TL;DR: Test OpenAI/ChatGPT ads only if you have money to burn. The platform is too limited and the audience too narrow to justify real budget, but if you want to be an early mover, now’s your window.
OpenAI launched ads inside ChatGPT on February 9, 2026, and the marketing world collectively leaned forward. After years of Sam Altman calling advertising a “last resort” for the platform, he once described the idea of ads mixed with AI as “uniquely unsettling”; the company flipped the switch anyway. The reason isn’t hard to figure out. OpenAI signed over $1.4 trillion in infrastructure contracts in 2025 and is targeting $20 billion in annual revenue. The subscription model alone isn’t going to get them there. So here we are.
If you’re an ecommerce brand or digital marketer trying to figure out whether to allocate budget here, I’ve spent time inside the ads manager. My honest take: it’s a great way to burn money right now. But if you’re the type who likes planting flags early, there’s an argument for testing it.
What the Platform Actually Looks Like
The self-serve ads manager opened to all U.S. advertisers in early May with no minimum spend requirement. Before that, entry required a $200,000 minimum commitment, so the fact that it’s now accessible to any budget is genuinely new. That said, accessible doesn’t mean sophisticated.
Right now you can optimize for clicks or reach. That’s the whole menu. Conversions are still labeled “coming soon,” though CPA bidding is rolling out via conversion-optimized campaigns in early June 2026, so that’s at least moving. Instead of targeting keywords, you’re targeting conversations, which is a conceptually interesting idea. In practice, you can add context hints to help the system understand when your ad is relevant, but there’s no audience segmentation, no theme-level targeting, and no meaningful reporting.
You’re flying mostly blind, which sucks as a data-driven performance marketer
Bidding is its own kind of strange. The recommended CPC bid is $3–$5. Drop to $2.99, and the platform warns your ad “may not deliver.” That’s a hard floor dressed up as a recommendation, not an actual bid strategy. And even when your ad wins the auction, it runs at the very bottom of the response… clearly labeled “Sponsored” and separated from the actual answer.
On CPM pricing, ChatGPT’s CPM runs $25–$60, which even after a recent 50% price drop is still 3x Meta and 5x TikTok. You’re paying a premium for an audience you can barely define and can’t retarget. Yeah… no thanks.
The Audience Problem Is the Real Issue
Here’s what should give most advertisers pause before committing real budget: ads only show to users on the free tier and the newly launched ChatGPT Go plan at $8/month. Plus, Pro, and Enterprise subscribers see no ads at all.
Think about who uses the free tier. For consumer ecommerce brands selling broadly, that population might be fine. But if your buyers are business owners, marketers, or anyone with purchasing authority, that crowd is overwhelmingly on paid plans. They’ll never see your ad. For B2B or higher-consideration purchases, you’re essentially advertising to an audience that doesn’t match your ICP, with no tools to fix it. Honestly, I would never put a single dollar of my own money into this kind of advertising for Stryde. It just doesn’t make sense.
There’s also the click-through problem. Early advertisers have flagged lower click-through rates as one of the key performance challenges on the platform. When ads sit at the bottom of a completed AI response, users have already gotten what they came for. The motivation to keep engaging, let alone click an ad, is low. Again… no thanks.
So Why Would Anyone Test It?
Because the scale is real, even if the tools aren’t. ChatGPT now reaches approximately 920 million weekly active users in the U.S. alone. That’s not a niche channel; it’s one of the highest-traffic destinations on the internet. And AI-powered advertising is projected to grow from $35 billion (about 8% of U.S. ad revenue) in 2025 to $142 billion by 2030. The money is coming. The question is just whether it’s here yet for most advertisers.
The platform is also maturing faster than you might expect. OpenAI is expanding the pilot to the UK, Mexico, Brazil, Japan, and South Korea in the coming weeks, and the product roadmap is clearly accelerating. When financial pressure is this intense, ad products tend to improve quickly.
Being early on a platform that eventually scales has real value: you learn the system before your competitors do, you help inform what the product becomes, and you’re positioned to move fast when the tools actually catch up to the audience size.
The Verdict
For clients, I’m not recommending budget here yet. If the choice is ChatGPT ads or more spend into Google and Meta, Google and Meta win today and it’s not close. The targeting is better, the reporting is real, and you know who you’re reaching.
But if you have a small budget you can genuinely afford to lose, call it a learning budget, not a performance budget; there’s an argument for getting your hands dirty now. Go in with zero ROI expectations, treat it as education, and pay attention to what changes over the next six months. The ad product will mature. The audience targeting will improve. Conversions are already on the roadmap.
The window to be an early mover is open. Just be honest with yourself about what “early” actually means: you’re not buying results yet. You’re buying a front-row seat to what might become a significant channel. Whether that’s worth it depends entirely on how much you value being first.
Greg is the founder and CEO of Stryde and a seasoned digital marketer who has worked with thousands of businesses, large and small, to generate more revenue via online marketing strategy and execution. Greg has written hundreds of blog posts as well as spoken at many events about online marketing strategy. You can follow Greg on Twitter and connect with him on LinkedIn.