a cost line the announcement does not explain. A buyer who has those four answers in writing knows what they are buying. A buyer who has only a headline price does not.

One comparison is worth keeping in mind during that negotiation. Anthropic’s Claude Enterprise is also sold on AWS Marketplace, and its listing does show a public price: a 12-month contract for 20 seats at $4,800, according to reporting on the Claude Enterprise listing. That works out to $20 per seat per month for a small team, and it gives a buyer a concrete anchor for what an enterprise AI seat can cost through the same store.

It is not a like-for-like comparison — the products differ, the seat counts differ and the included capabilities differ — but it does show that public pricing on AWS Marketplace is possible, and that OpenAI’s choice not to publish one is a choice, not a limitation of the store. Use the Anthropic figure as a negotiating reference point, not as a price for ChatGPT Enterprise, and the conversation with OpenAI’s sales team becomes a much more grounded one.

ChatGPT Enterprise AWS Marketplace vs Buying Direct

For most buyers, the real decision is not whether to buy ChatGPT Enterprise — it is which of the two channels to buy it through. The ChatGPT Enterprise AWS Marketplace route and the direct OpenAI route sell the same product, so the comparison comes down to procurement, billing and control. The table below summarises the differences that the published sources actually establish.

Factor AWS Marketplace route Direct OpenAI route
Purchasing relationship Uses the organisation’s existing AWS purchasing channel and vendor approvals A separate contract negotiated directly with OpenAI
Billing OpenAI is the seller; AWS bills the purchase on the AWS invoice Billed by OpenAI, outside the AWS invoice
Models and features Same as direct, according to OpenAI The baseline for OpenAI’s parity promise
New feature timing Same release timing as direct customers Same release timing
Price Private offer; no public price published Negotiated; no public per-seat price published either
AWS commitment treatment Eligible spend may count toward EDP or PPA, subject to the AWS agreement The announcement makes no corresponding commitment-drawdown promise
Data and hosting OpenAI remains data processor; purchase does not guarantee data stays on AWS infrastructure Same OpenAI processing; no AWS billing relationship involved
Migration for existing customers No automatic transfer published; renewal and migration arrangements need confirmation Existing relationship continues unchanged

Three points in that comparison deserve emphasis. First, the product parity promise is explicit and unusually strong: same models, same features, same release timing. A buyer does not trade product quality for procurement convenience, and OpenAI has put that in writing. Second, the commitment treatment is the only structural advantage of the ChatGPT Enterprise AWS Marketplace route — everything else is convenience.

If your organisation has no AWS commitment, or has already met it, the ChatGPT Enterprise AWS Marketplace route offers little beyond a consolidated invoice.

Third, existing direct customers should not assume they can switch channels at renewal without a conversation. The announcement does not establish an automatic transfer for existing direct customers, so anyone already on a direct contract who wants to move to Marketplace billing needs to confirm the migration and renewal arrangements with both OpenAI and AWS before the renewal date arrives. That is a small administrative step, but it is the kind of step that causes a missed renewal if it is left until the last week.

There is also a control question that the table does not capture. When a purchase runs through AWS Marketplace, the buying organisation’s AWS administrators gain a degree of visibility and control over the subscription through the Marketplace console — who bought it, what was spent, when it renews. For a large company that already governs all of its software spend through AWS, that visibility is a genuine benefit: the AI spend sits next to the cloud spend, in the same reports, under the same approval workflows.

For a small company with a single AWS account and a finance team of one, it makes almost no difference. As with the rest of this listing, the value of the channel depends on the size and shape of the buyer.

ChatGPT Enterprise AWS Marketplace vs Claude Enterprise on AWS

No buyer evaluates a purchase in isolation, and the ChatGPT Enterprise AWS Marketplace listing has an obvious neighbour on the same storefront: Anthropic’s Claude Enterprise, which is also sold through AWS Marketplace. The two listings are the clearest head-to-head in enterprise AI procurement right now, and comparing them honestly is useful — with one important caveat. They are different products with different strengths, and a price comparison that ignores that fact will mislead more than it helps.

The most visible difference is transparency. Claude Enterprise publishes a public price on its AWS Marketplace listing — $4,800 for 20 seats over 12 months, according to reporting on the Claude Enterprise listing. That is $20 per seat per month, for a defined small-team package, and a buyer can see it before speaking to anyone.

The ChatGPT Enterprise AWS Marketplace listing publishes nothing comparable: private offer only, price on application. For a buyer who values a quick, transparent starting point, Anthropic’s approach is plainly easier to evaluate. For a buyer who expects to negotiate a large, multi-year deal anyway, the difference matters less — but it still tells you something about the two vendors’ sales cultures. Anthropic is willing to anchor the conversation with a number. OpenAI is not.

