OpenAI is turning ChatGPT into a real ad channel, and the bigger question is how much search budget it can pull away from Google.
OpenAI's ad experiment is no longer just a theory about how AI companies might make money. It is now a live commercial test inside ChatGPT, aimed at logged-in users on the Free and Go tiers in the U.S., and advertisers are beginning to treat it as a new place to capture intent before a user ever reaches a search results page.
The company began testing ads in ChatGPT on February 9, 2026, after outlining its advertising principles in January. OpenAI says the placements are clearly labeled and may be matched to the topic of a conversation, past chats, and prior ad interactions, while advertisers do not get access to private conversations or control ChatGPT's answers. That distinction matters. If users believe ads are shaping the answer itself, trust becomes the product risk.
As Axios recently noted, OpenAI took the next step in early May by launching a beta self-service Ads Manager for U.S. advertisers, with agency partners including Dentsu, Omnicom, Publicis, and WPP. That move changes the story from a limited pilot for large brands into the beginning of a broader ad marketplace, even if the channel is still young and measurement is still catching up.
Advertisers are interested, but still careful
The early appeal is straightforward. ChatGPT sits close to the moment when a user is asking for help, comparing products, planning a purchase, or trying to solve a problem. That is the same kind of intent that made Google search advertising so valuable, but the interface is different. There are no ten blue links. There is a conversation, and a sponsored suggestion has to feel useful without making the whole answer feel compromised.
OpenAI has already added more advertiser infrastructure around the pilot, including buying options, reporting, and partner integrations. The company's help materials describe support for CPM and CPC buying, along with metrics such as impressions, clicks, spend, click-through rate, average CPC, average CPM, and conversions. That is basic plumbing, but basic plumbing is what turns curiosity into budget.
The early revenue signal is also real. Reuters reported in late March that OpenAI's U.S. ChatGPT ads pilot had passed a $100 million annualized revenue pace within six weeks of launch, citing a company spokesperson. That is tiny next to Google's advertising machine, but it is not trivial for a product that had only recently begun showing paid placements.
The more important point is not the size of the number today. It is the type of money being tested. If advertisers can buy against commercial intent inside an AI assistant, then search budgets no longer have to begin and end with a traditional results page. That gives media buyers a reason to experiment, and it gives OpenAI a path to revenue beyond subscriptions, enterprise contracts, and API usage.
Google is not suddenly exposed, but the pressure is real
Google still has the advantages that matter in advertising: scale, measurement, merchant relationships, agency muscle, and years of habit built into marketing teams. A startup founder or growth lead can launch a search campaign today and understand exactly what the system is supposed to do. ChatGPT ads are not there yet.
Still, Google's risk is not that every advertiser walks away at once. The risk is that a meaningful share of high-intent queries starts somewhere else. If people ask ChatGPT which CRM to buy, where to book a trip, what laptop fits a budget, or which tax tool handles a specific use case, the first commercial touchpoint may move away from Google Search.
That would not destroy Google's ad business, but it could change the auction economics around some of its most valuable categories. Search advertising works because buyers compete for intent at scale. If AI assistants create new inventory around the same intent, even slowly, brands will begin comparing cost, conversion quality, and trust across channels.
OpenAI also has to manage a delicate user problem. ChatGPT became popular because people felt they were getting direct help, not being routed through a marketplace. Ads that look too much like paid answers could undermine that. Ads that are clearly labeled, relevant, and kept separate from the response have a better chance of becoming part of the experience rather than a reason to leave it.
What founders and investors should watch
For founders, this is a practical marketing issue before it is a grand platform battle. If your company depends on discovery, the smart move is to test assistant-native ads early, track them tightly, and avoid assuming that search behavior will stay fixed. Early channels are messy, but they can teach you where acquisition costs may move next.
For investors, the signal is more strategic. Google's advertising moat remains large, but it is no longer a static fact. AI assistants are becoming commercial surfaces, and OpenAI is moving quickly to prove that those surfaces can support paid discovery without breaking user trust.
The next few months will show whether ChatGPT ads can scale beyond novelty. Watch attribution quality, advertiser retention, user reaction, and whether OpenAI expands beyond the current U.S. focus. If the channel proves it can deliver comparable return on ad spend, the old habit of treating Google search as the default home for intent-driven budgets will start to look less automatic.
Also read: CXMT's IPO could test how far China's memory chip ambitions have come • Byju Raveendran's Singapore jail term deepens the lesson on founder accountability • ByteDance is preparing a huge AI spending surge as China races for compute
