OpenAI Enters the Ad Business: Everything We Know So Far

The financial pressure on OpenAI has been mounting for some time. The company is burning through an estimated $8 to $9 billion per year in compute costs alone and is projected to lose around $14 billion in 2026, even as revenue climbs.^¹ OpenAI’s annualized revenue hit $20 billion in 2025, up from $6 billion the year before — impressive growth by any measure, but still not enough to offset the sheer scale of its infrastructure commitments, which include major data center deals with Microsoft, Oracle, and Amazon.^¹

The subscription model, while meaningful, has a clear ceiling. Only about 5% of ChatGPT’s user base pays for Plus or Pro tiers.^¹ That leaves roughly 95% of hundreds of millions of users — including an enormous free-tier audience — generating engagement without generating revenue. Advertising was the logical answer. The only question was when and how.

The Announcement and Rollout

On January 16, 2026, OpenAI confirmed it was beginning to test advertisements inside ChatGPT in the United States.^² Ads would appear for logged-in adult users (18+) on the free tier and the newly launched ChatGPT Go plan, priced at $8 per month. Premium subscribers on Plus ($20/month), Pro ($200/month), and Enterprise plans would remain completely ad-free — a deliberate move to preserve the value proposition of paid tiers.^³

The official rollout began on February 9, 2026, accompanied by an updated privacy policy that OpenAI emailed directly to users — an unusual step that signaled the company understood the sensitivity of what it was doing.^⁴ By that point, ChatGPT had grown to more than 800 million weekly active users, making it one of the largest untapped advertising audiences in digital media.^³

What the Ads Actually Look Like

OpenAI has been emphatic about one design principle: ads will not contaminate ChatGPT’s answers. The company calls this “answer independence,” and it’s built into the platform architecture itself.^⁵ When a user asks a question, ChatGPT generates its response based purely on its training and context. A separate system then determines whether to display a relevant advertisement — clearly labeled as “Sponsored” — in a visually distinct section beneath the AI’s answer.^²

These are contextual text ads, matched to the topic of the conversation. During the initial testing phase, the ad selection mechanism drew on conversation topics, past chat history, and previous interactions with ads.^⁵ Critically, OpenAI pledged not to sell user data to advertisers or give advertisers access to individual conversations.^² Advertisers in the early phase received only aggregated performance data — impressions and clicks — without detailed conversion tracking or demographic breakdowns.^³

The ad pricing reflected the premium nature of the platform. ChatGPT commanded a CPM (cost per thousand impressions) of around $60 during initial testing, significantly higher than Google’s search ads at approximately $38 — a sign that advertisers viewed the conversational context as more valuable than traditional keyword search.^³

Access and the Advertiser Ecosystem

Early access to the platform was not self-serve. In the initial phase, inventory was available only through direct outreach to select enterprise partners.^³ By May 2026, however, OpenAI had opened the platform more broadly through a self-serve OpenAI Ads Manager supporting both CPC (cost per click) and CPM bidding, with no minimum spend requirement at entry level.^⁶

Major advertising holding companies signed on quickly. Dentsu, Omnicom, Publicis, and WPP all became agency partners, and integrations were established with ad tech platforms including Adobe, Criteo, Kargo, Pacvue, and StackAdapt.^⁶ The platform generated a reported $100 million in revenue during its first six weeks, demonstrating that advertiser demand was real and immediate.^⁶

OpenAI also expanded its commercial video footprint separately — its video model Sora 2 became licensed for commercial advertising, opening a new creative channel for brands.^⁷

The Revenue Vision

OpenAI’s financial projections for its ad business are staggering in their ambition. According to reporting from Axios in April 2026, the company’s internal targets run as follows: $2.5 billion in advertising revenue for 2026, rising to $11 billion in 2027, $25 billion in 2028, $53 billion in 2029, and $100 billion by 2030.^⁷

To understand what’s driving that confidence, consider the broader context. The AI-powered advertising market is projected to grow from $35 billion in 2025 to $142 billion by 2030 — a 306% increase.^⁸ Within that, eMarketer projects AI-driven search ad spending in the U.S. alone will surge from roughly $1.1 billion in 2025 to $26 billion by 2029.^⁸ OpenAI is positioning itself to capture a substantial slice of that expansion.

