TL;DR — OpenAI launched advertising on ChatGPT in February 2026, with a CPM of $60 and a minimum commitment of $200,000 per advertiser. WPP estimates ad revenue between $500 and $800 million in the first year, but OpenAI's internal documents project $1 billion in 2026 and up to $25 billion by 2029. Meanwhile, Perplexity AI has abandoned advertising entirely to protect user trust. Only 14% of consumers trust AI recommendations for making purchases (Salsify, 2026), and 46% globally are willing to trust AI systems in general (KPMG, 2025). We are witnessing the greatest transformation in the relationship between users and artificial intelligence since ChatGPT's launch.
The day the impartial advisor stopped being impartial
There was a moment, brief but real, when millions of people believed they had access to a neutral advisor. A digital oracle with no hidden agenda. ChatGPT became, in just a few months, the place to ask "which laptop to buy," "which restaurant to choose in Milan," "which CRM software is best for an SME." And the answers came without asterisks, without banners, without the suspicion that someone had paid to appear at the top of the list.
That moment ended on February 9, 2026.
OpenAI officially activated advertising on ChatGPT, starting with Free and Go users in the United States. At Deep Marketing, we have been following this evolution since the first rumors emerged in 2025, and what we see today confirms a dynamic we know well: every platform that reaches critical mass ends up monetizing its users' attention. It happened with Google, with Facebook, with Instagram. Now it's happening with AI.
But this time there is a fundamental difference. Google displayed results and then added ads alongside them. ChatGPT talks to you. And when someone who talks to you starts recommending sponsored products, the relational dynamic changes radically.
How ChatGPT ads work: anatomy of a new channel
According to OpenAI's official announcement (January 2026) and data confirmed by Adweek and TechCrunch, ChatGPT's advertising system works as follows:
The point that deserves attention is the targeting. Unlike Google (keyword-based) and Meta (interest/behavior-based), ChatGPT combines the intent signal from the current question with the behavioral context of the entire conversation history and saved memories. As ALM Corp wrote in its analysis, "it's a targeting combination that has never existed before in advertising."
In practice, if you asked ChatGPT for wedding planning advice three weeks ago and today you're asking about a trip, the system knows who you are, what you're planning, and how to interconnect your needs. This is not a banner on a website. It is something profoundly different.
The numbers: how much is the AI ads market worth?
The projected numbers are impressive, but they need to be contextualized rigorously.
According to Kate Scott-Dawkins of WPP, OpenAI could generate between $500 million and $800 million in the first year of advertising, capturing between 0.1% and 0.3% of the global digital advertising market. But OpenAI's internal documents, reported by several specialized outlets, project much higher revenues: $1 billion from "free user monetization" in 2026, scaling up to nearly $25 billion by 2029.
To understand the scale: ChatGPT now has over 900 million weekly active users and approximately 50 million paying subscribers (February 2026 data). OpenAI generated $13.1 billion in total revenue in 2025, exceeding its $10 billion target, and is aiming for $29.4 billion in 2026. Advertising is the lever to monetize that 90%+ of users who don't pay for a subscription.
As we explain to our clients: when over 90% of your user base is free, advertising isn't an option. It's an economic inevitability.
The Google precedent: a story that repeats itself (but worse)
In 2000, Google AdWords launched with 350 advertisers and text ads on a cost-per-click basis. Today Google generates over $116 billion in advertising revenue. But the journey was not painless for user trust.
In 2013, the Federal Trade Commission (FTC) flagged a "decline in compliance" by search engines in clearly differentiating ads from organic results. Since then, the situation has worsened: paid ads have become visually almost identical to organic results, with increasingly discreet labels.
But there is a crucial difference. With Google, users see a list of results and know (or should know) that the top ones are sponsored. With ChatGPT, users receive a conversational response that they perceive as personalized advice. Inserting sponsored content in that context is epistemologically different: you're not contaminating a list, you're contaminating a dialogue.
The Perplexity case: when the alternative says "no, thanks"
The most instructive contrast comes from within the AI world itself. Perplexity AI, an AI search platform valued at $18 billion with annualized revenues of approximately $200 million, did exactly the opposite of OpenAI.
After launching ads in November 2024 with formats like "Sponsored Follow-Up Questions" and sponsored videos, Perplexity first stopped accepting new advertisers (October 2025) and then completely abandoned the advertising model in February 2026.
The reasoning was explicit and devastating for OpenAI's model. A Perplexity executive told the Financial Times: "A user needs to believe this is the best possible answer. Once ads appear in the results, users inevitably start wondering whether the answers maintain their integrity or contain subtle commercial influence."