The product comparison is a different question, and it is one this article cannot settle with a scorecard, because the two products are built around different strengths. ChatGPT Enterprise bundles four components — Chat, ChatGPT Work, Codex and dots — into a single seat, and its headline advantage is breadth: one subscription that covers conversation, project work, coding and always-on agents.

Claude’s enterprise offering is built around the Claude models and the Claude Code agent workflow, and its headline advantage is depth in long, careful, multi-step work — the kind of work Anthropic has spent the last year optimising for, including its newer agent orchestration features. A team whose main need is coding assistance will weigh Codex against Claude Code. A team whose main need is research and document production will weigh ChatGPT Work against Claude’s project and document capabilities. A team that needs both will need to trial both, because no published benchmark settles the question for a specific organisation’s workload.

There is also a strategic point worth noting. The fact that both vendors now sell through AWS Marketplace means the store has become a genuine marketplace for enterprise AI, not a single-vendor shelf. That is good news for buyers: it creates a credible alternative in every negotiation, and it gives procurement teams a standard channel for comparing offers.

A buyer who obtains a private offer from OpenAI and a public price from Anthropic, for the same seat count and term, has a real negotiating position with both vendors. The ChatGPT Enterprise AWS Marketplace listing is therefore valuable even to a buyer who ultimately chooses Claude — because its existence makes the Claude price negotiable, and vice versa. Competition through a shared storefront is exactly how enterprise software markets are supposed to work.

The Residency Catch: US-Only Inference

The most consequential detail in the ChatGPT Enterprise AWS Marketplace listing is not about price. It is about where the AI actually runs. OpenAI’s documentation draws a distinction that every buyer in a regulated industry needs to understand, because the two halves of it are controlled separately and the ChatGPT Enterprise AWS Marketplace route restricts one of them in a way that surprises many buyers.

Data residency controls where supported customer content is stored at rest — the files, conversations and documents that sit on disk between sessions. Inference residency separately controls where model inference on that content runs on GPUs — the actual processing, the moment the model reads the data and produces an answer. These are different controls, they apply only when enabled, and they are not available in all regions. The distinction matters because a buyer can easily satisfy one and fail the other without realising it.

For workspaces bought through the ChatGPT Enterprise AWS Marketplace listing, published reporting on OpenAI’s residency documentation identifies a specific restriction: those workspaces can select inference residency only in the United States. Data residency, by contrast, can use the listed regions — reported examples include Europe, India and the UK — so a European buyer can store its content at rest in Europe.

But the processing itself, the GPU inference, is available only in the US for Marketplace workspaces. Reporting adds a further condition: inference residency must be enabled, and it requires data residency in the same region. The practical consequence is straightforward. For a buyer that requires both European storage and European model processing — which is exactly what many European banks, healthcare providers and public-sector bodies require — selecting European storage alone does not satisfy both requirements. The Marketplace route’s documented inference-residency option does not provide a European processing guarantee.

There are also boundaries around the residency controls that the reporting describes. Supported GPU execution is not every processing step, and external integrations, some non-GPU processing and system information have separate treatment. These findings come from published accounts of OpenAI’s documentation, not from an independently inspected enterprise deployment, and a buyer with strict requirements should verify them directly with OpenAI before signing.

But the headline is clear enough to act on: if your organisation’s compliance policy requires AI processing to happen inside a specific region outside the US, the ChatGPT Enterprise AWS Marketplace route may not meet it, and the direct route — or a different vendor — may be the safer choice.

That single fact could be decisive for a European or UK buyer, and it is the kind of detail that is easy to miss in the excitement of a commitment-drawdown saving.

ChatGPT Enterprise AWS Marketplace — two business people shaking hands over a signed agreement
A Marketplace purchase is a negotiated private offer, not a self-service checkout — get the price, the commitment treatment and the residency terms in writing.

Who Should Buy Through AWS — and Who Shouldn’t

The ChatGPT Enterprise AWS Marketplace listing is a good deal for some organisations and a neutral one for others. The difference is almost entirely about the buyer’s existing AWS relationship, not about the product. This section is deliberately blunt, because the wrong buyer can waste weeks on a procurement route that was never going to help them.

