On August 31, OpenAI stated that ChatGPT advertising reached a $1 billion annualized run rate "in under 200 days" from launch. However, the same announcement did not specify the exact starting point used to define when advertising began. Dividing this figure by 12 months yields a monthly equivalent of roughly $83.3 million. This differs from actual single-month revenue booked. It's a figure calculated by assuming the most recent sales pace continues for a full year, and it needs to be distinguished from cumulative revenue or audited annual sales.
Still, this figure carries meaning. Viewed not as a cumulative total but as a measure of sales velocity, it reveals how quickly a pilot program—once selling impressions to a small number of companies—transformed into a marketplace where advertisers themselves can purchase, bid, and measure results. Interpreting the $1 billion annualized figure becomes an exercise in separating out just how far that sales infrastructure has actually developed.
What Emerges When You Convert $1 Billion Back to a Monthly Figure
The $1 billion annualized revenue figure, divided by 12 months, equals a monthly equivalent of approximately $83.3 million. This is an equivalent value normalizing the same sales pace to a monthly basis—it is not a figure OpenAI has disclosed as actual single-month revenue or cumulative revenue. The company has not revealed the observation period used to calculate the annualized figure, nor has it published contracted annual revenue. The costs and profits of the advertising business also remain unknown.
Therefore, this figure cannot be used to determine 2026 advertising revenue with certainty. What it does reveal, however, is roughly how fast sales had accelerated by that point in time. OpenAI states that tens of thousands of advertisers can now purchase ads across more than 40 countries, with ChatGPT's weekly active users exceeding 1 billion. These figures suggest that both the buying side and the audience reach have grown well beyond the scale of a small-scale pilot for a handful of companies.
On March 31, the company had announced that the advertising pilot reached an annualized run rate exceeding $100 million in under six weeks. By late August, that threshold had risen by an order of magnitude. However, since the March figure was described only as "exceeding $100 million," it cannot be calculated as having grown precisely tenfold. What can be confirmed here is simply that the annualized sales velocity stood at markedly different levels at these two points in time.
Annualized figures are convenient for quickly comparing the momentum of a newly launched business. On the other hand, if the observed sales pace changes, the figure extrapolated out to a full year shifts just as quickly. This metric cannot distinguish whether advertisers continued buying the following month or whether growth was concentrated in a particular month due to seasonal factors. This is precisely why the roughly $83.3 million monthly equivalent should be treated as a yardstick for comparison rather than as proof of sustained performance.
The Sales Infrastructure That May Have Driven Revenue Growth
ChatGPT advertising expanded its target audience, purchasing methods, and measurement tools in stages: the January policy announcement, the February U.S. test, the run rate exceeding $100 million by the end of March, the introduction of Ads Manager and click-based pricing in May, and the expansion to over 40 countries alongside the $1 billion run rate by the end of August. Advertising started out separated from answers and offered to a small number of advertisers on an impression-based pricing model, but after the U.S. test launched on February 9, both ad inventory and the pool of buyers expanded simultaneously.
| Date | Change Announced by OpenAI |
|---|---|
| Jan 16 | Announced policy to test ads on Free and Go tiers |
| Feb 9 | Launched ad testing in the United States |
| Mar 31 | Exceeded $100M annualized run rate in under 6 weeks |
| May 5 | Introduced Ads Manager, click-based pricing, and performance measurement |
| Aug 31 | Expanded to 40+ countries, tens of thousands of companies, $1B annualized run rate |
This sequence reveals that, alongside user growth, a system enabling advertisers to purchase directly and measure results also came together. After establishing the venues where ads could appear, tools were added in sequence that let advertisers purchase directly, measure outcomes, and feed those results back into subsequent campaigns. The period when sales velocity increased overlaps with the period when purchasing options expanded. However, the extent to which each feature contributed to revenue has not been disclosed.
The turning point was the Ads Manager beta launched on May 5. Advertisers can complete registration and payment, then set their own budgets and bids. After launching a campaign, they can check performance results in the same dashboard. While sales through agencies and technology partners continue, the fact that advertisers can now handle the process themselves means OpenAI can expand its customer base without relying on individualized support.
