OpenAI Opens ChatGPT Ads Publicly With Best Buy and Lowe's

OpenAI's "Advertise in ChatGPT" portal opened publicly this week, and the honest way to read the news is that a chatbot most CMOs had filed under "wait and see" is now a live self-serve performance channel with named retail case studies on day one. Best Buy, Lowe's and VistaPrint were cited as early advertisers, with Best Buy's VP of Media Amy Adams framing it as reaching customers "in those moments" of research that increasingly happen inside a chat window rather than a search bar.

The mechanics are the part worth paying attention to. Ads render as labeled sponsored boxes at the bottom of AI responses rather than woven into the answer itself, and OpenAI's line is that "ads do not influence ChatGPT's answers." Targeting is described as based on conversational context signals rather than keywords, which is the substantive break from a decade of search advertising. Placement is restricted to Free and Go tier users on the adult side of an age check, with under-18 users excluded and Plus, Pro and Business subscribers staying ad-free. OpenAI has also disclosed category bans covering real-money gambling, betting, lotteries and cash-prize gaming.

Why this matters outside the ad-tech bubble: the earlier pilot, per coverage of the self-serve rollout, required a $50,000 minimum spend that gated participation. Removing that floor and putting the platform behind a self-serve portal collapses a distribution moat, and the same reporting has OpenAI past 600 advertisers and roughly $100 million in annualized revenue within about six weeks of the May beta. That is the trajectory Google's search team should be modeling more carefully than any model benchmark: the second search-shaped intent surface is now a real ad business with retail majors on the books.

The honest caveat is that a lot of the numbers a media planner actually wants are not in the reporting yet. There is no settled public floor on minimum spend, no full CPM or CTR disclosure beyond a low single-digit click-through figure floating in one outlet's coverage, and no detail on how brand safety is enforced inside a context that turns over sentence by sentence. Take the launch as a real product ship, not a settled auction.

The part I keep coming back to is that Sam Altman was on record in 2024 calling ads "sort of uniquely unsettling" as an AI business model. Shipping a self-serve buy anyway, in front of a reported 900-million-weekly-user product, is the story: the economics of that scale apparently do not leave you the luxury of principled reluctance, and the brands that show up first get to shape what a "sponsored" placement inside an AI answer looks like before the norms harden around them.