ChatGPT now sells ads. $100 million in 6 weeks. And your conversations are the product.

· 13 min read

I want to walk through something that happened in the last eight weeks, because I think most people are underestimating how fast it moved.

On February 9, 2026, OpenAI turned on advertising inside ChatGPT. Sponsored placements, sitting below AI-generated responses, clearly labeled, visually separated. Fairly standard, if you squint. The kind of thing Google has been doing for two decades.

Six weeks later, those ads crossed $100 million in annualized revenue. More than 600 advertisers. A $60 CPM. And then, on April 1, OpenAI announced something much more interesting than banner placements: conversational ads. Ads you can talk to.

This is the moment ChatGPT stopped being a tool and started becoming a marketplace.

What actually happened on April 1

OpenAI signed Smartly, the creative adtech company, as its first partner for building interactive ad units inside ChatGPT. The partnership announcement was straightforward, but what they’re building is not.

The new ad format lets users click a sponsored placement and enter a chatbot-style dialogue with a brand. You’re already talking to ChatGPT. Now you can also talk to Nike inside ChatGPT. Or Sephora. Or your airline. The ad becomes a second conversation that can answer product questions and guide you toward a purchase.

This is different from every ad format that came before it. A Google search ad hopes you click. A Facebook ad hopes you pause your scroll. A ChatGPT conversational ad hopes you keep talking, and it has the context of everything you’ve already said to the AI to make that conversation feel relevant.

Smartly’s role is performance optimization: adjusting creative and targeting in real time based on how users interact with these new formats. The initial rollout targets entertainment, retail, and sports brands. Self-serve tools, which would remove the current $200,000 minimum commitment and open the platform to anyone with a credit card, are scheduled for this month.

The targeting is the product

I keep coming back to the targeting, because it changes the economics of online advertising in a way that I don’t think the marketing industry has fully internalized yet.

ChatGPT’s ad matching draws on three signals: the topic of your current conversation, your past chat history, and your previous interactions with ads. Think about what that means in practice.

When you search Google for “best running shoes,” Google knows you want running shoes. When you ask ChatGPT to help you train for your first marathon, plan your nutrition, build a weekly schedule, and then ask about gear, ChatGPT knows you’re a beginning runner who’s training for an endurance event, what your weekly mileage is, whether you have knee problems, and how much time you have before race day. The ad that shows up isn’t guessing. It has context that no search query could ever provide.

Criteo, OpenAI’s other adtech partner, published numbers on this. Users referred from ChatGPT and similar LLM platforms convert at 1.5 times the rate of users from other referral channels. That number comes from a sample of 500 US retailers tracked through February 2026. A 50% conversion lift is the kind of thing that makes CMOs rearrange their budgets.

And here’s the part that should make you uncomfortable: the system can also receive information from advertisers about purchases you make after clicking. Conversion data flows back to OpenAI. Your conversation context flows to the advertiser’s chatbot. The feedback loop closes.

$100 million in six weeks, with a $200,000 minimum

I want to sit with those numbers for a second. ChatGPT launched ads on February 9. By late March, annualized revenue was over $100 million. The platform had 600+ advertisers, all paying at least $200,000 to participate in a managed pilot. No self-serve. No small business access. Just large brands working through agencies like Omnicom, WPP, and Dentsu.

For context, Google Search took several years to reach $100 million in annual ad revenue after launching AdWords in 2000. Facebook took about two years after introducing News Feed ads. ChatGPT did it in six weeks, with a waitlist and a $200K floor.

The obvious objection: ChatGPT has 920 million weekly active users. Of course the numbers are big. But the speed says something about demand on the advertiser side. Brands were so eager to get into conversational AI that they committed six-figure budgets to a pilot with no self-serve tools and limited targeting controls.

When the self-serve tools launch this month, the $200,000 minimum disappears. Any business can buy ChatGPT ads. If you run a Shopify store, a SaaS product, a consulting practice, you’ll be able to advertise inside AI conversations. That’s a different world than managed enterprise campaigns.

The three-way split in AI search advertising

What I find most interesting about this moment is that the three major AI platforms have chosen completely different paths, and we’ll find out within a year which one was right.

OpenAI went aggressive. Ads for free users, conversational formats, self-serve tools, international expansion into Canada, Australia, and New Zealand. They told investors to expect $17 billion in consumer revenue for 2026, with advertising as a growing share. Premium tiers stay ad-free, creating clear “pay to escape” incentives.

Google went even more aggressive. According to recent data, Google is weaving ads into 25.5% of AI-generated results in its AI Mode. They’ve also launched shopping ads with “Direct Offers” inside AI search, monetizing AI responses at roughly the same rate as traditional search ads. For Google, AI is just a new surface for the same business model.

