The ChatGPT-Ads Land Grab: Ad Networks Cut Minimums

The ChatGPT ads land grab moved into its land-rush phase in June 2026: StackAdapt dropped its ChatGPT ad spend minimum to $0 in May, and rival Criteo cut its own minimum from $50,000 to $10,000 in early June. Read against the pilot's $200,000–$250,000 entry commitment in February, that is a compression from roughly a quarter of a million dollars to nothing in under four months.

Vendor coverage frames this as access democratisation — and for a small advertiser who wants to test conversational AI inventory, it genuinely is. But the more useful frame for a marketing team is the one nobody is writing: this is a textbook ad-tech land grab. Ad networks are racing to lock in advertiser relationships, reporting integrations, and creative workflows before OpenAI builds those capabilities natively and potentially disintermediates the same partners.

This guide maps the full minimum-spend timeline, compares every entry route into ChatGPT ads side by side, separates the real reach from the unverified performance claims, and gives you a structured way to test the channel now without treating it as a scaled acquisition line. For the mechanics of how the ads are priced and targeted, our companion guide on how ChatGPT ads are priced and targeted covers the auction in depth.

The speed of the change is the story. When the ChatGPT ad pilot launched on February 9, 2026, the entry commitment was reportedly $200,000–$250,000 at a $60 CPM. By June, an SMB could route a campaign through StackAdapt with no minimum at all. Three separate actors drove the compression, and it is worth keeping them distinct because their terms are not interchangeable.

First, OpenAI launched its self-serve Ads Manager on May 5, 2026, opening CPC bidding alongside the existing CPM model and removing the platform-level minimum spend requirement for US businesses. Second, StackAdapt — which started its ChatGPT pilot with a $50,000 minimum in April — opened its self-serve model to roughly all of its ~1,000 advertisers in mid-May with no minimum at all. Third, Criteo cut its ChatGPT campaign minimum from $50,000 to $10,000 in early June, pairing the cut with media-match incentives and simplified product-feed integration to court retail brands.

No single source has assembled the full arc, so we built it. The table below tracks each lever that moved the cost of entry between February and June 2026. The delta column is the change in stated minimum at each step, computed from the before/after figures in each row.

The arithmetic is stark. From the February baseline near $250,000 to the May–June floor of $0, the cost of getting into ChatGPT ads fell to nothing across self-serve routes — with Criteo holding a $10,000 managed-retail tier in between. The pricing pressure ran in parallel with the entry-cost pressure: the pilot's $60 CPM was reported as low as $25 in buyer feedback by mid-April, a figure we cite as a historical low point rather than a live floor, because auction pricing on a new exchange moves continuously. For the UK-specific mechanics of the June 19 expansion, see our first-mover playbook for ChatGPT ads.

There is no single front door. You can buy direct through OpenAI's self-serve Ads Manager, or you can route through one of the original ad-tech partners — StackAdapt and Criteo are the two most relevant for most advertisers, alongside Adobe, Kargo, and Pacvue. The capability bars below contrast what each route gives you, so you can pick by profile rather than by headline minimum.

Han's framing is the cleanest articulation of the routing decision. If ChatGPT is the only place you want to spend, going direct is reasonable and now costs nothing to start. If ChatGPT is one line in a broader programmatic plan, a buying layer that consolidates the audience view earns its margin — which is precisely the value proposition the ad-tech partners are racing to entrench before OpenAI can match it natively.

Most coverage reports the minimum numbers and stops. The more useful question is why a given advertiser should pick one partner over the other — and the answer is that they are not really competing for the same buyer. Criteo is a retail-and-commerce play; StackAdapt is a multi-channel buying play. The choice matrix below sorts the decision by advertiser profile.

One detail worth weighing: Criteo's commerce footprint — the scale of shoppers and transactions it already touches — is the asset StackAdapt cannot replicate, and StackAdapt's cross-channel audience graph is the asset Criteo cannot replicate. Neither is strictly better; they map to different buyer shapes. For the verification and identity layer that increasingly sits underneath both, see our breakdown of the agentic ad-tech buying layer.

