To turn ChatGPT into an advertising juggernaut, OpenAI will have to convince Madison Avenue that a chatbot can actually sell.
Mark Mahaney, Evercore ISI’s internet research analyst, said this week that OpenAI could reasonably generate several billion dollars in ad revenue this year, and as much as $25 billion by 2030 if its executives play their cards right.
So, what exactly will OpenAI need to do to turn ChatGPT ads into a multibillion-dollar business? (For context, EMARKETER forecasts that TikTok will register around $43 billion in net digital ad revenue this year, Microsoft $19 billion, and Apple $13 billion.)
Business Insider spoke with about a half-dozen advertising experts, including adtech CEOs, agency buyers, analysts, and academics. They said OpenAI needs a plan to ramp up beyond simple placements and to offer a data-driven platform that advertisers can trust to deliver outcomes.
Right now, ads are only being tested among a select group of brands to users in the US at the bottom of ChatGPT’s answers on its free and $8-a-month Go tier. In an interview on CNBC this week, OpenAI CFO Sarah Friar gave a couple of examples: A user looking for a place to stay in Dallas next weekend might see an ad from Expedia, Booking.com, or Airbnb. An expectant mother living in San Francisco researching the best strollers that won’t roll down a hill might get an ad from a baby specialty retailer.
Every time Lara publishes a story, you’ll get an alert straight to your inbox!
Stay connected to Lara and get more of their work as it publishes.
So far, so basic.
But advertising experts, including Gartner analyst Nicole Greene, said that OpenAI has a huge opportunity to match the wealth of data it has from users’ natural language queries with the products they actually want to buy.
“That’s marketing and advertising gold,” Greene told Business Insider. “Imagine actually getting a relevant ad!”
Context is nice. Performance is better.
Michael Cohen, EVP of performance media at the ad buying agency Horizon Media, described OpenAI’s initial foray into advertising as “a fairly innocuous injection of ads.”
“I wouldn’t expect it to perform because most contextual advertising is OK, it’s adequate, but not necessarily the top performers,” Cohen said. (Contextual advertising refers to ads targeted based on the content a user is viewing, rather than their personal data or browsing history.)
OpenAI
That’s because the most sophisticated advertising companies — think Google, Meta, and Amazon — have fine-tuned their algorithms to precision-target users with ads right at the moment they are likely to buy a product, no chat prompt required. And they’ve nailed the measurement side of the equation, demonstrating that the ads actually work.
A small-scale, contextual ad business is “at best a $1 to $2 billion revenue model in 2027” when you factor in the number of weekly users on ChatGPT’s free and Go tier, and the number of searches that are likely to trigger an ad, said Michael Komasinski, CEO of the adtech company Criteo.
To scale the business, OpenAI will need to develop a conversion API, Komasinski said. This is a type of “application programming interface” that allows advertisers to upload their data to report when a user takes an action, such as buying a product or downloading an app.
“They will have to create a performance ad unit that doesn’t ruin the customer experience but does allow advertisers to drive” incremental results, Komasinski said.
Maintaining consumer trust
OpenAI can’t risk users feeling squeamish that the app they used as their on-demand mental health therapist is now trying to sell them sneakers. And it must reassure users that ads won’t bias the answers OpenAI responds with.
“How are they going to do this balancing act?” said Ram Bala, associate professor of AI and analytics at Santa Clara University’s Leavey School of Business.
OpenAI has published its “ads principles,” which state that ads will be separate, clearly labeled, and won’t influence ChatGPT’s organic answers. The company said it will keep user conversations with ChatGPT private from advertisers and won’t sell user data to them.
Adam Edwards, chief product officer at the digital media agency Brainlabs, said consumers are accustomed to seeing ads alongside organic search results.
“As long as the ads are clearly labeled and distinct from the ‘organic’ advice — for example, a ‘Sponsored Option’ appearing alongside my restaurant recommendations — users will likely accept the trade-off for a free service,” Edwards said.
Increasing the surface area
Google built the world’s largest advertising business, generating more than $200 billion in annual ad revenue, by parlaying its enormous reach into a data-driven ad machine. A massive logged-in audience flows across Chrome, Android, Gmail, Search, and YouTube, all underpinned by a powerful adtech stack. That infrastructure spans advertiser tools, publisher monetization platforms, and an exchange that connects buyers and sellers.
While OpenAI isn’t likely to be interested in monetizing the web, it may need to think beyond simple ads in the ChatGPT app if it wants to build $25 billion ad business, advertising experts said.
Its AI-generated video app Sora would be an obvious destination for TikTok-like vertical video ads, for example. OpenAI is also pushing deeper into commerce through its agentic commerce protocol and instant checkout feature, areas where sponsored listings might not feel out of place. And then there’s its Atlas browser and whatever AI device former Apple design chief Jony Ive is working on.
“That’s where it gets interesting: How big can the company get outside the core app itself? That’s a multiplier on ads,” Criteo’s Komasinski said.
Relationships matter in the ad business
When companies like Uber and Netflix officially launched their ad businesses, the announcements usually coincided with the hiring of a well-known Madison Avenue face to lead them. That hasn’t yet been the case at OpenAI.
Dan DeLong
OpenAI’s ad-related hires to date have largely focused on building infrastructure rather than building relationships.
On Wednesday, The Information broke the news that Vijaye Raji would lead OpenAI’s ads team, reporting to Fidji Simo, CEO of applications, citing an internal memo. Previously the company’s CTO of apps, Raji joined OpenAI when it acquired his product development startup for $1.1 billion in September last year.
“In a few words: he’s the GOAT,” said Julien Codorniou, a general partner at the venture capital firm 20VC. Codorniou, an investor in Statsig and a former colleague of Raji’s at Meta, said Raji was ideally placed to help build OpenAI’s ad platform.
As a software engineer at Meta, Raji worked on products including its Audience Network and mobile app install, per his LinkedIn profile. He had roles leading Meta’s gaming and entertainment efforts and oversaw the expansion of its Seattle office.
Now that OpenAI has a point person for ads, advertising insiders predicted the company would ramp up its dedicated advertising staff. Further acquisitions could be one way to get there.
“I would think they would do more because they need more talent,” said Criteo’s Komasinski, who suggested OpenAI might look at acquisitions in hard-to-build areas like ad serving, bidding, and personalization engines. “Organic hiring is slow, even when you’re a destination employer like OpenAI, but they certainly do seem to be very selective.”