Ads in ChatGPT: OpenAI Finally Hits the Gas. What It Means for You

For two months, OpenAI tested ads in ChatGPT on a handful of users while major agencies nervously watched their budgets sit idle. Then — in the span of a single week — the company announced a partnership with Criteo and confirmed to Reuters that ads would roll out to all free users in the US. Here’s what happened, what the first numbers look like, and why this matters even if you don’t sell to Americans.

Two Months of Piloting: Why Agencies Were Frustrated

It all started on February 9, 2026. OpenAI launched an ad pilot in ChatGPT for users on the Free and the new Go tier ($8/month) in the United States. Three of the world’s largest advertising holding companies — WPP, Omnicom, and Dentsu — got seats in the program. The minimum buy-in was roughly $200,000 — double what’s typically required for alpha tests of new ad formats.

The agencies paid up. And then they waited.

According to Sensor Tower, by mid-March ads had reached only about 5% of ChatGPT’s mobile users. At the start of the month, it was just 1%. Yes, the number of impressions grew roughly 600% in two weeks — but that’s 600% of a very small number. For holding companies that had reserved six-figure sums from their Q1 budgets, the situation was uncomfortable: money allocated, pilot ending in late March, and physically nowhere to spend it.

OpenAI responded calmly: the slow pace was intentional. The goal was to first understand how ads affect user experience before scaling. Meredith Spitz, EVP and Head of Paid Search at Dentsu, confirmed that delivery volumes were building week over week, and that the holding company had gone in with realistic expectations, drawing from a pool dedicated to testing and innovation.

Industry insiders acknowledged that the caution was probably a good sign rather than a bad one. But when you have $200K earmarked for a specific test and the test is barely breathing, enthusiasm turns to frustration fast.

Then Everything Accelerated

On March 21, Reuters published the news: OpenAI confirmed that in the coming weeks, it would show ads to all Free and Go users in the US. The original source was The Information; an OpenAI spokesperson confirmed on the record.

So the pilot formally wrapping up at the end of March wasn’t a wind-down — it was a transition to scale.

What stays the same: ads appear below ChatGPT’s responses, clearly labeled as “Sponsored,” don’t influence model outputs, aren’t shown to minors, and don’t appear near sensitive topics (health, mental health, politics). Plus, Pro, Business, and Enterprise tiers remain ad-free.

Criteo: The First AdTech Partner, and a Game Changer

Between “slow pilot” and “ads for everyone,” there’s an important link: on March 2, Criteo announced it had become the first AdTech company integrated into ChatGPT’s advertising infrastructure.

Why does this matter? Criteo operates a network of 17,000+ advertisers, processes $4B+ in annual ad spend, and offers the established buying workflows that any media buyer knows. Where previously you needed a direct relationship with OpenAI and a $200K minimum, Criteo brings the entry point down to $50,000–$100,000.

In effect, ChatGPT’s ad inventory is starting to go programmatic. This is no longer an exclusive club for the chosen few — it’s the beginning of a standard advertising channel.

First Numbers: $60 CPM, 1.5x Conversion, $30B Forecast

Now for the specifics you’re actually here for.

CPM ~$60. That’s roughly three times Meta’s rates and comparable to premium TV inventory like NFL games or Netflix’s launch CPM. OpenAI positions this as premium placement: ads are based not on browsing history but on current intent — what the user is asking right now.

1.5x conversion. Aggregated Criteo data across 500 US retailers (February 2026) showed that traffic from LLM platforms like ChatGPT converts at approximately one and a half times the rate of other referral channels. The sample is limited and the observation period short, but the direction is telling: people coming from an AI chat are already deeper in the decision funnel.

Minimal analytics. Advertisers receive only aggregated data — total views and clicks. No conversion tracking, no demographic breakdowns, no purchase data. For anyone used to Google’s or Meta’s analytical depth, this is a painful trade-off.

Forecasts. Truist analysts called 2026 an “inflection year” for LLM advertising. Their estimate: under $1B in OpenAI ad revenue in 2026, but over $30B by 2030. Evercore ISI analyst Mark Mahaney published a similar figure — $25B annually by 2030. If these projections hold, advertising alone would double OpenAI’s entire business.

Context: Why OpenAI Went Into Ads in the First Place

The short answer: subscriptions aren’t enough.

