OpenAI's $100 billion ad forecast is fantasy. Here's the math that says so.

OpenAI has reportedly told investors that advertising could grow into a US$100 billion global business by 2030. eMarketer's forecast for the entire US chatbot ad market that same year, covering ChatGPT, Copilot, Google AI Mode and Alexa for Shopping combined, is US$5.41 billion.

Adjust for the fact that one number is global and the other is US only. Be as generous as you like. The gap still doesn't close. What it actually shows is the real state of chatbot advertising today: enormous usage, and monetisation that isn't anywhere near catching up.

Ask ChatGPT to book you a flight and watch what happens. It'll cheerfully suggest options, get a detail wrong, and send you off to actually book it somewhere else. That's the product advertisers are being asked to bet a media budget on: something famous for confidently inventing things now pitching itself as a measurement platform.

Whether OpenAI hits its number isn't really the interesting question. It won't, not on this timeline. What's worth asking is what ChatGPT becomes instead, and whether a generation now forming its search habits ever gives it room to become something bigger.

Usage isn't revenue

Few products manage to change a basic human habit. ChatGPT changed how millions of people look things up, in under three years. But shifting user behaviour and shifting advertiser behaviour are different fights, and OpenAI has only won one of them so far.

Advertisers don't move budget because a platform is popular. They move budget when a channel proves it delivers on cost, reach, measurement and return. Google wins that argument today not because anyone loves it, but because it's proven. Every dollar in Google Ads carries years of attribution behind it. A dollar in ChatGPT ads right now is still closer to a bet.

To be fair to OpenAI, this is roughly what Google itself looked like in 2001. Massive usage, no real ad business yet, skeptics everywhere. The difference is that Google built its ad model on intent data it already had at unmatched scale, search queries with commercial signal baked in. ChatGPT's usage skews toward answers and conversation, which is a much harder signal to monetise cleanly without wrecking the product's credibility. That's not a reason to write ChatGPT off. It's a reason its 2001-to-Google-2004 timeline could take a lot longer than five years.

The revenue numbers back this up. OpenAI reportedly expects around US$2.5 billion in ad revenue this year. eMarketer puts the entire US chatbot ad category at under US$1 billion. Not a clean comparison, but the direction is consistent: the market is pricing this a lot more cautiously than OpenAI is pricing itself.

A better test than "will it beat Google"

Most coverage keeps asking the wrong question. It isn't a binary between ChatGPT and Google. For anyone holding an actual media budget, a better approach is running any emerging AI platform through three checks before spend goes near it.

Cost: is the cost per outcome close to what proven channels deliver, or are you paying a premium for novelty? Measurement: can you attribute results with the confidence you have in Search or Meta, or are you trusting the platform's own numbers? Reach: is the audience genuinely incremental, or the same people you already reach through a different door?

ChatGPT fails two of three today. Cost data is thin. Measurement is immature, and there's no independent fraud or brand-safety tooling built out yet, which matters more than people testing small budgets tend to notice. For advertisers targeting anyone over 30, reach mostly overlaps with Search and Meta already. None of that makes it worthless. It means it belongs in the test budget, not the scale budget. Treat it the way performance marketers treated Bing a decade ago: 3 to 5 percent of budget, scaled only once cost parity holds for two straight quarters.

Google doesn't need you to switch. ChatGPT needs you to quit.

That's the whole battle in one line. Google brought generative AI to people inside the products they already open every day. OpenAI is asking people to change where they start. Historically, that's the harder trade to win.

The wildcard: the younger and newer generations

All of this assumes today's habits hold steady. They might not, and it has nothing to do with ad tech.

It's evidence a generation is forming its first search habits right now, and habits formed early are the ones that stick. Google's real moat was never the algorithm. It's that most people never consciously chose Google. By the time they were old enough to form a habit, it already was one. If a real share of under-25s are forming that same reflex around ChatGPT instead, the three-part test above stops being fixed. It shifts as this generation ages into real spending power.

Google clearly sees this coming. AI Overviews, AI Mode and Gemini all exist so a 17-year-old never has a reason to leave Google's ecosystem for an AI-native answer. That's the actual fight. Not chatbot versus search engine, but who becomes the automatic reflex for a generation that hasn't locked in its habits.

A prediction, and three things to actually do

Here's a number worth holding me to: ChatGPT's global ad revenue lands under US$8 billion in 2027, well off pace for the $100 billion 2030 target, unless OpenAI ships a fundamentally better measurement and attribution layer than it has today.

In the meantime, three moves worth making:

Fund a test line, not a strategy overhaul. Put 3 to 5 percent of a channel's budget into ChatGPT ads with a hard rule: it earns more only once cost per outcome sits within roughly 20 percent of your Search benchmark, for two consecutive quarters.

Track SEO/GEO visibility now, regardless of spend. Your brand is already showing up, or not, in AI-generated answers across ChatGPT, Gemini and AI Overviews. Audit it the way you'd audit organic search visibility, and be skeptical of anyone selling you a fixed playbook for it. The rules aren't stable yet, and most "GEO strategies" on the market right now are guesses dressed up as expertise.

Break out under-25 performance separately from the rest of your reporting. If younger audiences are quietly forming different habits, blended data will hide it. Split it out and you'll likely see the shift a year before your competitors do.

Google owns today's search habit, comprehensively and for good reason. The fight isn't over that habit. It's over the next one, and it's being decided right now by people too young to fit most advertisers' target demographics yet.