Underlying the ad push is a major shift in OpenAI’s consumer subscription model. According to reporting from The Information in April 2026, ChatGPT Plus is projected to decline sharply — from 44 million subscribers to around 9 million — as the ChatGPT Go tier grows from 3 million to an estimated 112 million subscribers.^⁷ The company is deliberately trading high-revenue subscribers for massive scale, then monetizing that scale through advertising.

Privacy, Trust, and the Backlash

The reaction to OpenAI’s ad rollout was predictably mixed. Privacy advocacy groups raised concerns about how conversation data would be used to target ads, even with OpenAI’s pledges about data sharing.^¹ Tech commentators questioned whether commercial interests could ever be truly separated from AI-generated answers over time. Some called it inevitable; others called it a betrayal of what made the platform valuable in the first place.^⁹

OpenAI has been careful to address these concerns procedurally. Its ad policies include restrictions on placement in sensitive contexts — conversations about mental health, crisis situations, and similar topics are ineligible for ad display.^¹⁰ Ads are also prohibited near content that would damage brand safety. A May 2026 update to the policy added clearer guidance on review and enforcement standards.^¹⁰ One earlier restriction was quietly relaxed: medical, legal, and financial advice contexts, which were initially blocked by default, are now eligible for ads provided the conversation is deemed appropriate.^¹⁰

Consumer sentiment research has offered a nuanced picture. IAB research from early 2026 found emerging skepticism among Gen Z users toward AI-delivered ads, despite that demographic’s general comfort with AI technology overall.^⁸ Separately, research from Verve Group found consumers increasingly willing to accept ads in exchange for free AI access, but simultaneously growing more anxious about how their data might be used.^⁸ OpenAI has reported no measurable decline in consumer trust metrics during beta testing, attributing that to the clear visual separation between answers and ads.^⁸

The Competitive Landscape

OpenAI is not alone in building out AI advertising. Google has been integrating advertising into its Gemini AI products, leveraging its dominant position in search and its enormous existing advertiser relationships. With over $200 billion in annual ad revenue and decades of infrastructure, Google possesses significant advantages in AI advertising that no competitor can match overnight.^¹¹

Google’s approach with Gemini has been notably more cautious — and more contradictory — than OpenAI’s. In December 2025, Adweek reported that Google representatives had told at least two advertising clients in direct calls that ad placements in Gemini were targeted for a 2026 rollout.^¹² Google’s VP of Global Ads, Dan Taylor, quickly pushed back, stating publicly that there were “no ads in the Gemini app and no current plans to change that.”^¹² Then, at the World Economic Forum in Davos in January 2026, Google DeepMind CEO Demis Hassabis echoed that position, telling reporters the company had “no plans” to introduce ads into Gemini.^¹² The official stance shifted again, however, when Google’s Senior VP of Knowledge and Information, Nick Fox, told WIRED the company was “not ruling them out,” adding that learnings from ads in AI Mode in Search would “likely carry over” to Gemini down the road.^¹² What is clear is that Gemini’s advertising strategy is evolving in a different direction than a traditional chatbot ad rollout. At Google Marketing Live 2026, the company unveiled Gemini-powered ad formats built specifically for AI Mode in Search — including conversational discovery ads that generate customized creatives based on complex user questions, and highlighted answers that let advertisers appear as featured recommendations in AI-curated lists.^¹³ Google is also piloting Direct Offers, which surfaces exclusive promotions for users showing clear purchase intent within AI conversations, and is building toward enabling checkout directly inside AI Mode.^¹² In short, rather than placing ads inside Gemini as a standalone chatbot, Google appears to be weaving Gemini’s capabilities into its existing — and enormously lucrative — search and commerce advertising infrastructure.

Meanwhile, Anthropic — the maker of Claude — has taken a conspicuously different path. The company ran a Super Bowl campaign in early February 2026 positioning Claude as an ad-free alternative, with the pointed tagline aimed squarely at the moment OpenAI’s ads went live.^⁵ Whether that positioning proves durable or merely a temporary competitive differentiation remains to be seen.