The route makes most sense for:

  • A large US company with an unmet AWS commitment. If your organisation owes AWS a substantial sum under an EDP or PPA and is at risk of leaving part of it unused, routing a ChatGPT Enterprise purchase through Marketplace converts a sunk obligation into useful software. This is the strongest case in the whole listing, and it is the buyer OpenAI and AWS had in mind.
  • A company that already governs all software spend through AWS. If every vendor approval, purchase order and invoice already flows through the AWS relationship, adding ChatGPT Enterprise to the same channel removes a whole layer of procurement friction — one vendor to approve, one invoice to process, one renewal to track.
  • A regulated US buyer that can accept US inference residency. A US healthcare or financial-services organisation that is eligible for the Healthcare or Financial Services editions, and that has no requirement for processing outside the US, can use the ChatGPT Enterprise AWS Marketplace route without hitting the residency catch described above.
  • A team that will actually use the whole bundle. If your staff will use Chat, ChatGPT Work, Codex and dots together, the single-seat bundle is genuinely broad value. The more components a team uses, the stronger the case for the bundled seat over separate tools.

The route makes least sense for:

  • A company with no AWS commitment, or one already met. Without an unmet commitment, the drawdown benefit disappears, and the ChatGPT Enterprise AWS Marketplace route offers little beyond invoice consolidation. A direct quote may be simpler to negotiate and easier to compare.
  • A European or UK buyer that needs local AI processing. The US-only inference residency for Marketplace workspaces is a hard limit. If compliance requires processing inside Europe or the UK, verify the direct route or another vendor before committing to Marketplace billing.
  • A small team that wants a quick, transparent price. The private-offer process is a negotiated enterprise sale. A team of twenty that wants to see a price and buy this week will find Anthropic’s published $4,800-for-20-seats listing on the same store far easier to evaluate — and may find that it fits better anyway.
  • An existing direct customer who assumes a painless switch. There is no published automatic transfer from a direct contract to Marketplace billing. The migration needs a conversation with both vendors, and it should be planned before the renewal date, not after it.

If you recognise your organisation in the first list, the ChatGPT Enterprise AWS Marketplace route is worth pursuing seriously. If you recognise it in the second, the ChatGPT Enterprise AWS Marketplace listing is still useful as a negotiating reference — but the purchase itself may be better made elsewhere.

How to Buy ChatGPT Enterprise Through AWS Marketplace

Buying through the ChatGPT Enterprise AWS Marketplace listing is a short process on paper and a negotiation in practice. The steps below follow the route OpenAI itself describes, with the checks that the published reporting suggests a careful buyer should add at each stage.

  1. Confirm the route fits. Before contacting anyone, check the two gating questions: does your organisation hold an AWS commitment with room left in it, and can you accept US-only inference residency? If the answer to the first is no and the second is a problem, the direct route is probably better, and you have saved yourself a sales cycle.
  2. Contact your OpenAI or AWS account team. OpenAI directs interested organisations to their OpenAI or AWS account team to confirm eligibility and get started. There is no self-service purchase. If you do not already have an account team, the AWS Marketplace listing page is the starting point for the request.
  3. Request the private offer. Ask for the offer in writing, with the seat count, the four included components, the usage allowances for Codex and dots, the contract term and the renewal terms all stated. Ask specifically how dots usage will be billed, since OpenAI’s usage billing for dots is reported to begin on 18 November 2026 and the announcement does not explain how it maps to the AWS invoice.
  4. Get the commitment treatment in writing from AWS. Ask your AWS account team exactly how much of the purchase will count toward your EDP or PPA, under which clause, and whether any cap applies. Do not rely on a verbal assurance or a general guide figure. The listing makes the treatment subject to your agreement, so the agreement is the document that matters.
  5. Verify the residency configuration. Confirm which data-residency regions are available to your workspace and confirm in writing that inference residency is US-only for Marketplace workspaces. If your compliance team requires processing elsewhere, raise it now, not after the contract is signed.
  6. Compare with the direct quote. Obtain an equivalent direct offer from OpenAI on the same seat count, components, usage allowances and term, and compare the two side by side. The Marketplace route should win on the strength of the commitment treatment and the procurement convenience — not on an assumed price advantage that nobody has published.
  7. Check the Healthcare or Financial Services editions if relevant. If you are a hospital system, insurer or bank, ask whether you are eligible for the regulated-industry editions through Marketplace, and what additional terms apply. Eligibility is confirmed by the account team, and the criteria have not been published.

The whole process is designed for organisations that already buy enterprise software through a sales team, and it will feel slow to anyone used to buying software with a credit card. That is the nature of the product.

The time is worth spending, because the two things that most commonly go wrong — a commitment that does not actually count, and a residency configuration that does not meet compliance — are both cheap to check before signing and expensive to fix afterwards.

