The CEO has already sent the Slack message. Someone on the team screenshot the headline. And now you're being asked whether this is real or whether it's another platform promising the moon before delivering a thin slice of it two years later. OpenAI's self-serve Ads Manager launched on May 5, 2026. It is real. The platform works. And you probably shouldn't shift serious budget to it yet.

That tension is worth sitting with, because the noise around this launch runs in both directions. There are marketers who see nine figures of annualized revenue in six weeks and assume the floodgates are open. There are skeptics who see a favicon-sized ad unit and dismiss it entirely. Both positions miss the actual story.

What Launched and What It Actually Does

OpenAI's Ads Manager opened to all U.S. advertisers on May 5 in beta. The minimum spend threshold, previously $50,000 in the expanded pilot phase, is gone. Any business can register, set a budget, upload creative, and launch a campaign. The platform supports two bidding models: cost-per-thousand-impressions (CPM) at roughly $60, and cost-per-click (CPC) with a recommended starting bid of $3 to $5. Both are live now.

Measurement has moved further than the early pilot suggested. The Conversions API tracks post-click events including landing page views, catalog views, and add-to-cart actions. A tracking pixel is live for site-side event measurement. What remains unbuilt: third-party measurement partnerships, cost-per-action bidding, and granular geo-targeting beyond country level. OpenAI's ads and monetization lead Asad Awan confirmed all three are in development but declined to give timelines.

The ad format is a sponsored product card appearing below ChatGPT's organic response. It includes a brand logo, product name, price, and delivery information. It does not appear within the response itself, does not influence what ChatGPT says, and is clearly labeled as an ad. Targeting for logged-in users draws on conversational context and user profile; for logged-out users, it functions closer to keyword matching.

Why OpenAI Needs This to Work

ChatGPT has 900 million weekly active users as of February 2026, the majority on free accounts. Training and serving frontier AI models at that scale is extraordinarily expensive. The Stargate infrastructure joint venture with SoftBank carries a $500 billion price tag over four years. OpenAI's compute spending for 2026 alone is projected at $50 billion.

Subscriptions cover paying users. Advertising monetizes everyone else. The pilot crossing $100 million in annualized revenue within six weeks of its February 9 launch was enough to confirm the demand signal. OpenAI is now building fast: self-serve tools, CPC bidding, CAPI, and the Ads Manager all followed within three months of that first paid placement.

For marketers, the implication is straightforward. OpenAI is highly motivated. The platform will iterate quickly, and it will court your budget aggressively. The question is whether the infrastructure is mature enough to justify moving budget now, rather than in six to twelve months when measurement catches up to intent.

The Signal Is Genuinely Different

Google captures what people search for. Meta, which posted $56.3 billion in Q1 2026 revenue on 33% year-over-year growth, captures who people are and what they engage with. OpenAI captures something neither does as cleanly: what people are actively reasoning through at the moment they need help deciding.

When someone asks ChatGPT "what's the best running shoe for flat feet under $150," that is not a keyword. It is a purchase decision in progress, with constraints and intent embedded in a single prompt. The conversion data supports this: Criteo's analysis of 500 U.S. retailers in February 2026 found that users referred from large language model platforms converted at roughly 1.5 times the rate of other referral channels.

That is the structural argument for this channel. It is not that ChatGPT has more users than Google or a better ad auction than Meta. It is that the moment of placement is qualitatively different from anything in a standard media plan. The person is already in the middle of a decision conversation. That is not a reach metric. That is purchase proximity.

What the Measurement Gap Actually Means

Here is the practical constraint holding back serious budget allocation. Without third-party measurement, you cannot run a clean incrementality test against your existing channels. You cannot de-duplicate conversions. Your marketing mix model has a blind spot the size of the platform you're evaluating.

OpenAI's self-reported metrics are the only metrics available. Awan said third-party measurement partners are coming but gave no timeline and named no partners. Until that infrastructure exists, any spend is a learning investment, not a performance allocation. Your CFO is right to ask for third-party verified return on ad spend before approving a line item. The answer does not exist yet.

This is not unusual for a platform in its first year. YouTube, podcast advertising, and connected TV all went through a period where the intent signal was real but the measurement wasn't. The brands that built creative fluency and audience learning during that window had a meaningful advantage when the attribution infrastructure caught up. That is the actual argument for testing now.

How the Incumbents Are Responding

The competitive reactions are more telling than the launch itself. Google announced Meridian Studio (an enterprise marketing mix modeling tool) and an updated Meridian GeoX incrementality product at Google Marketing Live 2026. It has also rebuilt its search ad infrastructure around conversational intent via Gemini. Google is not watching this from the sidelines.

Meta's move is subtler and strategically interesting. Rather than building its own conversational AI ad surface to compete with ChatGPT, it opened its ecosystem via AI connectors, letting advertisers manage Meta campaigns through third-party AI tools. The bet is that Meta's ad inventory is more valuable than any single interface it lives in. If you can buy Meta ads from inside ChatGPT, Meta wins regardless of where the session starts.

Neither response reads as complacent. Both read as incumbents who have done the math on what a $100M ARR pilot in six weeks implies at scale.

The Privacy Variable Marketers Should Watch

OpenAI updated its U.S. privacy policy on April 30, 2026, to formally disclose that it receives purchase data from advertisers and shares data with marketing partners. Marketing cookies are enabled by default for free users. Cookie identifiers and device identifiers are shared with marketing partners for measurement. Email addresses may be sent to ad platforms for conversion matching.

OpenAI's stated position is that conversations remain private from advertisers and that conversational data is not sold. But regulators are not waiting. Canadian regulators have flagged OpenAI's data practices as non-compliant. The gap between OpenAI's intent and third-party legal interpretation of its data practices is a live variable in every market outside the U.S.

For marketers in regulated categories, this is not a hypothetical risk. If regulatory action restricts OpenAI's ability to use conversational context for ad targeting, the platform's primary value proposition weakens significantly. You would be buying reach without the intent signal that justifies the premium CPM.

Should You Test Now?

What to Do This Quarter

If you decide to test, the mechanics are straightforward. The Ads Manager is live at ads.openai.com. Verification takes a few business days. There is no minimum spend. CPM and CPC both available from day one.

Run a small awareness-frame budget and treat measurement independently. Use your own UTM parameters. Run post-purchase surveys asking customers how they heard about you. Do not rely solely on OpenAI's reporting dashboard. The CAPI integration is useful, but you need triangulation until third-party measurement arrives.

The creative constraint is real. The sponsored product card format is small. What wins is a clear product name, a specific value proposition, and a landing page that continues the conversation rather than interrupting it. Thinking about the person who is 70% through a buying decision and needs one clean answer is a better creative brief than standard display thinking.

Watch the Q3 and Q4 roadmap. CPA bidding and third-party measurement are the two unlocks that will move serious performance budgets onto this platform. When both arrive, the brands that already know the creative and conversion mechanics will have a measurable head start.

The channel is real. The intent signal is real. The measurement gap is also real