This is a point we at Deep Marketing consider fundamental. It's not just a matter of "Sponsored" labels or visual separation. It's a question of epistemic trust: can I trust what this machine tells me, knowing that someone is paying to be there?
Consumer trust in AI: what the data says
The data on consumer trust in AI recommendations is, to put it mildly, complex.
According to the KPMG global report (2025), conducted with over 48,000 people across 47 countries, only 46% of people globally are willing to trust AI systems. Furthermore, 66% use AI output without evaluating its accuracy, and 56% make mistakes in their work because of AI. A paradox: we use AI but don't trust it enough to verify what it tells us.
The Salsify research (2026) is even more specific to our topic: only 14% of consumers trust AI recommendations alone to make a purchase. 27% use them but verify with other sources, and a third (33%) simply don't use them at all.
BCG (2026) offers a different perspective: over 60% of consumers express high trust in GenAI results, and the use of GenAI for shopping grew by 35% between February and November 2025. Consumers describe GenAI tools as "direct, objective, transparent, and personalized."
The point is that these perceptions were built in a pre-advertising environment. When ChatGPT was perceived as a neutral advisor, not as an advertising channel. The question that none of these studies can yet answer with certainty is: what happens to that 60% trust when users discover that some recommendations are sponsored?
The impact on purchasing decisions: who loses and who wins
The introduction of advertising on ChatGPT is not a neutral event for the market. It creates specific winners and losers.
According to BCG (2026), GenAI assistants and chat tools are the second most influential touchpoint in the purchase journey among users, and the most influential touchpoint among daily users. 41% of consumers have purchased a product recommended by AI in the last six months (CX Dive, 2025). More than one in five now start with AI tools when making decisions.
This means that advertising on ChatGPT is not a marginal channel. It is a direct interference in the most powerful decision-making touchpoint for a growing segment of consumers.
Who wins? Large brands with minimum $200,000 budgets — Adobe, Ford, Target, Albertsons — that can afford to enter the pilot program. Who loses? SMEs and emerging brands that don't have access to that channel, and whose products may be potentially excluded from ChatGPT's "organic" recommendations.
As we tell our clients: the game is changing, and the barrier to entry is high. But the opportunity for those who act now with an AI-optimized content strategy is real.
What businesses should do (and SMEs in particular)
At Deep Marketing, we work primarily with SMEs and companies in the European market. The question we ask ourselves — and that our clients ask us — is practical: what changes for us?
First: ChatGPT advertising is currently only in the USA. But international expansion is inevitable, likely by late 2026 or early 2027. Businesses have a window of preparation, not immunity.
Second: the real battle is not about ads, but about presence in AI's organic results. When a user asks ChatGPT "which marketing agency in Milan is the best," the answer is based on training data and, increasingly, on real-time information. Being present, authoritative, and citable in the sources that AI uses is the new SEO.
Third: trust becomes the real competitive asset. If 14% of consumers trust AI for purchasing (Salsify, 2026), but 67% say that product quality and value are the main factors for brand trust, then the strategy is not buying space on ChatGPT. It is building a brand that ChatGPT recommends spontaneously.
Fourth: diversify touchpoints. Companies that depend on a single channel — whether it's Google, Meta, or tomorrow ChatGPT — are structurally vulnerable. Resilience comes from multi-channel presence and the quality of proprietary content.
The transparency paradox: declared ads that nobody notices
OpenAI states that "ads do not influence the answers ChatGPT gives you" and that "conversations remain private from advertisers." But this is a technical distinction that ignores consumer psychology.
KPMG's research (2025) reveals that 66% of people use AI outputs without evaluating their accuracy. If most users don't even verify whether an AI response is correct, how many will truly notice the difference between an organic suggestion and a sponsored one?
Google's experience is instructive. The FTC documented how advertising labels became progressively less visible over time. In 2004, Google's ads had an obvious colored background. By 2025, the visual difference between an ad and an organic result is minimal. The same trajectory is predictable for ChatGPT: today the "Sponsored" label is clear, tomorrow it could be a discreet icon in a corner.
Furthermore, ChatGPT's "conversational banner" format introduces something unprecedented: users can ask the chatbot questions about the sponsored product. This creates a hybrid between advertising and content that is much harder to distinguish than a traditional banner.
The two-speed model: those who pay are "immune"
One aspect that deserves critical reflection is the two-speed model introduced by OpenAI. Free and Go users see advertising. Plus ($20/month), Pro ($200/month), Team, Business, Enterprise, and Education users do not.
This creates an unprecedented information stratification: those who can afford a subscription receive "pure" advice. Those who don't pay receive advice potentially influenced by commercial interests. It is the digital version of a disparity that already exists in access to information, but made explicit and systematic by a technological architecture.