ChatGPT Enterprise AWS Marketplace — professionals collaborating around laptops in a conference room
For a large team that already runs on AWS, buying ChatGPT Enterprise through the Marketplace can consolidate procurement — but only if the commitment math actually works.

What This Means for the Wider AI Market

The ChatGPT Enterprise AWS Marketplace listing is a small announcement with a large implication: enterprise AI is now being sold through the same channels as every other kind of enterprise software. That sounds unremarkable, but it marks a real change in how the AI industry works. For the first two years of the generative-AI boom, the leading labs sold directly, on their own terms, through their own sales teams. The product was new, the buyers were experimenting, and the channel did not matter much. Now the buyers are no longer experimenting.

They are standardising, consolidating and negotiating — and the vendors are meeting them where the procurement already happens.

That shift is visible across the industry, not just in this one listing. OpenAI is also building commerce directly into ChatGPT itself — the company recently launched ChatGPT Wallet, which lets users pay inside conversations — and it is opening business profiles and agents inside ChatGPT for companies that want to reach customers there. Anthropic, meanwhile, keeps shipping flagship models such as Claude Fable 5.2 while selling its enterprise edition through the same AWS store.

The pattern is consistent: the labs are moving from selling a chatbot to selling a workplace platform, and from selling it directly to selling it through the channels where enterprises already buy. AWS Marketplace is simply the most important of those channels, because so many large companies already route their software spend through it.

For AWS, the ChatGPT Enterprise AWS Marketplace listing is a win in a quieter competition. Amazon’s cloud store earns its position by being the place where the software a company needs is already available, and every frontier AI vendor that lists there makes the store harder to leave.

For OpenAI, the ChatGPT Enterprise AWS Marketplace listing is a way to reach buyers who would otherwise face a long internal procurement process for a new vendor — a process that, in a large company, can take months and can kill a deal that the users themselves want. And for buyers, the ChatGPT Enterprise AWS Marketplace listing is a small but real increase in leverage: when two frontier vendors sell through the same store, on comparable terms, the buyer’s negotiating position improves with both of them.

The longer-term question is whether the ChatGPT Enterprise AWS Marketplace route becomes the default way enterprises buy AI, or whether it stays a niche for companies with large AWS commitments. The answer will depend on two things that have not happened yet.

The first is pricing transparency: if OpenAI and its rivals keep hiding behind private offers, the store will remain a place for negotiated deals rather than a place for comparison shopping. The second is residency: if the US-only inference restriction for Marketplace workspaces persists, a large part of the European market will simply not be able to use the route, and the direct channel will remain essential. Both are fixable. Neither has been fixed yet. Until they are, the ChatGPT Enterprise AWS Marketplace listing is best understood as a useful new option for a specific kind of buyer — not as the new default for everyone.

Limitations and Risks to Know Before You Sign

A fair guide should end the sales pitch where the sales pitch ends, and the ChatGPT Enterprise AWS Marketplace listing has several limitations that a buyer should weigh alongside its benefits. None of them is a reason to reject the ChatGPT Enterprise AWS Marketplace listing outright. All of them are reasons to ask questions before signing.

The price is unknown until you negotiate. There is no public price, no seat minimum and no published term. A buyer who values transparency, or who needs a budget figure before an internal approval, will find the process slower than a listing with a published rate. The Anthropic comparison in this guide exists precisely because a buyer needs some anchor — and an anchor from a rival is a poor substitute for a price from the vendor itself.

The commitment benefit is conditional. Eligible spend may count toward an EDP or PPA, subject to the customer’s agreement. The word “may” carries the whole risk: a buyer who assumes the full purchase will count, and then discovers that only part of it does — or that a cap applies, or that the ChatGPT Enterprise AWS Marketplace listing does not qualify under the post-May-2025 PPA rules — has made a budgeting error that only becomes visible after the contract is signed.

Get the treatment in writing, from AWS, before you sign.

Inference residency is US-only for Marketplace workspaces. This is the single most important limitation for non-US buyers, and it is described in detail in the residency section above. It is a documented restriction, not a rumour, and it should be treated as a hard constraint until OpenAI says otherwise.

Your data is not automatically on AWS. Buying through AWS Marketplace does not guarantee that all customer data stays on AWS infrastructure or inside the buyer’s AWS environment. OpenAI is the data processor and engages subprocessors including AWS, but the service is OpenAI’s, and the data-processing arrangements are OpenAI’s. A buyer whose security review assumes AWS hosting will fail that review — and will have to redo it.

Dots billing is about to start, and the invoice mapping is unexplained. Reporting indicates that OpenAI starts billing for dots on 18 November 2026. The announcement does not explain how dots credit use maps to the AWS invoice. A buyer signing a Marketplace agreement in November should ask for that mapping explicitly, in writing, rather than discovering it on the first invoice.

