OpenAI Launches Self-Serve ChatGPT Ads Manager With CPC Bidding and Pixel Tracking

Photo by Jo Lin on Unsplash

OpenAI opened the ChatGPT ads manager to U.S. advertisers on May 5, 2026, killing the $50,000 minimum spend that locked the pilot to holding companies and replacing CPM-only buying with cost-per-click bidding. Any verified U.S. business can now register, fund a budget, upload creative, and run a campaign inside ChatGPT without booking an agency. The launch also introduces a pixel and a Conversions API so advertisers can finally track what happens after a click.

The pivot from a tightly managed pilot to a self-serve platform is the cleanest signal yet that OpenAI is treating advertising as a permanent revenue line, not an experiment.

The $50,000 Floor Just Disappeared

Until this week, testing ChatGPT inventory required a $50,000 commitment, a gate that priced out every SMB and steered budgets into the holding companies that had early integration access. Axios reported that the threshold has been removed entirely with the self-serve beta, and OpenAI's monetization head Asad Awan confirmed the rollout to U.S. advertisers on the same day.

That single change reframes the audience. OpenAI's own announcement describes the portal as built for "companies of all sizes, from SMBs and startups to global brands," with the full self-service loop in one place: register, add payment, set budgets, bids and pacing, upload ads, launch campaigns, and view performance.

Access is not instant. Vizup's onboarding writeup notes the Ads Manager is rolling out as a beta where advertisers register interest at ads.openai.com and then wait for approval, with business verification heavier than a typical social platform, requiring legal name, address, and tax documentation. Teams have reported sitting in the queue for weeks.

The pilot itself only began publicly in February 2026, with a Reuters report cited by Axios suggesting OpenAI had banked more than $100 million in U.S. ad revenue after just six weeks. That number is what makes the $50,000 floor disappear so quickly. Demand is already there.

How to Advertise on ChatGPT Through the New Ads Manager

Here is the practical sequence for new advertisers, drawn from OpenAI's documentation and reporting around the launch:

1. Join the waitlist at the OpenAI for advertisers page and submit business details.
2. Complete business verification with legal name, address, and tax information.
3. Once approved, add billing and set user access in the dashboard.
4. Choose a bidding model. OpenAI confirmed both CPM and cost-per-click bidding are supported.
5. Upload creative, set budget and pacing, and define the contextual targeting.
6. Install the pixel or Conversions API to attribute downstream actions.

Advertisers who want to keep existing workflows can still buy through agency holding partners Dentsu, Omnicom, Publicis, and WPP, or through ad-tech and measurement partners including Adobe, Criteo, Kargo, Pacvue, and StackAdapt, per OpenAI's partner list.

ChatGPT CPC Bidding and the Conversions API Change the Math

The original February 2026 pilot ran on CPM only. That format works for awareness, but a conversational interface where users are mid-decision produces far stronger signals than impressions. ChatGPT CPC bidding lets advertisers pay only when a user actually clicks, which is the unit performance marketers actually budget against.

Measurement was the other gap. Search Engine Land's coverage describes it bluntly: without conversion tracking, advertisers struggled to justify spend. OpenAI is closing that with a pixel and a Conversions API that captures purchases, leads, and sign-ups after an ad interaction.

There is a meaningful catch on the pixel. Jellyfish CSO Jai Amin told Digiday that OpenAI controls pixel creation itself, and advertisers cannot self-implement, which limits flexibility for teams that prefer to wire measurement into their own tag managers. For larger advertisers running custom attribution, that dependency is a real constraint.

Debra Aho Williamson, founder and chief analyst at Sonata Insights, framed the broader move to Digiday as OpenAI demonstrating it understands the basic building blocks advertisers need to feel comfortable testing on ChatGPT. The translation: this is table stakes catch-up work, not an innovation. Meta, Google, and TikTok shipped this stack years ago.

The Creative Format Is Still a Favicon and a Line of Text

The ad unit itself is the most under-discussed limitation of the launch. The current creative is a small favicon paired with accompanying text, a constraint flagged by Digiday as a key restriction for brand advertisers. There is no video unit. No carousel. No interactive product card. No native commerce checkout.

