ChatGPT Crossed $100M in Ad Revenue in 60 Days. The CPM Fell From $60 to $25. Here's the Business Model.

On February 9, 2026, OpenAI launched its first advertising test inside ChatGPT at a $60 CPM — one of the highest opening price points in digital advertising history, roughly 4× what a premium YouTube placement costs. The pitch to brands: access to 800 million weekly ChatGPT users with stated intent so crisp that even Google's search team would envy it. Someone asking ChatGPT "what's the best project management software for a 50-person engineering team" is not browsing. They are buying.

Less than 60 days later, OpenAI's ads pilot crossed $100 million in annualized revenue. Six hundred advertisers participated. The privacy trust metrics held. The press wrote that OpenAI had pulled off the cleanest advertising launch in memory.

There was just one problem: the $60 CPM couldn't hold.

By mid-April 2026 — ten weeks after launch — clearing CPMs had fallen to approximately $25. OpenAI simultaneously cut the minimum spend requirement from $200,000 to $50,000, launched cost-per-click bidding at $3 to $5 per click, and opened a self-serve Ads Manager to all US advertisers on May 5, 2026. The combination of premium launch pricing collapsing under supply pressure and a pivot to lower-barrier self-serve buying tells you almost everything about the structural challenge OpenAI is navigating as it attempts to build advertising into a $2.5 billion business this year and a $100 billion business by 2030.

Understanding why requires understanding what ChatGPT ads actually are, what the CPM compression means in context, and whether the revenue targets reflect ambition or evidence.

What ChatGPT Ads Actually Are

The ChatGPT advertising format is structurally different from both Google search ads and social media display advertising, though it borrows elements from both.

Ads appear as clearly labeled sponsored placements within ChatGPT's response interface. When a user asks a question in commercial intent categories — software recommendations, travel planning, financial services, consumer products — a "Sponsored" labeled response or recommendation may appear alongside or within the AI-generated answer. The placement is conversational, not banner-style.

The targeting model is intent-based rather than behavioral. OpenAI does not use third-party data or tracking pixels. Targeting happens at the query level: what did the user ask, in what category, and which advertiser's product matches that intent? This is closer to keyword targeting in Google Search than to Meta's audience targeting — but without the granular keyword bidding complexity. Advertisers select categories, not specific terms.

The format has two things going for it that Google and Meta cannot fully replicate: the user is actively seeking an answer rather than passively scrolling, and the ad can be delivered after the AI has already demonstrated value by providing a useful response. That sequence — value first, sponsorship second — is the inverse of most digital advertising.

The format has one significant structural problem: inventory scarcity relative to demand has defined its early economics, and the compression from $60 to $25 CPM is the market resolving that scarcity.

Why the $60 CPM Couldn't Hold

A $60 CPM implies that every 1,000 ChatGPT query impressions where an ad appears is worth $60 to advertisers. At launch, with 600 curated advertisers competing for limited inventory in a high-intent format, that pricing was defensible. The scarcity was real: OpenAI had not opened the platform widely, and early advertisers were buying novelty premium as much as performance.

Inventory expansion and CPM compression are the same event, viewed from two perspectives. As OpenAI opened more query categories to advertising, invited more advertisers, and launched the self-serve platform, the supply of available ad placements grew faster than performance data could justify premium pricing.

The clearing CPM of approximately $25 — where the market has settled — is not a failure. It is a more defensible number than $60 for two reasons.

First, $25 CPM remains significantly above industry averages for premium digital placements ($8-$15 for Google Search equivalents, $7-$12 for social), meaning ChatGPT ads still price at a premium that could be justified by intent quality and engagement if conversion data supports it.

Second, the compression from $60 to $25 in 10 weeks maps to the standard pattern for new advertising inventory categories. Google Sponsored Links launched in 2002 at pricing that looked unsustainable, then found its equilibrium. Facebook launched ads in 2007 at effective CPMs that seemed high relative to display advertising and then defined a new category. New advertising surfaces routinely command novelty premiums that compress as the market expands and performance expectations become concrete.