Existing direct customers have no published migration path. If you already buy ChatGPT Enterprise directly and want to move to Marketplace billing, the announcement does not describe an automatic transfer. Plan the migration as a project with both vendors, and start it well before your renewal date.

The listing is a purchasing change, not a product change. This is worth repeating because it is the most common misunderstanding. The ChatGPT Enterprise AWS Marketplace listing adds no models, no features and no capabilities. If your team does not need ChatGPT Enterprise, the ChatGPT Enterprise AWS Marketplace route does not make it worth buying. If your team does need it, the route may make it easier to buy. That is the whole of it — and, for the right buyer, it is enough.

Frequently Asked Questions

Can I buy ChatGPT Enterprise at a public price on AWS Marketplace?
No. The listing has no public price, according to reporting on it. Buyers must request a private offer and confirm eligibility through an OpenAI or AWS account team. There is no published per-seat rate, seat minimum or term length.

Is the product the same as buying directly from OpenAI?
Yes, according to OpenAI. Marketplace customers receive the same models, features and functionality as direct customers, with new features released at the same time. The bundle is ChatGPT, ChatGPT Work, Codex and dots in both cases.

Does buying through AWS Marketplace move my data onto AWS?
No. OpenAI remains the data processor and engages subprocessors including AWS, but the ChatGPT Enterprise AWS Marketplace listing warns that buying through Marketplace does not guarantee all customer data stays on AWS infrastructure or inside your AWS environment. Business data is encrypted at rest and in transit and is not used to train models by default.

How much of my AWS commitment will a ChatGPT Enterprise purchase retire?
It depends on your own EDP or PPA agreement with AWS. The listing says eligible consumption counts toward the commitment, subject to your agreement. Third-party guides commonly quote a cap of around 25% of the annual commitment for eligible Marketplace purchases, but that is a general guide figure, not a promise about this listing. Confirm the exact treatment with your AWS account team in writing.

Can a European company use the ChatGPT Enterprise AWS Marketplace route?
It can buy through the route, but with a restriction: published reporting on OpenAI’s residency documentation says Marketplace workspaces can select inference residency only in the United States, even though data can be stored at rest in regions such as Europe, India and the UK. A buyer that needs AI processing inside Europe should verify the direct route or another vendor.

What is the difference between ChatGPT Enterprise on AWS Marketplace and OpenAI models on Amazon Bedrock?
Bedrock is for developers building applications and agents on top of OpenAI’s models. The Marketplace listing is the finished ChatGPT Enterprise workplace product — Chat, ChatGPT Work, Codex and dots — for teams using AI in their daily work. OpenAI says the two complement each other.

Are ChatGPT for Healthcare and ChatGPT for Financial Services available through Marketplace?
Yes, for eligible customers. OpenAI says eligible customers can access both editions through AWS Marketplace, but it has not published the qualification criteria. Eligibility must be confirmed through an OpenAI or AWS account team.

The Verdict

The ChatGPT Enterprise AWS Marketplace listing is a sensible, narrowly useful change that has been oversold in some coverage and undersold in other. It is not a new product, not a price cut and not a hosting move. It is a second checkout lane for the same enterprise product, and its value depends almost entirely on who is standing in it.

For a large US company with an unmet AWS commitment, a procurement team that already buys through AWS, and a workforce that will use Chat, ChatGPT Work, Codex and dots together, it is a genuinely good way to buy — it retires an obligation, consolidates the invoice and delivers the same product on the same schedule as a direct purchase. For a small team, a European buyer with residency requirements, or a company with no AWS commitment to retire, it changes very little, and the private-offer process may cost more time than the consolidated invoice saves.

The honest advice is therefore conditional, and it fits in one paragraph. If you hold an AWS commitment with room in it, ask for the private offer and the written commitment treatment, compare it with an equivalent direct quote, and confirm the residency configuration with your compliance team.

If those four answers line up, buy through Marketplace. If any one of them does not, buy direct — or buy from the rival whose published price you can already see. The listing’s real significance is not what it does for any single buyer.

It is that enterprise AI has now joined the ordinary machinery of enterprise software buying, alongside the other workplace agents companies are already deploying, from Agentforce in Microsoft Teams to the coding agents on every engineering roadmap. The product is no longer the novelty. The procurement is. And on that measure, OpenAI’s move onto AWS Marketplace is the most normal thing that has happened to enterprise AI all year — which is exactly why it matters.

Leave a Comment

Your email address will not be published. Required fields are marked *