OpenAI announced that its ChatGPT Ads platform reached a $1 billion annual revenue run rate in under 200 days after its US debut. Ads within ChatGPT are already being shown to an audience of over 1 billion weekly users, and self-serve ad buying through Ads Manager is open far beyond the US.

image

What Happened

ChatGPT ads launched in the US in February 2026, Ads Manager opened for self-serve buying in May, and starting August 31 it is expanding to India, Europe, the Middle East, and North Africa. In total, the platform is available in over 40 countries and relies on an ecosystem of at least 50 technology and measurement partners. According to CNBC, the annual run rate in March was only about $100 million, meaning the figure grew roughly tenfold in six months. Among the first advertisers are Newegg, Best Buy, Lowe's, and VistaPrint; OpenAI cites an e-commerce client with a 3x ROAS over 28 days and offers new advertisers a $500 credit upon spending $500.

Context

Before ads, OpenAI earned revenue from subscriptions, API access, and enterprise products, and ChatGPT Ads is integrated into this mix as a separate revenue stream. The bet is on a high-intent format: instead of keyword targeting, the context of the conversation is considered, which OpenAI describes as "more than keywords." The platform's expansion is happening against the backdrop of the company's preparation for an IPO, expected no earlier than 2027, and a valuation of $852 billion according to CNBC. On the other end of the competitive landscape is Anthropic, which criticized AI ads in its first Super Bowl ad: the rejection of such monetization itself becomes an argument in the battle between labs.

Why This Matters for the Industry

For the industry, this confirms that monetizing LLM chats through high-intent ads works at scale: the format with conversation-context targeting has passed its first commercial test with major retail brands. The growth to a $1 billion run rate in about six months, according to the source's interpretation, is an order of magnitude faster than the launches of Google's and Facebook's ad businesses. For the ad ecosystem, the speed of self-serve model rollout is also significant: Ads Manager is already opening in new regions where inventory is just forming and competition for impressions is minimal. However, from an engineering perspective, this is an early-stage platform: there is no public API, and integrations rely on Ads Manager and the ecosystem's measurement partners.

Why This Matters for Users

Free use of ChatGPT is becoming sustainable — but at the cost of ads within conversations, which are seen by free-tier users and Go subscribers; free users make up the majority of the audience. OpenAI states that ads are labeled, do not affect the model's responses, and advertisers do not get access to private conversations; personalization management is available via a toggle in settings. For marketers and self-promotion enthusiasts, self-serve has become more accessible without intermediaries: in new regions, you can independently set up a campaign with CPC bidding, conversion optimization through Pixel and Conversions API, and campaign upload via CSV, and the "$500 credit for $500 spend" promo significantly reduces the cost of the first experiment.

What Is Still Unknown / Limitations

The key figure is a vendor claim: the $1 billion run rate is an extrapolation, not recognized revenue, the calculation methodology and structure (signed budgets vs. pace projection) are not disclosed, and so far only CNBC has confirmed it from external sources. The comparison of growth speed with Google's and Facebook's ad businesses is an interpretation, not a measurement. The claim that "ads do not affect the model's responses" is presented without methodology: there is no independent audit or metrics of commercial bias, and the mechanics of conversation-context targeting are not disclosed. The 3x ROAS case is a single example from an interested party, without a baseline and without an efficiency distribution across all advertisers, against the backdrop of an active self-promotion promo.

Sources

Author

Look at AI, editorial team