OpenAI launches advertising platform: a wealthy business selling to the poor

Sam Altman once described advertising as ChatGPT’s “last resort.”

For a long time, this was a form of restraint. OpenAI still brands itself as a research company, an infrastructure company, and a company trying to democratize AI capabilities for everyone. Advertising, the most familiar monetization method of the old internet, was considered an alternative.

But the shift to advertising became rapid.

On May 5, OpenAI launched the self-serve Ads Manager platform, allowing advertisers to place ads directly or through agencies like Dentsu, Omnicom, Publicis, WPP on ChatGPT. Less than three months after the initial ad pilot launched on February 9.

The platform is still in testing, but the direction is clear: ChatGPT is no longer just a conversational product; it is beginning to become ad inventory. OpenAI aims to reach $2.5 billion in ad revenue by 2026 and push ad income to $100 billion by 2030.

With 900 million users, ChatGPT finds that the free route is becoming increasingly difficult.

Billions in annual losses, recovering via ads

OpenAI is growing rapidly, so fast that traditional internet companies find it hard to benchmark.

But it also burns money quickly.

HSBC analysts estimated that by the end of 2025, OpenAI could still face a funding gap of $207 billion. Its cloud and AI infrastructure expenses from late 2025 to 2030 might reach $792 billion, with long-term compute commitments possibly approaching $1.4 trillion by 2033.

These figures explain why they are building out an advertising business.

Subscription revenue proves users are willing to pay but struggles to cover the reasoning costs of all free users. Enterprise APIs can generate cash flow but face price wars and model convergence. Capital financing can extend life but dilutes equity and pushes higher valuation pressures back into the company.

Advertising is the fastest non-dilutive revenue source. It doesn’t require free users to pay, doesn’t need to re-educate the market, and is easier to pitch to investors.

According to Reuters, OpenAI’s ad pilot generated over $100 million in annualized revenue within six weeks. Ads are only shown to free and Go plan users, do not affect ChatGPT responses, and user data is not shared with marketers.

Setting aside user privacy, there’s a more fundamental question behind this strategy: Ads are sold to free users, but advertisers are after paying users.

ChatGPT has 900 million weekly active users, about 50 million paid subscribers, and less than 6% conversion from free to paid. Ads are only shown to free users, meaning OpenAI’s ad inventory is entirely from the 94% who are unwilling to pay.

The problem is, advertisers willing to spend at least $50k often sell products not aimed at individual consumers. Decision-makers for enterprise software, SaaS tools, B2B services—these high-ticket categories—are most likely to be ChatGPT’s paying users. They spend $20 to $200 monthly on better models and larger context windows, yet their screens will never display ads.

Beyond audience mismatch, there’s a deeper issue: even if ads successfully reach free users, how much ad value can their usage scenarios support?

High intent doesn’t equal high conversion

OpenAI’s ad narrative is built on a core assumption: ChatGPT users enter conversations with genuine intent, making high-intent ad impressions worth more.

This assumption is only half correct.

Over the past twenty years, brands most eager to capture the search box because it signals intent. When users search for hotels, they might be booking; searching for enterprise tax software suggests procurement; searching for the best noise-canceling headphones indicates they’re close to a purchase decision.

Google built its ad empire on this.

After ChatGPT appeared, users directly entrust decision-making to AI. For advertisers, this is more enticing—and more frightening—than search ads.

The appeal is that ChatGPT sees an entire demand segment; it not only knows what users want to buy but also why they buy it this way. The danger is that if AI provides direct answers, users might skip the search results page altogether.

But “help me buy a pair of running shoes” and “help me write an email” are two entirely different intents. The former is a consumption scenario; the latter is productivity. In daily use, productivity tasks dominate ChatGPT—writing, translating, coding, planning, emotional regulation—high frequency but not naturally linked to product purchases.

This will directly lower ad effectiveness metrics. Advertisers are willing to pay high prices for high-certainty purchase intent. Google search ads are expensive because users often have clear intentions to buy, compare, book, or order. Meta ads are cheaper but leverage social profiles and vast conversion data to repeatedly filter low-intent users into potential consumers.

ChatGPT sits in the middle.

It’s more like a demand gateway than social, but harder to judge commercial intent than search. It’s more private than search but more difficult to attribute. It can solve user problems but may not generate ad clicks.

This is why OpenAI is shifting from CPM (cost per thousand impressions) to CPC (cost per click). It’s not just a product upgrade; advertisers are reluctant to pay long-term based on the “next-generation search” concept.

They will ultimately ask: who brought this click? Where does the conversion happen? How much budget should be shifted from Google, Meta, TikTok to ChatGPT?

Category fit is also an issue. Low-risk categories like home, travel, education, software tools can experiment first. High-margin categories—often highly regulated, such as finance, healthcare, insurance, recruitment—pose risks beyond ad effectiveness: misguidance, discrimination, and compliance.

