The AI Advertising Split: OpenAI Goes All In on Ads While Rivals Refuse

Three companies. Three business models. One question that will determine who controls the next generation of consumer technology: should your AI assistant sell you things?

In the span of two weeks this February, the AI industry fractured along a fault line that has nothing to do with model performance, context windows, or benchmark scores. It is about money - specifically, where it comes from.

Company	Ad Strategy	Revenue Model	Minimum Ad Spend	CPM
OpenAI (ChatGPT)	Ads live since Feb 9	Subscriptions + Ads	$200,000	$60
Anthropic (Claude)	Ad-free pledge, Super Bowl campaign	Subscriptions + Enterprise	N/A	N/A
Perplexity	Abandoned ads Feb 18	Subscriptions + Enterprise	N/A	N/A
Google (Gemini)	Denies plans despite advertiser briefings	Ads (Search) + Subscriptions	Unknown	Unknown

OpenAI Pulls the Trigger

On February 9, OpenAI began showing sponsored content to free-tier and Go ($8/month) ChatGPT users in the United States. The ads appear below responses, clearly labeled as "sponsored," and are matched to conversation topics. Someone researching recipes might see a grocery delivery promotion. A coding question could surface a developer tool.

The pricing tells you everything about how OpenAI views its position. At $60 CPM - three times what Meta charges - and a $200,000 minimum commitment, this is not a product for small businesses. It is a premium play targeting brands willing to bet that conversational context is worth more than a social media scroll.

"Ads do not influence ChatGPT's answers. Ads run on separate systems from the chat model, and advertisers have no ability to shape, rank, or alter ChatGPT's responses," OpenAI stated in its announcement.

The Financial Pressure

The math explains the move. OpenAI's annual recurring revenue hit $20 billion in 2025, tripling from $6 billion the year prior. But the company's own internal forecasts project a $14 billion loss in 2026 - roughly three times worse than 2025 losses. Revenue projections of $30 billion for 2026 still leave a massive gap when you are burning cash at this rate. When your costs grow faster than your subscriptions, advertising is the obvious lever.

Anthropic Fires Back With a Super Bowl Ad

Five days before ChatGPT ads went live, Anthropic launched what may be the most expensive counter-positioning move in AI history: a multi-million dollar Super Bowl campaign titled "A Time and a Place."

The four spots - titled "Deception," "Betrayal," "Treachery," and "Violation" - each showed a helpful AI assistant mid-conversation, abruptly pivoting to pitch unrelated products. One featured a man asking for workout advice and getting a robotic sales pitch for shoe insoles. The tagline: "Ads are coming to AI. But not to Claude."

Anthropic's reasoning is straightforward. Internal analysis found that a significant share of Claude conversations involve sensitive or deeply personal topics. An advertising model would create incentives to optimize for engagement and time spent rather than usefulness.

"We want Claude to act unambiguously in our users' interests," Anthropic stated. "A conversation with Claude should be a space to think, not a space to be sold to."

Altman Responds

OpenAI CEO Sam Altman did not take it well. He called Anthropic's ads "funny" but "clearly dishonest" and "deceptive," arguing that OpenAI would "obviously never run ads in the way Anthropic depicts them." The exchange devolved into what TechCrunch described as a "novella-sized rant" that included calling his rival "authoritarian." The two CEOs later refused to hold hands at the India AI Impact Summit - a moment that perfectly captured the state of the relationship.

Perplexity Walks Away From the Table

Then, on February 18, Perplexity quietly did something more significant than either Super Bowl ad or ad launch: it abandoned advertising entirely.

Perplexity was among the first generative AI companies to test ads back in 2024, placing sponsored answers beneath chatbot responses. It generated roughly $20,000 in ad revenue that year - less than 0.1% of its $34 million total revenue. By late 2024, the company was already phasing ads out. This month, executives confirmed no plans to revisit them.

The reasoning cuts to the core of what makes AI different from traditional search, as covered in our AI search engine comparison. Google serves a list of links. Users have spent two decades learning to distinguish organic results from paid placements. An AI chatbot gives you one synthesized answer. There is no list to scroll past.

"A user needs to believe this is the best possible answer, to keep using the product and be willing to pay for it," a Perplexity executive explained. "We are in the accuracy business, and the business is giving the truth, the right answers."

The Numbers Work Without Ads

Perplexity now reports 780 million monthly queries, over 100 million users, and annualized revenue of approximately $200 million, driven by subscription tiers ranging from $20 to $200 per month. The company is ramping enterprise sales, targeting finance professionals, CEOs, and doctors - high-value users willing to pay for reliable answers. Five people on the enterprise sales team, with plans to expand aggressively.

The Google Wildcard

Google occupies the strangest position in this split. Adweek reported in December 2025 that Google representatives told at least two advertising clients that Gemini ad placements were "targeted for a 2026 rollout." Google's VP of Global Ads Dan Taylor then publicly denied it, calling the report the product of "uninformed, anonymous sources."

The contradiction is telling. Google already runs ads in AI Overviews and recently launched a shopping ad format inside AI Mode, which reaches over 75 million daily users. But Gemini - the standalone chatbot competing with ChatGPT and Claude - remains ad-free for now. Google appears to be hedging, watching whether OpenAI's ad experiment alienates users before committing.

Counter-Argument

The bull case for AI advertising is simple: it works everywhere else. Google built a $300 billion business on search ads. Meta built a $130 billion business on social ads. The pattern is consistent - free access funded by advertising beats paid access for mass adoption every time.

OpenAI's $60 CPM also suggests something that subscription purists miss: conversational context may be genuinely more valuable to advertisers than any other format. When a user tells ChatGPT "I'm planning a trip to Japan in March," the intent signal is stronger than any search query or social media profile. If the ads truly do not influence the model's answers, the user experience trade-off could be minimal.

And there is the access argument. OpenAI explicitly frames advertising as a way to keep ChatGPT free for hundreds of millions of users who cannot or will not pay $20 per month. Anthropic's ad-free pledge sounds principled until you ask who gets locked out when the only business model is subscriptions.

What the Market Is Missing

The real story is not about ads. It is about what advertising does to product incentives over time.

Every ad-supported platform in history has eventually optimized for engagement over utility. Facebook did not start with algorithmic feeds designed to maximize time-on-site. Google did not start with sponsored results that look nearly identical to organic ones. These changes happened gradually, driven by the same quarterly revenue pressure that will eventually apply to OpenAI.

Anthropic and Perplexity are making a bet that AI is different - that the relationship between user and model is intimate enough that advertising breaks the product in ways it never broke search or social media. It is an expensive bet. Anthropic spent millions on a Super Bowl campaign to make it. Perplexity walked away from an entire revenue category.

The market will decide who was right. But if you want to know where an AI company is headed, do not read its benchmark scores. Read its revenue model. The money always tells the truth.

Sources: