OpenAI says ChatGPT Ads reached a $1 billion annualized run rate in under 200 days, and opened self-serve buying across India, Europe, MEA — company-stated (OpenAI, 31 Aug 2026)

Why it matters (spine). The cabinet's marketing thread has been tracking where advertising money goes as discovery moves into AI answer surfaces: Alphabet's filed numbers showed the owned search funnel compounding while the syndicated open web stalls (google-owned-search-vs-open-web-ad-split), and Reddit's filing showed a human-conversation corpus being monetised by selling attention around it (reddit-q2-2026-human-corpus-ad-premium). This post is the third datapoint in that series and the first from inside a conversational assistant itself: the place where, in OpenAI's words, "discovery, consideration, and decision-making come together in a single experience" now carries ads, at a claimed billion-dollar annual pace, with self-serve buying. The marketing vein was flagged as under-swept in the last two runs; this is its card.

What the document says — growth and rollout. "In less than 200 days after launch, ChatGPT Ads has reached $1 billion in annualized revenue run rate. The platform is now used by tens of thousands of advertisers and continues to expand globally. Starting later today, advertisers can purchase ChatGPT ads directly via Ads Manager across India, Europe, the Middle East, and North Africa." Context the post gives for the model: advertising is "one pillar of OpenAI's diversified business model, alongside consumer subscriptions, enterprise offerings, and usage-based APIs," supporting "an advertising-supported free tier that helps keep ChatGPT available to more than 1 billion weekly active users." Reach: "ChatGPT Ads are now available in over 40 countries." Maturity claims: "The introduction of Ads Manager in May opened the platform to small and medium-sized businesses, which now represent a material share of the business," and "CPC and outcome-optimized bidding now account for the majority of campaigns, while Pixel and Conversions API have become important foundations for measurement and optimization," with "more than 50 technology and measurement partners." Its two outcome examples, quoted as the anecdotes they are: "an ecommerce advertiser achieving a 3x return on ad spend across its campaigns over 28 days, while a technology partner reported that more than 80% of ad-driven ChatGPT traffic is from new customers."

What the document says — targeting and the trust claims. The targeting description is the operationally important passage: "Our ads system uses the context of the current conversation to show an ad relevant to what someone is exploring. Depending on the country and a user's settings, the ads system may also use context from the user's broader ChatGPT experience." The trust principles, quoted whole because their falsification would be the bigger story: "Ads in ChatGPT are always clearly labeled and separate from ChatGPT's answers. Advertising does not influence the answers ChatGPT provides. Advertisers do not receive access to people's private conversations, and users can control how their ad experience is personalized." And the next-phase sentence: "We will also explore new ways for businesses to interact with consumers in more native ways in ChatGPT."

Every figure here is company-stated. OpenAI does not file public financial statements; nothing in this post is audited, no methodology is given for the run rate, the advertiser count, or the weekly-active-user figure, and the two outcome examples are unnamed single cases supplied by the seller of the ad product. The card is verified as a first-party document — OpenAI stated these things on this date — not as independent confirmation that they are true.

Derived figures, enumerated so a missing one is detectable. Three derivations appear on this card, and only these. (1) Scale against Google: Alphabet's filed Q2 2026 Google advertising total was $81,629 million (read from google-owned-search-vs-open-web-ad-split, whose source is Alphabet's Exhibit 99.1); annualized as ×4 that is $326.5 billion, and $1 billion ÷ $326.5 billion = 0.31% — under one-third of one percent. (2) Scale against Reddit: Reddit's filed Q2 2026 advertising revenue was $762 million (from reddit-q2-2026-human-corpus-ad-premium); ×4 is $3.05 billion annualized, of which a $1 billion run rate is roughly a third. Both comparisons set a company-stated forward run rate against annualized filed quarterly actuals — an approximation in both directions, stated as such. (3) The launch-date bound: a post dated 31 August (day 243 of 2026) claiming "less than 200 days after launch" implies launch after day 43, i.e. after roughly 12 February 2026; the post itself gives no launch date and this card asserts only the bound.

Cross-links, in the direction that complicates. The Google card's thesis — money concentrating into owned surfaces while the syndicated web stalls — is supported by a new owned surface monetising, but that card's caveat cuts here too: revenue says nothing about query mix or user behaviour, and a run-rate claim says even less than a filed revenue line. The Reddit card's correction is the sharper one: it established that the demonstrated business model for human-AI conversation platforms is selling attention around the conversation, not licensing data — and OpenAI's post confirms that pattern from the assistant's side of the market rather than complicating it. What this card does complicate is any MktKG assumption that ChatGPT advertising is still an experiment: self-serve buying across four more regions, majority CPC/outcome-optimized bidding, and conversion APIs is the description of an ad platform, not a pilot — as described by its seller.

Garden-code note (L41, flagged not resolved). This card's content sits where the marketing vein meets the machine layer. TKG carries two live meanings in this repo — Toxicology per the Frontier Map, the machine layer per this beat's own wording — so, as on the seven earlier abstaining cards, TKG is left out of gardens and the machine-layer link is carried in domains. The founder's call on the code stands open.

Money angle. Directly: conversational ad inventory is now purchasable self-serve across most of the world's large markets, and MktKG's clients can test it at small budgets — the post says SMBs are already "a material share." For RDKG's own thesis, the sharper reading is that the answer surface is monetising attention while the underlying answers remain, per OpenAI's stated principle, uninfluenced — a separation whose durability is exactly what a verified-ground-truth vendor should track, because ad-influenced answers would make verified, citable sources more valuable, not less. Metacrisis lever: a billion people using a free tier funded by conversation-context advertising is a new attention economy forming inside the most personal interface yet built; the post's own control-and-labeling claims are the terms to hold it to.

What would prove it wrong (falsifier). Four limbs.

• The document fails if the post is withdrawn, redated or materially revised, or if these quotations do not appear in it.
• The figures fail if OpenAI restates the run rate or user numbers, or if any later audited or filed disclosure contradicts them — none exists today, which is why every figure is labeled company-stated.
• The rollout fails if self-serve Ads Manager access in India, Europe, the Middle East and North Africa does not in fact open or is withdrawn — testable against OpenAI's own availability page (linked from the post, not opened for this card).
• The separation principle fails if labeling weakens or if credible evidence emerges that advertising influences answers — the claim is quoted whole above so a future run can test it limb by limb; its failure would flip this card's so_we_should from "plan for the surface" to "the surface broke its own terms."

Provenance (what I actually read). One first-party document, fetched from the main thread on 1 September 2026: openai.com/index/expanding-access-to-ai-with-chatgpt-ads/, titled "A milestone in expanding access to AI," carrying the visible dateline "August 31, 2026" and the tag "Product." Every quotation above comes from that page — enumerated: the run-rate/launch sentence, the advertisers sentence, the Ads Manager rollout sentence, the business-model pillar sentence, the free-tier/WAU sentence, the 40-countries sentence, the Ads-Manager-in-May sentence, the bidding/measurement sentence, the partners figure, the two outcome examples, the targeting passage, the four trust-principle sentences, and the native-interaction sentence. The two comparison figures are read from the two named cabinet cards, both opened in the repository this run, and their ultimate sources are those cards' SEC exhibits. I did not open OpenAI's ads-principles post, its Ads Manager availability help page, or any coverage of this announcement; a Forbes article dated 31 August circulated in search and nothing on this card comes from it.