Google’s approach is a mirror. In Q1 2026, Google’s search ad revenue hit $77.25 billion. Yet even so, ad placements in AI Mode and AI Overviews remain cautious; the standalone Gemini app has yet to run official ads.

OpenAI expanding into advertising explores broader commercial models for large models.

It aims to make users feel AI is approachable while convincing advertisers there’s enough commercial intent. If this balance tips, ChatGPT risks losing both: users feel it’s inauthentic, advertisers see no conversion.

But the impact of advertising goes beyond that; it’s reshaping how brands compete.

GEO’s focus is shifting

Over the past year, brands worried about disappearing from AI responses. This concern is often framed as GEO, but it’s not a new concept; it’s just the old search marketing anxiety repackaged for the AI era.

OpenAI’s launch of Ads Manager hits this anxiety precisely but also shifts its focus.

In a no-ads era, the core issue for GEO is “how to enter AI’s context.” Brands compete by providing product documentation, media coverage, third-party reviews, community discussions—trying to be referenced by the model, based on information quality and data structuring.

With the ad platform, targeted traffic can be directly purchased, reducing reliance on organic mentions. But the competition focus doesn’t revert to “buy more exposure.” Instead, it shifts from “how to enter AI’s answer” to “what does AI say about my product.”

The reason is simple: after seeing ads, the most natural next step for users is to ask AI, “Is this product really good?” AI’s response becomes the real conversion gate. Brands can buy exposure but not positive AI reviews. If AI, based on public data, gives negative feedback, every dollar spent on ads accelerates user loss rather than conversion.

This means brands need to build positive reputation within AI’s evaluation system. Product quality, user review density, third-party coverage—these signals AI can read—will matter more for conversion than the ads themselves.

GEO shifting from “entering context” to “winning evaluation” is a key trend to watch as OpenAI launches its new ad platform.

Not advertising in 2026 will be the most expensive ad

Talking about OpenAI, one must mention its rival Anthropic, which is taking a completely different “ad model” path.

On February 4, 2026, two days before the Super Bowl, Anthropic published a blog stating that Claude will never run ads. No sponsorships, no third-party placements.

This statement itself is a costly ad.

Super Bowl ads are expensive; Anthropic spent heavily to tell users it doesn’t sell ads, essentially buying brand recognition through the absence of ads.

Not running ads is never just a moral stance; it’s a business positioning. It signals to enterprise clients, professional users, and high-sensitivity scenarios that Claude’s answers are unaffected by advertisers, and its product direction isn’t optimized for ad inventory. Its revenue comes from what you pay.

The effect is immediate. Claude’s ranking in the US App Store rose from 42nd at the start of the year to the top. On February 28, after OpenAI’s Pentagon contract sparked the QuitGPT movement, Claude hit #1 in the US App Store’s free apps, surpassing ChatGPT for the first time ever. Free active users increased by 60%, daily sign-ups quadrupled, and paid users doubled within a week.

Anthropic’s revenue structure is entirely different from OpenAI’s: over 80% comes from enterprise clients, with annual recurring revenue soaring from about $9 billion to $19 billion. Claude Code and Cowork enterprise tools have contributed at least $1 billion. Anthropic doesn’t rely on free user ad value; it depends on enterprise trust that data won’t be used for advertising, allowing higher subscription prices.

Choosing not to advertise is a precise business decision—by sacrificing ad revenue, it reinforces trust with enterprise clients, supporting higher subscription pricing.

However, “not advertising” isn’t an eternal virtue.

According to Stanford’s AI Index, the cost to reach GPT-3.5-level performance has dropped 280 times in two years—from $20 per million tokens in November 2022 to $0.07 in October 2024.

If model capabilities continue converging and API price wars erupt, Anthropic’s current enterprise subscription premium might be eroded over time. When model costs fall enough for all competitors to offer similar performance, why would enterprise clients keep paying more for Claude?

No conclusion exists yet, but time will tell what this choice is worth.

There’s no free lunch in the world

OpenAI chooses ads; Anthropic turns “no ads” into a premium. They seem opposite but are actually answering the same question: when AI product reasoning costs can’t be covered long-term by free models, who pays?

OpenAI’s Ads Manager isn’t just an ad product; it’s a signal that the AI industry is shifting from free expansion to cost recovery.

But OpenAI’s way of stopping the bleeding exposes the most fragile part of this business: it relies on a user base with the least consumer intent—paying three times more than Meta—to sustain a higher ad price.

This isn’t a problem that can be solved by user scale alone. 900 million active users is a nice number, but if they come to ChatGPT to write emails rather than buy things, advertisers will eventually vote with their feet.

Ads can be a revenue source for AI products, but they shouldn’t be the only answer. Because if a product’s business model depends on users staying as long as possible and exposing as many intents as possible, that product is no longer a user’s assistant—it’s an advertiser’s assistant.