A number is worth sitting with: ChatGPT’s ad program crossed $100 million in annualized revenue within two months of its self-serve launch, at roughly $60 CPM with over 800 million weekly users behind it. Analysts have called it the fastest-growing new ad platform since TikTok’s early days.
Meanwhile, Perplexity walked away from advertising entirely earlier this year, choosing to stay ad-free and lean into subscription revenue instead, positioning itself as the trust-first alternative. Two AI platforms, two opposite bets, both made in the same few months.
Why the growth rate matters more than the total number
$100 million in annualized revenue is still a small fraction of what Google or Meta generate. But the speed of the ramp is the signal, not the size. It tells you that advertiser demand for AI-native inventory existed the moment a credible, brand-safe placement became available, before the measurement tooling had even caught up to justify it on a spreadsheet.
That is a pattern marketing leaders have seen before: early movers on a fast-growing channel tend to get disproportionately cheap inventory and disproportionately valuable platform learning, precisely because the channel has not yet been priced up by the crowd that waits for perfect attribution data.
What this doesn’t mean
It does not mean you should reallocate meaningful budget away from channels with proven attribution. Truist’s own analyst estimate put OpenAI’s 2026 ad revenue under $1 billion, with the real inflection point projected several years out. This is early-innings growth, not a channel shift you need to bet the quarter on.
What $60 CPM actually tells you
The $60 CPM figure is worth unpacking rather than skimming past. That’s a premium rate compared to typical paid social or programmatic display, which usually run well below that. A premium CPM this early in a channel’s life typically means one of two things: either the inventory is genuinely scarce relative to demand (advertisers are bidding it up because the placement works), or the platform is deliberately pricing high to filter for advertisers willing to pay for brand-safe, high-intent placement rather than opening the floodgates to programmatic-style volume. Early signals suggest it’s closer to the former — a conversational interface where the ad appears inside an active research query is a fundamentally different placement than a banner competing for attention against a feed, and advertisers are pricing that difference in.
That matters for how you think about testing budget. A high CPM channel isn’t necessarily an expensive channel on a cost-per-outcome basis if the intent signal is strong enough — it just means your test needs to be structured around conversion quality from day one, not raw impression volume. Testing with a tiny budget purely to “learn the platform” without a real conversion goal attached will mostly teach you that CPMs are high, which you already know.
What it does mean for your planning
Put a placeholder in next year’s budget planning conversation now, even if the line item is small. Platforms that hit this kind of growth curve tend to tighten availability and raise minimum spend requirements as they mature. The advertisers testing now, even modestly, are the ones who will have platform relationships and creative learnings in place before it gets more competitive and more expensive to enter.
For a growth-stage B2B SaaS team specifically: this is a channel to watch and lightly test, not one to build a Q3 plan around yet. Structure any early test around a real conversion event, not impressions, and revisit the measurement question every quarter, because it is the piece most likely to change fastest.