As of August 31, direct purchasing through Ads Manager had expanded beyond India and Europe to include the Middle East and North Africa. According to OpenAI, the share of revenue coming from advertisers outside the United States is also growing. However, region-by-region figures and ratios have not been disclosed. The expansion to over 40 available markets demonstrates reach, but it remains unclear which markets actually generated the revenue.
How Click-Based Pricing and Performance Measurement Changed Advertisers' Work
Under the original impression-based pricing model, advertisers were paying for the volume of ad impressions displayed. The May expansion introduced click-based pricing, along with the ability to use the Conversions API and pixel-based measurement. Advertisers can now manage budgets and bids while monitoring the outcomes that occur after a click, adjusting their campaigns accordingly.
The two changes serve different roles. Click-based pricing ties charges not to how many times an ad was viewed, but to the actual action of a user opening the ad. The Conversions API and pixel measurement give advertisers a way to verify, on their own end, whether a purchase or sign-up occurred afterward. Being able to distinguish between impressions, clicks, and post-click outcomes lets advertisers move beyond simply buying exposure volume. That said, this system doesn't guarantee purchases will occur.
By late August, campaigns using click-based pricing and automated bidding aligned with performance outcomes accounted for the majority of campaigns. The latter refers to a bidding method that adjusts bids according to the outcomes advertisers seek, such as clicks or purchases. This has made it easier to tie budgets to behavioral outcomes rather than exposure volume, shifting operations from a trial run of buying impressions toward a model that adjusts campaigns based on post-ad behavior.
The number of technology and measurement partners has surpassed 50, and OpenAI says small and medium-sized businesses account for a meaningful share of the business. That said, the company has not disclosed the actual proportion represented by small and medium-sized businesses. There are also accounts of one e-commerce advertiser achieving 3x return on ad spend within 28 days, and one technology partner reporting that over 80% of ad-driven traffic consisted of new customers. However, company names and sample sizes were not disclosed, nor were the comparison conditions made public. These are individual case examples and cannot be substituted for the average performance of ChatGPT advertising as a whole.
The extent to which Ads Manager, click-based pricing, and performance measurement contributed to revenue growth remains undisclosed. Even so, a pathway through which advertisers can purchase directly, verify results, and feed that information back into subsequent bids came together over the same period. It's reasonable to view the $1 billion annualized figure as reflecting the velocity of sales flowing through that pathway.
Measuring the Independence of Answers, Separately From Revenue
Ads appear on the Free and Go tiers and are clearly separated, displayed below ChatGPT's answers. OpenAI states that advertising does not influence the answers themselves. The company also says it does not share chats or history with advertisers, nor does it share memory or personal information. What advertisers receive, according to OpenAI, are aggregate figures such as total impressions and click counts.
OpenAI counts ad-supported free usage as one of its business models, explaining that it supports access to ChatGPT for the more than 1 billion weekly users. While this scale demonstrates the potential reach for delivering ads, it does not necessarily mean users have accepted the ads. This is precisely why revenue figures need to be kept separate from evaluations of user experience.
That said, ad selection does draw on the context of the current conversation. For users who have personalization enabled, past chats and memory may also be used. Users can disable personalization and can delete their advertising data. Growth in sales figures cannot serve as a metric for the independence of answers or for how users perceive the experience.
OpenAI states that metrics for trust, relevance, and overall user experience tracked during its global rollout have been favorable. However, the August 31 announcement did not disclose the definitions or actual values of these metrics, nor is it clear what methodology was used or whether any third-party verification took place. Advertising revenue and trust in answers require separate yardsticks entirely.
Before assessing how much further the annualized revenue figure will climb, it will be necessary to verify actual cumulative revenue, advertiser retention rates, region-by-region revenue, and the distribution of returns on ad spend. Regarding the independence of answers as well, disclosures are needed—separate from OpenAI's own explanations—that measure how much choice and control users genuinely experience.