For performance marketers chasing direct response, a favicon and a sentence may be enough when it lands inside a buying-intent conversation. For brand budgets accustomed to YouTube CTV or Meta Reels, the format offers almost nothing to work with. That asymmetry will shape who actually moves spend into the platform first, and it explains why OpenAI is leaning hard into CPC and conversion tracking rather than brand-lift studies.

The second constraint sits on category eligibility. OpenAI's policy restricts the beta to a limited set of low-risk verticals: household and consumer goods, local services, travel and entertainment, digital products, and education. The company says categories will expand as safeguards, review systems, and compliance infrastructure mature. Regulated sectors like finance, pharma, and crypto are not part of the launch, which closes off some of the highest-CPC categories in traditional search.

A detail that has gone largely unmentioned in the launch coverage: ChatGPT's data sharing inverts the Google Search model. Advertisers receive only aggregated performance data, with individual conversations, memories, and personal details never exposed. This is the opposite of Google Search Ads, which expose exact query-level data and let buyers optimize against specific search terms. ChatGPT advertisers will optimize against intent clusters they cannot see, which is a measurement and creative discipline very few performance teams have built muscle for.

The $2.5 Billion Target and the $100 Billion Story

Axios reports that OpenAI is targeting $2.5 billion in ad revenue for full-year 2026 and $100 billion annually by 2030, both company-stated goals. The same piece notes that AI search advertising is still relatively nascent and is not expected to overtake traditional search advertising in the near term, a direct challenge to the long-term narrative.

The near-term number is achievable. The far-term one requires a category that does not yet exist at that scale. Google has reported that ads placed inside AI Overviews monetize at roughly the same rate as traditional Google Search ads, as Maas notes in its launch breakdown, which gives OpenAI a benchmark to chase but also a much larger competitor occupying the same buying intent.

The competitive context for the OpenAI advertising platform is thinner than it looks. Anthropic has explicitly chosen not to run ads on Claude. Perplexity launched sponsored follow-up questions in November 2024 with brands including Whole Foods, paused new advertisers by October 2025, and killed the program entirely in February 2026, per Maas, before pivoting to a subscription-only model. That leaves ChatGPT and Google AI Overviews as the only two AI ad ecosystems available to performance marketers in 2026.

Perplexity's collapse is the cautionary tale OpenAI is trying to outrun. A clean ad product is not enough on its own. Volume, retention, and a creative format that brands can actually use are what compound.

What OpenAI Says About Answer Independence

The most defensible claim in the announcement is structural. OpenAI states that ads "do not influence the core organic model" and that delivery decisions run through a system physically separate from the model generating ChatGPT answers. Aggregated performance data is the only thing advertisers receive.

Whether users believe that separation over time is the question that determines whether ChatGPT's ad business reaches the $100 billion projection or stalls at a few billion. Google spent two decades insulating organic results from paid placements and still faces regulatory pressure on the boundary. OpenAI is starting that work from zero, inside a product where the line between answer and recommendation is harder to see than a blue link versus a sponsored result.

The pilot is also moving abroad. Digiday confirmed expansion to the U.K., Mexico, Japan, Brazil, and South Korea alongside the U.S. self-serve rollout, which broadens the test surface significantly. Advertisers outside the U.S. should expect inventory in their markets before the end of 2026, even if self-serve buying takes longer to reach them.

The specific thing to watch: whether OpenAI publishes its first auction-level performance data, average CPCs by category, and conversion-rate benchmarks before the end of Q3 2026. Without that, the $2.5 billion target stays a press release number rather than a verifiable trajectory. The first earnings-style disclosure will decide whether the next wave of advertisers commits real budget or keeps treating ChatGPT as a test line. For more on how AI infrastructure costs are shaping these monetization decisions, see our analysis of why Big Tech is redirecting capital into GPUs rather than headcount, and follow ongoing coverage in our AI Industry articles and News articles sections.