The risk is not that $25 is too low for the format. The risk is that $25 is sustainable only if ChatGPT ads demonstrate conversion performance that justifies the premium over established channels — and that data does not yet widely exist.

The CPC Pivot: What $3-$5 Per Click Actually Means

The shift from CPM-only to CPC bidding is more strategically significant than it appears at the surface level. It signals that OpenAI understands the fundamental challenge of scaling advertising beyond brand-awareness budgets: premium CPMs require either blind faith in brand value or demonstrated click-through-to-conversion data. Brand-awareness advertisers (who pay CPM regardless of clicks) require large budgets and patience. Performance advertisers (who pay per outcome) are where most advertising growth has come from over the past decade.

The $3-$5 per click recommended starting bid is competitive with Google search CPC for many commercial categories. Google's average CPC across all categories is approximately $2-$3, with premium B2B software categories running $5-$15. A $3-$5 starting bid for ChatGPT positions the format as comparable to mid-tier Google Search inventory while offering the intent quality of a conversational query.

The critical question for CPC bidding is: what converts? Google's CPC model is built on 25 years of conversion data at massive scale. Advertisers know their conversion rates, cost-per-acquisition, and return on ad spend. ChatGPT's conversion benchmarks across categories are nascent — the Ads Manager launched measurement tools alongside CPC bidding, but performance norms do not yet exist.

For brands evaluating whether to allocate budget toward OpenAI's self-serve ads platform, the practical playbook for early-stage testing:

1. Start with high-intent commercial categories where your product is a natural recommendation. If someone asks ChatGPT to recommend a project management tool and you're a project management company, the intent match supports a defensible conversion thesis.

2. Set conservative bids and run alongside Google for comparison. The data does not yet support paying a premium to ChatGPT over proven channels. Treat it as a new channel test with a capped budget — not a substitution for existing spend.

3. Measure click quality, not just click volume. Early reports suggest ChatGPT clicks convert differently than Google clicks: users may be further along in the consideration phase because they've already received an AI recommendation, which can mean higher intent but also a smaller total audience pool.

4. Watch your organic presence as a signal. If ChatGPT already cites your company in organic responses for relevant queries, paid placements in those same categories may amplify an existing signal. If your product doesn't appear organically, a paid placement may feel incongruous to users who see it.

5. Audit brand safety regularly. Category-level targeting means your ad can appear in query contexts you didn't explicitly select. Regular review of actual placement contexts is essential until OpenAI provides more granular placement transparency.

The Revenue Math: Is $2.5 Billion in 2026 Real?

OpenAI's stated $2.5 billion advertising revenue target for 2026 requires context to evaluate honestly.

The company crossed $100 million in annualized revenue within two months of launch with approximately 600 advertisers. Scaling to $2.5 billion requires one of three things: reaching 15,000 advertisers at similar average spend; maintaining advertiser count while dramatically increasing spend per advertiser; or some combination that the self-serve Ads Manager is designed to enable by removing the spend minimums that currently constrain the pool of eligible advertisers.

ChatGPT's 800 million weekly active users generate an enormous query volume daily. Even with conservative monetization assumptions — ads shown on 5-10% of queries, CPMs at $25 — the theoretical inventory value is large. The constraint is not user volume. The constraint is advertiser adoption at sufficient scale and demonstrated conversion performance that justifies sustained budgets.

For context: Google's own AI Overviews have begun compressing its advertising revenue by reducing clickable results in favor of direct answers — a dynamic that simultaneously hurts Google's core business and creates an opening for ChatGPT to attract ad spend from brands seeking presence in AI-generated answers.

The $2.5 billion target for 2026 is aggressive but not implausible. The $100 billion target for 2030 is best understood as a strategic positioning statement: OpenAI intends to compete for the global advertising market, not build a supplementary revenue stream. Whether that ambition translates to market share depends on factors that are genuinely unknown today — user retention trends in ChatGPT, long-run advertiser performance data, and whether the format can expand beyond the US market.

Zero-click search has already pushed advertisers to diversify beyond traditional search placements. The brands most affected by that shift are among the most likely candidates to experiment with ChatGPT advertising as an alternative means of reaching high-intent audiences.

