By Kimeko McCoy • February 17, 2026 •
Subscribe: Apple Podcasts • Spotify
The other shoe has finally dropped. After months of speculation, OpenAI officially began to test ads in ChatGPT in the U.S. last Monday.
Here’s what we know so far:
- Ads appear as sponsored placements at the bottom of answers on free and Go tiers of ChatGPT.
- The tech behemoth is mainly approaching “the biggest of the big” brands as opposed to agencies.
- OpenAI is asking some marketers to commit a share of their total search budgets — a $200,000 minimum commitment in some cases.
- CPMs are said to be $60, on par with streaming and premium TV inventory.
Experiments are already underway with brands like Williams Sonoma, Bed Bath and Beyond, and others via major agency holding companies. Seemingly, OpenAI has learned from their competitor’s missteps — namely Perplexity, which put a pin it its ad business last fall.
“They’re clearly trying to see themselves as very separate to any of the other platforms out there,” said Krystal Scanlon, Digiday senior platforms reporter, who joined the Digiday Podcast to discuss OpenAI’s ad plans. “They’re not necessarily saying that other platforms have failed, but they’ve obviously come with different learnings as to what has worked and what hasn’t.”
Meanwhile, marketers are still trying to read the tea leaves.
What’s less clear is how much insight advertisers will have beyond views and clicks to justify premium CPM pricing. There’s also OpenAI’s small, still-forming ads organization and what that says about monetization efforts going forward — especially as the landscape gets more competitive between the tech platforms.
Scanlon joined the Digiday Podcast to help co-hosts Tim Peterson and Kimeko McCoy make sense of OpenAI’s road map.
Here are a few highlights from the conversation, which have been edited for length and clarity.
Controlling the narrative
Even in the guidelines, OpenAI was very clear in the sense that if [ad pilot partners are] going to say anything publicly about this pilot — number one: you make sure that you only talk about it as a pilot. It is not a launch product. They were very clear about that. And then the ultimate thing was: any public messaging, we have to approve it before it goes out, and you can only do yours after we’ve actually said publicly that we’re doing ads.
So it’s OpenAI controlling the narrative, which makes sense that they would want to, because they already know bringing ads in is already going to be a risky business for them in terms of whether or not it’s going to alienate users.
Paying for clicks vs. impressions
In terms of why they’ve chosen CPM, anyone that I’d spoken to before literally was saying for a search-type model, you’ve always expected it to be a CPC (cost-per-click), but CPM is actually the model that is leaned on if you’re doing brand sponsorships and things like that — which is obviously where they’re gearing this. Ironically, albeit, that they are still trying to go for search budgets. It’s probably just the most succinct, easiest way to at least test something to see if it’s worth their time, how it’s going to work and if users are going to be OK with it, before starting to delve into more formats that might be more complex.
Justifying $60 CPMs
Apparently the starting CPMs, because we’re going with CPMs for this model, is $60. Anyone I’ve spoken to [says] that’s putting it in the same region as premium TV. We already saw when Perplexity tried to come out with high prices, no one liked it. Even when Netflix first did it, when they had their ad product, no one liked it and they had to lower theirs. We know what bucket that they want to be considered as: this is a very separate model and very different from ones already out there.
