Technology
ChatGPT Ads Hit $100 Million in Six Weeks. Here Is What That Number Actually Means.
OpenAI's advertising platform reached $100 million in annualised revenue within six weeks of launching on 9 February 2026. CPMs have dropped from $60 to $25 as the platform opened to self-serve advertisers on 5 May. The economics of advertising inside a conversation are different from anything that has come before.
By Patrick T ·
OpenAI launched advertising in ChatGPT on 9 February 2026, for Free and Go tier users in the United States. Within six weeks, the platform had reached $100 million in annualised revenue. On 5 May 2026, OpenAI removed the minimum spend requirement and opened the platform to all US advertisers through a self-serve Ads Manager at ads.openai.com. The CPM has dropped from $60 at launch to approximately $25 today. CPC bidding with $3 to $5 bid floors by category is now live alongside the CPM model. The full journey from invite-only enterprise pilot to open self-serve took 86 days.
The AI search advertising market is estimated at $15 billion to $25 billion in 2026. ChatGPT ads reached $100 million in annualised revenue in under two months — before the platform was even open to self-serve advertisers.
The $100 million figure is impressive, but it needs context. OpenAI processed 2.5 billion daily prompts at the time of launch, but a significant share of those users are on paid tiers — Plus ($20 per month), Pro ($200 per month), Team, Business, Enterprise, and Education — and will never see an ad. OpenAI has permanently positioned ad-free access as a feature of paid tiers. The addressable ad audience is the Free and Go tier user base in the United States, which is a meaningful but bounded segment of ChatGPT's total user base.
The chat_card: The Only Format That Exists Right Now
The only confirmed ad format at launch is the chat_card: a sponsored card that appears below the AI response, with a title of 3 to 50 characters, body copy of up to 100 characters, an image, a favicon, and a destination URL. The placement context is the defining characteristic of this format. A user mid-conversation with ChatGPT is not scrolling a social feed or scanning search results. They are describing a problem in natural language, actively in research or decision mode. When that conversation aligns with an advertiser's targeting criteria, the sponsored card appears after ChatGPT finishes its response.
OpenAI's content policy is explicit on one point: paid placements cannot influence what ChatGPT says in its organic response. The company calls this the Answer Independence principle. What ChatGPT says is determined entirely by accuracy and relevance to the user's question, not by which brands are spending on the platform. This has a direct impact on creative strategy: advertisers buy visibility alongside the organic answer, not a better position inside it.
Targeting Without Demographics: Context Hints
ChatGPT uses context hints for targeting. The system reads the semantic content and intent of the active conversation, combined with prior chat history, and matches relevant ads from advertiser inventory. There are no keywords, no demographic profiles, and no third-party behavioural data. Advertisers provide context hints describing the conversations where their product is relevant, and the platform handles matching. This is a fundamentally different model from Google Search (keyword intent) and Meta (demographic and behavioural audiences), and it comes with real tradeoffs: less granularity, no negative context exclusions, no placement-level transparency, and no retargeting.
The reduced granularity is uncomfortable for advertisers who rely on precise keyword segmentation. The upside is access to users who are actively mid-problem, describing their needs in more detail than a search query ever captures. Early data from Criteo shows that referrals from large language model platforms convert at 1.5 times the rate of other digital channels. ChatGPT referral visitors also spend 60 to 80 percent more time on-site than social traffic, with conversion rates comparable to high-intent Google search in tested categories.
The Pricing Trajectory
The platform moved through three distinct pricing phases in 86 days. At launch on 9 February, the CPM was $60 with a $200,000 minimum spend, available only through managed service via agency partners including Dentsu, Omnicom, Publicis, and WPP. On 10 April, the self-serve Ads Manager launched in beta, the minimum was reduced to $50,000, and CPC bidding was introduced at $3 to $5 floors by category. On 5 May, the platform opened to all US advertisers with no minimum spend, and CPMs had dropped to approximately $25.
The CTR Problem and Why It Is the Wrong Metric
The early ChatGPT ad pilot recorded a 0.91 percent click-through rate, compared to a 6.4 percent benchmark for Google search ads. That gap is real, but CTR is the wrong primary metric for a channel where users continue their conversation after seeing an ad instead of ending the session with a click. In search, a click typically ends the interaction. On ChatGPT, the ad appears mid-conversation and the user keeps talking. That session depth changes what performance looks like and how early results should be read.
The standout early adopter verticals share one trait: users who ask detailed research questions before making a purchase decision. B2B SaaS and software, financial services, travel, education, and home services have all shown strong early performance. E-commerce categories with short purchase cycles and lower average order values face a harder equation at current pricing. The right question for any advertiser evaluating the platform is cost per acquisition, not cost per thousand impressions — and the answer to that question will vary significantly by category and by the quality of the creative.
What This Means for Google
ChatGPT's advertising launch is the most direct competitive challenge to Google's search advertising business since the emergence of social media advertising in the early 2010s. The AI search advertising market is estimated at $15 billion to $25 billion in 2026, and ChatGPT is not the only entrant: Google's own AI Mode in Search, Perplexity's advertising product, and Microsoft's Copilot all represent different bets on how advertising will work in an AI-mediated information environment. The difference is that ChatGPT's conversational context is qualitatively different from a search results page — and the early conversion data suggests that advertisers who can navigate the targeting model will find it a high-quality channel.