Out of ChatGPT’s ~800 million weekly active users, only about 5% pay for a subscription. The remaining 95% are free. Meanwhile, OpenAI has committed to spending over $1.4 trillion on data center infrastructure through the early 2030s, with a projected cash deficit of $14–17 billion in 2026 alone. The math is simple: when you have 760 million free users and trillion-dollar obligations, advertising isn’t a whim — it’s a necessity.

The irony? As recently as May 2024, Sam Altman called the idea of combining AI with advertising “uniquely unsettling” and described ads as a “last resort.” By February 2026, he launched the pilot. Financial reality has a way of being persuasive.

Anthropic vs OpenAI: A Business Model War at the Super Bowl

It would be dishonest to talk about ChatGPT ads without mentioning the elephant in the room. On February 4, 2026, Anthropic — the company behind Claude and OpenAI’s primary competitor — dropped a series of Super Bowl LX commercials with the tagline: “Ads are coming to AI. But not to Claude.”

The spots were openly sarcastic: in one, a user asks an AI therapist about their relationship with their mother, and mid-response the “therapist” pivots to pitching a dating app for meeting older women. Anthropic wasn’t just joking — it was positioning itself as the premium alternative for anyone who doesn’t want ads in their AI conversations.

Altman fired back with a lengthy post on X, calling Anthropic’s ads “funny but clearly dishonest,” and counterattacked: Anthropic “serves an expensive product to rich people,” while OpenAI wants to make AI accessible to billions. According to BNP Paribas data, Anthropic’s campaign worked: Claude’s daily active users jumped 11% post-Super Bowl.

Marketing professor Scott Galloway called Altman’s response a strategic blunder and compared Anthropic’s ad to Apple’s legendary “1984” spot — because a market leader should never publicly legitimize a challenger with a lengthy rebuttal.

For us as marketers, the important thing isn’t who’s right — it’s what’s happening: two fundamentally different business models are crystallizing in the AI industry. OpenAI is going hybrid — subscriptions plus ads, following the Google/Meta playbook. Anthropic is betting on pure subscriptions and enterprise contracts, closer to Spotify Premium. This will shape not just the product experience but which new advertising channels end up in your media mix over the next few years.

What to Do With This: A Quick Summary for Decision-Makers

If you’re targeting the US market — it’s time to watch closely, but don’t rush with budgets. The first pilot results are frank: impressions are low, analytics are limited, CPM is high. The 1.5x conversion rate is an encouraging signal, but it’s Criteo data from 500 retailers over a single month — not an established pattern. If you have budget for tests ($50–100K through Criteo), this could be a worthwhile early experiment. But putting ChatGPT into your Q2 media plan as a full-fledged channel is premature. Wait at least until summer, when data from the full audience rollout becomes available.

If your audience is outside the US — for now, this is your watchtower, not your battlefield. ChatGPT ads are available only in the United States, and OpenAI has confirmed that a global expansion is not currently planned. But “not currently” is the key phrase. LLM advertising is a new channel being built right now. And when (not “if” but “when”) ChatGPT opens targeting to other markets, the winners will be those who already understand the mechanics: CPM-based pricing, intent-based contextual targeting, limited analytics. Follow the case studies from US advertisers — that’s your homework for the next 6–12 months.

What you should definitely do right now — think about your brand’s presence in LLM responses. Regardless of paid ads, ChatGPT already recommends products organically. Optimizing for AI search (Answer Engine Optimization) is something you can and should be doing today, without waiting for ad inventory to open up.

The Bottom Line

March 2026 is the month when ChatGPT’s ad business stopped being an experiment and started becoming a channel. The slow pilot, agency frustration, the Criteo partnership, the confirmed rollout to all free users — these aren’t separate news stories, they’re stages of a single process.

The forecasts are ambitious ($30B+ by 2030), the early signals are encouraging (1.5x conversion), but there are more open questions than answers. Will user experience survive the scale-up? Will proper analytics emerge? Will users flee to Claude or Gemini?

We’ll be watching and keeping you posted. In the meantime — study the fundamentals, follow the early advertiser case studies, and prepare for AI advertising to become yet another pillar of the digital media mix alongside Search, Social, and Retail Media. The only question is how fast.

---

Ukrainian version of this article.