ChatGPT Ads Hit a $1 Billion Run Rate: What This Means for Your Marketing Budget

Less than a year ago, Sam Altman was still describing advertising inside ChatGPT as a last resort, something OpenAI would only touch if every other revenue idea failed. On 31 August the company announced that its ads business had crossed a $1 billion annualised revenue run rate, less than 200 days after launch. So much for the last resort.

The same announcement carried a second piece of news that matters more for most of our readers. OpenAI opened self-serve access to its Ads Manager for advertisers across Europe, India, the Middle East and North Africa. Until now, buying ChatGPT ads in these regions meant going through a select group of agency partners. From this week, a small company can open an account, pass the policy checks and run a campaign on its own, the same way it would on Google or Meta.

A new advertising channel with a billion weekly users behind it does not appear very often, so it is worth slowing down and looking at what is actually on offer. We went through the announcement and the first advertiser reports to see how the ads work, what the early results look like, and whether this deserves a slice of your marketing budget yet. Let's sink in!

From last resort to a billion dollars

The headline number is impressive on its own. A $1 billion annualised run rate means ChatGPT ads are currently bringing in roughly $83 million a month, and OpenAI says tens of thousands of advertisers are already buying. For comparison, it took Google and Meta considerably longer to reach the same milestone in their early years, although both launched into a much smaller online ad market.

There is a less flattering way to read the same figures. OpenAI's own target for 2026 is $2.5 billion in recognised revenue, and analysts estimate the company has actually booked somewhere around $330 million so far this year. Nate Elliott of eMarketer put it bluntly: the business has ramped with amazing speed, and it is still on course to miss its own goal by a wide margin. The long-term ambition of $100 billion in ad revenue by 2030 would require growth rates that no advertising platform has ever sustained.

Both readings are true at once. This is the fastest-growing ad platform in the market right now, and it is also a young product that has to prove it can keep scaling. For an advertiser, that combination usually means one thing: early access is cheap relative to what it will cost later, if the platform works out.

How the ads actually appear in ChatGPT

If you have not seen a ChatGPT ad yet, the format is fairly restrained so far. Ads are labelled and kept visually separate from the assistant's answers. OpenAI is firm that advertising does not influence what ChatGPT says, which is the line the whole model depends on. The moment users suspect the answer itself is sponsored, trust in the product collapses, and OpenAI knows it.

Targeting is built primarily on context. The system looks at what the current conversation is about and shows ads that fit the topic, a bit like search ads matching a query. Where users allow it, broader interests come into play as well. Advertisers do not get access to private conversations, and users can control how personalised their ad experience is.

What makes the placement interesting is the nature of the conversations themselves. People ask ChatGPT things they would never type into a search box in that form: long, specific, full of intent and constraints. Someone comparing accounting software for a five-person company in the UK has told the assistant more about their situation than any keyword ever could. That context is what advertisers are really paying for.

What the self-serve rollout changes

Until this week, the practical answer to "how do I buy ChatGPT ads in Europe" was "you probably don't, unless your agency is on the partner list". The Ads Manager, which launched for US small businesses back in May, is now open in beta across 31 European markets plus India, the Middle East and North Africa. That takes the channel from a managed product for large brands to something a small business can test with a modest budget.

The toolkit will feel familiar to anyone who has run performance campaigns. There is CPC and outcome-optimised bidding, which OpenAI says most campaigns now use. Measurement runs through a pixel and a Conversions API, product feeds are supported for ecommerce, and there are custom audiences and geographic targeting. More than 50 technology and measurement partners are already integrated.

There are gates, though. Advertisers have to sit within approved categories, and every ad passes policy checks before it runs. OpenAI is clearly trying to avoid the low-quality ad clutter that plagued early social platforms, at least while the product's reputation is still forming.

The early numbers are honestly mixed

OpenAI's own case studies are the ones you would expect. An ecommerce advertiser reports a 3x return on ad spend across campaigns over 28 days. A technology partner says more than 80% of ad-driven ChatGPT traffic comes from new customers, which is the statistic that should interest anyone whose Google and Meta campaigns keep reaching the same audience over and over.

Independent reports are less tidy. One Canadian agency told MediaPost it was seeing click-through rates around 0.6%, against a typical 2% on comparable search ads. Another advertiser found performance competitive with Meta and Google only after switching to the newer Max Results bidding strategy. That spread is normal for a young platform: the auction is thin, the formats are still settling, and nobody has years of optimisation data yet.

The honest summary is that ChatGPT ads today look like Facebook ads did around 2012. Results vary wildly, the measurement stack is immature, and the advertisers who benefit most are the ones treating it as an experiment with a defined budget rather than a replacement for a proven channel.

Where this leaves Google and Meta

Nobody should be moving their core budget this quarter. Google still owns the deepest pool of commercial intent, and Meta still owns scaled social discovery. But the direction of travel is hard to ignore. Google itself now shows ads in almost a third of commercial AI Mode answers, as we covered recently, and OpenAI is building a direct competitor for the same shift in behaviour: people increasingly start their research inside an assistant rather than on a results page.

For advertisers, the sensible frame is incrementality. If 80% of ChatGPT ad traffic really is new customers, the channel is not cannibalising your search campaigns, it is reaching people earlier in their decision, while they are still asking open questions. That stage of the funnel has always been expensive to reach through search and noisy to reach through social.

There is also a defensive angle. Your competitors can now appear inside conversations where a potential customer is comparing options in your category. Being absent from that surface has a cost too, even if it never shows up in a dashboard.

What businesses building digital products should do now

If you sell online and sit in an approved category, the practical move is a small structured test. Set aside a budget you can afford to learn with, install the pixel or wire up the Conversions API before spending anything, and run outcome-optimised campaigns against one clear conversion goal. Judge the channel on cost per new customer, not on click-through rate, because the early CTR numbers will look poor next to search and that is not the point of the placement.

If you run an ecommerce catalogue, set up the product feed properly. Feed-based formats are usually where conversational platforms monetise first, and clean product data will matter here just as it does on Google Shopping.

And whether or not you buy ads at all, this rollout is another reminder that assistants are becoming a real discovery surface. The same content and structured data work that earns you visibility in AI answers organically also shapes how relevant the paid side considers you. Businesses that treat their site as something AI systems read, not just people, keep coming out ahead of this shift.

Final notes

A billion-dollar run rate in 200 days settles one argument: advertising inside AI assistants is now a real market, not a thought experiment. Whether it becomes a core channel for mid-sized businesses depends on numbers we will only see over the next year, once the novelty wears off and the auction fills up.

Our advice is the same as with every new platform we have written about this year. Test early while attention is cheap, measure properly, and keep your expectations tied to data rather than press releases. And if you are not sure how your product data, site content or tracking setup would hold up on a channel like this, that is worth fixing now, before your competitors' ads start showing up in your customers' conversations.