ChatGPT Ads reaches $1bn run rate as OpenAI goes global

OpenAI's advertising business crossed a $1 billion annualised revenue run rate in mid-2025, a figure the company disclosed on its blog under the understated heading "A milestone in expanding access to AI." The milestone matters less as a financial curiosity than as a structural signal: ChatGPT is no longer a subscription product with advertising as an afterthought. It is becoming an ad-funded platform, with all the editorial and algorithmic consequences that implies.

The comparison to Google's early years is tempting and mostly wrong. Google's ad model was built around a fundamentally passive interface: a results page where brands competed for placement in a list. ChatGPT answers in prose. The unit of value is not a click-through rate on a ranked link; it is the content of the answer itself. When advertising enters that environment, the incentive structure changes in ways that search marketing playbooks do not yet account for.

What an ad-funded answer engine actually rewards

The conventional worry about AI advertising is that sponsored content will crowd out organic citations. That is a reasonable concern, but it misses the more consequential shift. Ad-funded models require scale, and scale requires free access. Free access expands the user base. A larger, more diverse user base generates more and more varied queries. That query diversity, in turn, increases the value of being cited organically across a wider range of topics, because the model must answer a wider range of questions. For brands in complex, high-trust sectors, such as financial services, multilateral institutions, or industrial groups with long sales cycles, this expansion of the query pool is both an opportunity and a threat, depending on who has invested in the conditions that make organic citation possible.

Organic citation in an LLM is not won by bidding. It is won by being the kind of source a model trusts to answer a question accurately. That trust is built through the quality, specificity, and corroboration of content across the open web. The arrival of paid placements does not change that mechanism; it makes the stakes of not having invested in it higher, because the alternative to organic citation is not neutrality. It is invisibility, or worse, displacement by a competitor who paid.

For a multilateral institution or a UN-adjacent body, which typically cannot and will not run paid advertising, this dynamic is stark. If ChatGPT's ad inventory grows to the point where sponsored responses dominate high-volume queries in policy-adjacent areas, the institutions that have failed to build organic citability will find their outputs and expertise simply absent from the answers that senior decision-makers increasingly rely on. IEEE, ISO, and similar standards bodies face exactly this problem: enormous intellectual authority, relatively modest investment in the content infrastructure that LLMs draw on.

The global expansion matters as much as the revenue number

OpenAI's blog describes a global rollout, not merely a US launch. That geographic breadth changes the competitive context for brands operating across multiple jurisdictions. A financial institution that is highly cited in English-language training data may be poorly represented in Portuguese, Arabic, or Mandarin corpora. An ad-funded model expanding into new markets will, over time, generate query data in those languages, reinforcing citation patterns in whatever content already exists. Brands that have not published substantive, model-accessible content in those languages are building a structural disadvantage that compounds with each passing month.