In Lagos, Buenos Aires, Karachi and Kyiv, a free ChatGPT user will start seeing sponsored links. OpenAI said Wednesday that it expanded advertising to 26 more countries, bringing the total to 89, roughly eight months after it began testing ads in the United States.
The new markets include 14 more Latin American countries, among them Argentina, Chile, Colombia and Peru, following earlier launches in Mexico and Brazil. OpenAI is also opening Nigeria, South Africa, Serbia, Ukraine, Cambodia, Bangladesh and Pakistan. Advertisers can buy through the company’s Ads Manager or through technology partners.
The rules are unchanged. Ads appear only on the Free and Go plans; Plus, Pro and Enterprise subscriptions stay ad-free. OpenAI says ads are labeled, kept separate from ChatGPT’s answers and do not influence those answers, and that conversations remain private from advertisers and customer data is not sold.
The scale it cites is large. More than 100,000 advertisers have run campaigns since the product launched, according to the company, and 1.2 billion people use ChatGPT each week. CNBC reported that the ad business reached a $1 billion annualized revenue run rate in August, up from $100 million in March.
That curve explains the geography. Advertising is now the fourth revenue line alongside enterprise contracts, consumer subscriptions and usage-based access to its models, and it is the only one that scales with free users rather than paying ones. OpenAI expanded ads across Europe in August and into Southeast Asia and Taiwan in September. The company said its next phase will include new formats, objectives, buying options and measurement.
For advertisers, the pitch is placement at the moment of consideration rather than after it. A person planning a first marathon asks about meals for race week, refines the options, and a grocery delivery service appears. That is a different inventory from a search result, where intent has already hardened into a query, and different from a social feed, where the interruption competes with friends.
L’Oréal is treating the format as a test bed. “Testing early with OpenAI enables us to pioneer helpful experiences and journeys, from discovery, to evaluation, to agentic commerce,” Asmita Dubey, the group’s chief digital and marketing officer, said in OpenAI’s announcement. The phrase agentic commerce is the part worth watching: an ad that a user can act on inside a conversation is closer to a transaction than to a banner.
The economics of the expansion are uneven by design. Ad rates in Nigeria, Pakistan or Bangladesh are a fraction of those in the United States, and revenue per free user in those markets will be small for years. What OpenAI gets is volume, a foothold in markets where its rivals are strong, and advertiser relationships that local businesses can form before a competitor teaches them otherwise. The company says the free tier exists to support low-cost access, and ads are how that access is paid for.
The measurement problem is the one advertisers will press on. Attribution from a conversation to a purchase is harder than counting clicks, and OpenAI lists measurement among the capabilities it still plans to build. Advertisers will also want targeting, which sits awkwardly with a promise that conversations stay private and data is not sold. Contextual placement around a stated need is the compromise the company is offering: less precise than a profile, more defensible.
There is a second tension, and it is the one Anthropic has made its marketing. The rival model maker mocked OpenAI’s advertising push in a Super Bowl campaign and pledged that Claude will remain ad-free. Its argument is that an assistant which recommends products cannot also sell them without degrading the recommendation. OpenAI’s answer is structural separation: a labeled block, outside the answer, with no effect on what the model says.
Separation is easier to promise than to prove. A search results page can put ads above organic links because the two are visibly different objects. A conversation is a single stream, and the answer itself is where the commercial intent lives. If a user asks which running shoes suit a bad knee, the value of the ad depends on the answer being good, and the advertiser’s interest and the user’s interest are adjacent in a way that is hard to police with formatting alone.
Regulators in Europe have already had one look. The August expansion put ads in front of users covered by rules on labeling, profiling and dark patterns, and the enforcement record there will shape how quickly OpenAI moves in the rest of the 89. The company’s next phase, new formats and buying options, is exactly where those rules bite.
Both OpenAI and Anthropic are preparing for widely expected listings, and OpenAI executives have signaled a 2027 debut. A billion-dollar ad run rate is a useful number to carry into that process, and an ad business in 89 countries is a more durable story than a subscription business concentrated in a handful of rich markets.
What remains unanswered is whether an ad placed next to a recommendation can stay outside it. OpenAI is betting that labeling and separation are enough, and that free access is worth the trade. The users in the 26 new markets will decide with their attention, and the advertisers will follow wherever it goes.