In a context where AI is becoming a daily decision-making tool for billions of people, this asymmetry is an issue that the European regulator — with the AI Act already in force — should monitor closely.
What will happen in the next 12 months
Based on available data and the historical trajectory of advertising platforms, our predictions for the next 12 months are:
1. Geographic expansion: ChatGPT ads will arrive in Europe by late 2026, with a likely launch in Q4 in the UK, Germany, and France.
2. Self-serve ad platform: the $200,000 minimum commitment is typical of beta phases. OpenAI will launch a self-service platform accessible to SMEs as well, likely in 2027.
3. Erosion of transparency: "Sponsored" labels will become progressively less visible, as happened with Google.
4. Ad-free competitors: Perplexity and other players will use the absence of advertising as a competitive differentiator, creating a segmented market between "sponsored" AI and "pure" AI.
5. European regulation: the AI Act and the DSA will likely require stricter disclosure than the American model.
6. New metrics: specific metrics will emerge to measure the ROI of conversational AI ads, different from traditional CTR.
Frequently asked questions
Does ChatGPT show ads in Europe?
As of March 2026, ChatGPT ads are active only for users in the United States on the Free and Go tiers. Expansion to Europe is expected by late 2026 or early 2027, but there are no official dates yet for European markets.
Do ads influence ChatGPT's responses?
OpenAI states that advertising does not influence the AI-generated responses and that ads are visually separated with the "Sponsored" label. However, the "conversational banner" format, which allows users to interact with the chatbot about the sponsored product, makes the distinction between organic and sponsored content more blurred compared to traditional search.
How much does it cost to advertise on ChatGPT?
The CPM is $60 (roughly three times Meta's average) with a minimum commitment of $200,000 to access the beta program. Initial partners include major advertising groups WPP, Omnicom, and Dentsu with brands like Adobe, Ford, Target, and Audible. Currently, it is a channel accessible only to large advertisers.
Why did Perplexity abandon ads while ChatGPT launched them?
Perplexity AI, valued at $18 billion, decided that advertising compromised user trust in the quality of responses. With annualized revenues of approximately $200 million from subscriptions ($20 to $200 per month), it chose a subscription-only model. OpenAI, with over 900 million weekly users of which 90%+ are non-paying, faces very different economic pressure to monetize its free user base.
Can SMEs advertise on ChatGPT?
Not at this time. The minimum budget of $200,000 and the limitation to the US market make ChatGPT Ads inaccessible for most SMEs today. When OpenAI launches a self-service platform (expected in 2027), costs will likely decrease. In the meantime, the best strategy for SMEs is to optimize their presence in the sources that AI uses to generate responses: authoritative content, citations from trusted sources, structured data.
Is my ChatGPT conversation data shared with advertisers?
According to OpenAI's policy, advertisers do not have access to individual chats, history, or personal details. They only receive aggregated data such as number of views and clicks. However, the targeting system internally uses the conversation topic, location, language, and — from February 2026 — also chat history and saved memories to determine which ads to show.
How can I protect my business from the impact of AI ads on recommendations?
The most effective strategy is to invest in building authority and visibility in the sources that feed AI models. This means: high-quality, citable content, presence on authoritative sources in your industry, structured data on your website, authentic reviews, and a digital PR strategy that generates mentions and citations. At Deep Marketing, we call this approach "AI-readiness": don't buy space in AI, earn it.
Sources and References
- OpenAI — Our approach to advertising and expanding access to ChatGPT (2026)
- OpenAI — Testing ads in ChatGPT (2026)
- TechCrunch — ChatGPT rolls out ads (February 2026)
- Adweek — OpenAI Confirms $200,000 Minimum Commitment for ChatGPT Ads (2026)
- Adweek — ChatGPT Gets Ads: Omnicom, WPP, and Dentsu Line Up Brands (2026)
- Euronews — ChatGPT will now show you adverts (2026)
- KPMG — Trust, attitudes and use of artificial intelligence: A global study (2025)
- Salsify — 2026 Consumer Research Report: AI Trust Gap (2026)
- BCG — Consumers Trust AI to Buy Better. Brands Need to Move Quickly (2026)
- YouGov — American trust in AI for retail: Consumer sentiment (2025)
- MacRumors — Perplexity Abandons AI Advertising Strategy Over Trust Worries (2026)
- Campaign Asia — Perplexity pulls the plug on ads, citing trust concerns (2026)
- ALM Corp — OpenAI's Bold Advertising Strategy: Monetizing 800 Million Weekly Users (2026)
- Search Engine Land — A visual history of Google ad labeling in search results
- Zendesk — Global survey reveals growing consumer trust in personal AI assistants (2025)