Perplexity went the other direction entirely. In February 2026, Perplexity abandoned advertising after testing sponsored follow-up questions through 2024 and 2025. A Perplexity executive told the Financial Times: “The challenge with ads is that a user would just start doubting everything.” The company is now targeting $500 million in annualized subscription revenue and positioning itself as the ad-free alternative.

An Ipsos survey found that nearly two-thirds of US adults say ads in AI search results make them trust the results less. Perplexity is betting that trust premium is worth more than ad revenue. OpenAI is betting the ad revenue is too large to leave on the table. Google is betting that users complain but don’t actually leave.

All three could be right, for their respective user bases. Or one of them is making a catastrophic miscalculation.

Why conversational ads are different from everything before

The reason I think the Smartly partnership matters more than the raw revenue numbers is the format itself. Every previous digital ad format was essentially a sign you walked past. Some were better designed, some were better targeted, but they were all variations of “here’s a message, hope you click.”

Conversational ads break that pattern. You don’t click and leave the AI. You click and stay, and the brand starts talking to you in the same interface you were already using to get advice, plan purchases, and solve problems. The brand has access to the context of what you were already doing. It can respond to your specific situation.

Imagine you’re asking ChatGPT to help you plan a trip to Japan. You’ve been going back and forth about itineraries, budget, timing. A conversational ad from an airline doesn’t just show you a fare. It enters the conversation knowing you’re planning two weeks in Tokyo and Kyoto in October, that your budget is around $3,000 for flights, and that you prefer window seats. It can offer something specific. It can negotiate.

This is the advertising equivalent of a knowledgeable salesperson walking up to you in a store while you’re already holding the product. Except the salesperson has read your diary.

What this means if you’re building a product

If you’re building an AI product, this changes the competitive picture in ways that aren’t immediately obvious.

The “AI wrapper” business model just got more complicated. If you built a product on top of ChatGPT’s API, your users might now encounter ads in the base product. OpenAI hasn’t said ads will appear in API responses (they currently don’t), but the incentive structure is obvious. 920 million free-tier users are a monetization surface that OpenAI is now actively exploiting. The API is a separate business, but watch this space.

Then there’s the positioning angle. “Ad-free AI” is becoming a real selling point. Perplexity is already doing it. Claude hasn’t introduced ads. If you’re building for professional users, “we will never put ads in your workflow” might matter more than any feature on your roadmap.

And if you’re building any kind of commerce product, ChatGPT ads are a new acquisition channel worth testing. The 1.5x conversion rate over other referral channels is hard to ignore. When self-serve launches, the cost of testing drops from $200,000 to whatever CPM-based spending looks like for your category. E-commerce, DTC, and SaaS founders should be paying attention.

What this means if you’re a marketer

The self-serve launch changes the math for marketing teams of every size.

The $60 CPM is expensive compared to most display advertising. Google Display Network averages around $2-5 CPM. Facebook is roughly $8-12. But if ChatGPT’s conversion rates really are 1.5x higher, the effective cost per acquisition might be lower despite the higher CPM. The intent signal in a conversation is richer than anything keyword targeting can provide.

The conversational format also creates something no other ad platform offers: a back-and-forth relationship before any money changes hands. Users can ask the brand’s chatbot questions and get personalized recommendations without ever leaving the AI interface. That’s closer to a good sales call than anything digital advertising has produced before.

The practical question is whether users will actually engage with conversational ads or treat them the way they treat every other ad format (by ignoring them). Banner blindness is real, and “conversational banner blindness” might develop just as quickly. The early Criteo data is encouraging, but it’s from a small sample during a novelty period.

The privacy problem nobody is solving

I’ve been writing about ChatGPT ads for 800 words now without addressing the thing that makes me most uneasy, so let me do that.

OpenAI’s privacy policy for free and Go tier users states that personalized ads will be shown “subject to your settings.” The system can use your conversation history for targeting. Advertisers can send conversion data back to OpenAI.

People share things with AI assistants that they wouldn’t type into a search bar. Medical questions. Relationship problems. Financial anxiety. Career doubts. The intimacy of a conversation with an AI is categorically different from the transactional nature of a search query. Targeting ads based on that conversational context feels different from targeting based on keywords, even if the technical mechanism is similar.

Google and Meta both started with strong privacy principles and gradually weakened them as revenue pressures grew. OpenAI is starting with a $200,000 minimum and managed campaigns. But when self-serve tools arrive and revenue targets tighten, the incentive to expand what data gets used for targeting will only grow.