The Trust Paradox: What OpenAI Is Risking

The most important structural feature of ChatGPT advertising is also the thing most likely to destroy it if mismanaged: users trust ChatGPT's recommendations in a way they have learned not to trust Google's ads.

This is a double-edged characteristic. On the positive side, a ChatGPT sponsored placement has higher perceived credibility than a "Sponsored" label on a Google search result — most users have learned to skip those entirely. On the negative side, that credibility is borrowed from the underlying trust users have placed in the AI as an honest broker. Using it to deliver paid recommendations risks the exact kind of trust erosion that degraded native advertising's effectiveness across media over the past decade.

LLMs are already the primary citation source across internet content, and user expectations around AI neutrality are high. If ChatGPT's paid placements become indistinguishable from its organic recommendations, the long-run consequence is that users stop trusting either. The format's value proposition depends entirely on maintaining that distinction.

OpenAI's current approach — explicit "Sponsored" labeling and separation of paid from organic content — is the right starting position. The design challenge is maintaining that discipline as revenue pressure scales. Advertising revenue targets that imply significant inventory expansion create an economic incentive to maximize monetizable query surface, which over time tends to blur the line between editorial and commercial content.

What This Means for Publishers, Agencies, and Brands

The ChatGPT advertising launch has implications that extend well beyond the direct advertiser relationship.

For publishers, the launch confirms a shift in the attention and ad dollar ecosystem that affiliate marketing operators are already experiencing: users are migrating toward AI interfaces for information-seeking, and advertising budgets follow attention. If ChatGPT scales to significant ad volumes, a portion of that spend will be redirected from traditional search and display advertising — affecting Google, programmatic networks, publisher direct buys, and newsletter advertising simultaneously.

For agencies, the Ads Manager launch creates a new channel to manage and report on — and a new creative discipline to develop. ChatGPT ads require copy that sounds conversational and contextually appropriate, not the truncated search-ad format or the visual-forward social format that most agencies have optimized for. The agencies that invest early in understanding ChatGPT's placement context and conversion patterns will have an advantage when spend scales.

For brands, the strategic question is funnel positioning. The format's high intent suggests upper-to-mid-consideration placement: users are actively evaluating options, which is more aligned with driving consideration than brand awareness or direct response. The brands positioned to benefit most are those in software, financial services, travel, and professional services — categories where ChatGPT recommendations already influence decisions and where a sponsored presence can feel additive rather than intrusive.

The Competitive Signal

The $60-to-$25 CPM compression is the market's first honest feedback about ChatGPT advertising's equilibrium value. It's telling OpenAI: you built something interesting, but demonstrate conversion performance before we sustain a premium for it. That is exactly the right signal to receive at this stage of a new format's development, and how OpenAI responds over the next 12 months will determine whether $2.5 billion is a stretch goal or a floor.

The format has structural advantages over every existing advertising surface. The intent quality is as high as premium search. The user trust is higher than any current digital channel. The creative format matches how people actually want to receive recommendations — in context, conversationally, after the AI has already proven useful. Those three characteristics together are, in theory, the highest-value advertising surface ever built.

The gap between "in theory" and "in practice" is where the CPM compression lives. Closing that gap requires conversion data, advertiser performance benchmarks, and enough at-bat experiences across categories to move brands from test-budget experimentation to sustained channel allocation. OpenAI has achieved the first step: a real advertising business, generating real revenue, with real advertiser participation. The next twelve months will determine whether that's the beginning of something genuinely new, or a premium format that finds its ceiling at a fraction of its stated ambitions.

Takeaway: ChatGPT's advertising business reached $100M ARR faster than any new format in digital advertising history. The CPM compression from $60 to $25 in ten weeks is not a failure — it is the market finding its floor before scale. The CPC launch and self-serve Ads Manager open the platform to the performance advertisers who will determine whether the format delivers the conversion ROI needed to sustain a $2.5B 2026 revenue target. For brands, the current window is an early-mover opportunity in a format that hasn't yet established its performance benchmarks — meaning higher risk and potentially higher reward than established channels. For OpenAI, the advertising pivot is the clearest signal yet that the company is building toward a revenue future that no subscription model can fund alone.