Here is the original analysis, the part the vendor press releases leave out. Cutting a minimum to zero is not charity — it is a customer-acquisition tactic with a strategic payoff. By onboarding advertisers now, ad-tech partners embed themselves into reporting integrations, creative workflows, and product-feed plumbing that are expensive to rip out later. The relationship, once formed, is sticky. That is the land being grabbed.

The forward-looking risk for those same partners is that they are building the on-ramp for a platform that may eventually want the traffic for itself. OpenAI's hire of Samantha Jacobson from The Trade Desk to lead its partnerships business signals fluency with the full ad-tech ecosystem — and an awareness of where the value concentrates. The Trade Desk's own CEO Jeff Green has compared AI-chatbot advertising to Netflix before it launched ads: platforms that spent heavily on infrastructure will eventually lean on advertising to offset the cost. The partners cutting minimums today are betting they can entrench before that shift.

None of this means the partners are doomed or that advertisers should avoid them — buying layers add genuine value, and the major holding companies will keep needing them. It means you should enter with eyes open: the relationship you form today is the asset two parties are competing to own, and your leverage is highest while minimums are at zero and everyone wants your test budget.

The audience is not in dispute. As of a June 2026 eMarketer forecast, ChatGPT is used by nearly three-quarters of the 152.9 million US generative-AI users — a genuinely large, mostly high-intent surface. The pull for advertisers is obvious. The complication is that the standout performance numbers come from the vendors who benefit from them, and consumer sentiment runs the other way.

That is a real tension, and it is the most important thing to internalise before you spend. The channel can perform well on early-adopter, high-intent traffic while consumers simultaneously say ads there make them trust the output less. The resolution most likely sits in timing: early advertisers are catching mid-funnel, high-intent users before broad skepticism compounds. That window is an advantage, but it is also a reason to keep creative non-intrusive and to measure brand impact, not just clicks.

The cleanest evidence that this is still a pilot, not a platform, is the gap between the projections. OpenAI projects $2.5 billion in ad revenue in 2026 — a vendor-stated figure that independent analysts have not confirmed; the closest corroboration is a Barclays research note (April 14, 2026) projecting $2.4 billion in 2026. Set against that, eMarketer expects ads in AI chatbots to generate less than $1 billion in 2026, roughly 3% of total AI advertising spend. That is the distance between ambition and current institutional scaling.

What sits in that gap is measurement infrastructure. The pilot crossed $100 million in annualised revenue within six weeks of launch, and well over 1,000 brands are live via Criteo — the demand signal is there. But enterprise buyers and agencies have said plainly that test budgets will not become real budgets until third-party verification, identity resolution, and trustworthy reporting are in place. DoubleVerify's CEO has positioned the company as exactly that trust layer for AI advertising.

Project forward and the shape is clear: 2026 is the bridge year. The partners cutting minimums are racing to onboard advertisers; the verification vendors are racing to make the inventory institutionally safe; and OpenAI is building the self-serve platform and measurement APIs that could one day make both groups optional. For a marketing team, the implication is not to wait for the dust to settle — it is to test now, cheaply, with measurement you control, so that you have real first-party data on whether the channel works for your offer before budgets and competition scale.

The minimums being at zero is the opportunity; the measurement gap is the reason to be disciplined. A structured test costs little and tells you what no vendor case study can — whether ChatGPT inventory converts for your specific offer. Here is the approach we use with clients.

Done this way, the test is close to free to start and bounded in risk, and it generates the one thing that actually matters when the channel scales: your own first-party evidence. If you want help scoping the experiment, wiring the measurement, and reading the results against your other channels, our paid media engagements are built for exactly this kind of structured channel test, and our agentic search and visibility work covers the broader question of how AI-assistant surfaces reshape discovery. For the wider context on how brands show up in AI search, our guide to brand visibility in AI search is a useful companion.