Users have already pushed back. OpenAI pulled aspects of a contact syncing feature after backlash following a policy update. The comparison to “Google 2.0” is showing up in user forums and social media. Some users are openly talking about switching to Claude or Gemini specifically to avoid ads.

The Ipsos data is worth repeating: nearly two-thirds of US adults say ads in AI results make them trust the results less. When the “result” is a conversation that feels personal and private, the trust damage could be even higher than in traditional search.

The $17 billion question

OpenAI told investors to expect $17 billion in ChatGPT consumer revenue for 2026. That includes subscriptions and advertising. The company is still burning through cash; its compute costs are enormous, and the $122 billion it raised in Q1 2026 came with expectations of revenue growth to match.

The advertising business is a direct response to that pressure. Free users are expensive to serve. They use GPU compute, they generate support load, and until February 2026, they generated zero revenue. Advertising turns the free tier from a cost center into a revenue engine. The faster it grows, the longer OpenAI can sustain its current burn rate.

But there’s a tension. The thing that makes ChatGPT valuable is trust. Users share sensitive information because they believe the AI is working for them, not for an advertiser. Every ad in the conversation chips at that perception. If enough users start self-censoring their queries because they know the conversation is being used for ad targeting, the product gets worse. And a worse product means less engagement, which means less ad revenue.

Google navigated this tension for 20 years and mostly got away with it, because search queries were already public-ish in nature. You type something, you get results, some are paid, most are organic. The contract is clear. But AI conversations are private-ish in nature. You’re sharing context, emotions, half-formed thoughts. The contract is murkier, and breaking it feels like more of a betrayal.

What I’d do with this information

If you’re a founder building an AI product: decide now whether you’re an ad-supported business or a subscription business. The middle ground is going to get uncomfortable fast. Perplexity chose. OpenAI chose. Not choosing is also a choice, and it’s usually the worst one.

If you’re a marketer with budget to experiment: get ready for the self-serve launch this month. The early movers on ChatGPT ads will have the advantage of lower competition and novelty-driven engagement. Test conversational formats when they become available. The 1.5x conversion rate won’t last forever, but the first-mover window is real.

If you’re a user who cares about privacy: go to your ChatGPT settings and check what data sharing options are available. Consider whether the free tier is still worth it, or whether paying $20/month for ad-free access is a better trade. If neither option works for you, Claude and Perplexity don’t show ads today. Whether that stays true depends on their economics over the next year.

If you’re watching the AI industry broadly: this is the most important strategic divergence since the open-source vs. closed-source debate. The three biggest AI platforms have placed radically different bets on how to make money. Within 12 months, we’ll know which bet was right. The answer will shape how AI products are built and funded for the next decade.

The advertising market for AI search is estimated at $15-25 billion in 2026 and growing at 35-50% annually. That’s too large for OpenAI to ignore and too risky for Perplexity to chase. The question is whether users will treat AI ads the way they treat search ads (tolerable) or the way they treat pop-ups (unforgivable).

Six weeks, $100 million. The answer is coming fast.

Frequently asked questions

How much does it cost to advertise on ChatGPT?

ChatGPT advertising uses impression-based CPM pricing at approximately $60 per 1,000 impressions. The current minimum commitment is $200,000, but OpenAI is launching self-serve tools in April 2026 that will remove this minimum and open the platform to businesses of all sizes.

How does ChatGPT target ads to users?

ChatGPT’s ad targeting uses three signals: the topic of your current conversation, your past chat history, and your previous interactions with ads. This gives advertisers access to intent data beyond what keyword targeting provides, because users share detailed context, preferences, and plans in their AI conversations.

What are conversational ads in ChatGPT?

Conversational ads are a new format announced April 1, 2026 through a partnership between OpenAI and Smartly. Unlike static sponsored placements, conversational ads let users click and enter a chatbot-style dialogue with a brand directly inside ChatGPT. The brand’s chatbot can offer tailored suggestions based on the context of your conversation.

Does ChatGPT Plus show ads?

No. ChatGPT Plus at $20/month and ChatGPT Pro at $200/month are completely ad-free. Advertising only targets free-tier and Go-tier users, which represent approximately 95% of ChatGPT’s 920 million weekly active users.

How much ad revenue has ChatGPT generated?

ChatGPT crossed $100 million in annualized advertising revenue within six weeks of launching ads on February 9, 2026, with more than 600 advertisers. OpenAI told investors it expects total consumer revenue to exceed $17 billion in 2026, with advertising as a growing portion.

Is Perplexity AI showing ads?

No. Perplexity abandoned advertising entirely in February 2026 after testing sponsored follow-up questions through 2024-2025. The company cited user trust concerns and is instead pursuing $500 million in annualized subscription revenue, positioning itself as the ad-free AI search alternative.