Chat is dead: OpenAI's super app and the $100B ad question

A senior OpenAI employee told the Financial Times that “chat is dead.” Source The company is rebuilding ChatGPT into a “super app”: coding tools, agents, and partner services bundled into one product, in what the FT calls its biggest overhaul since the 2022 launch. Source That sentence should stop any marketer mid-scroll, because the chat box is exactly where OpenAI told investors a $100 billion advertising business would grow.

What OpenAI is actually building

The pitch is an agent that does the work rather than a chatbot that answers questions. Chief Product Officer Thibault Sottiaux, who now has ChatGPT, Codex, and other product teams merged under him, described it this way: “It will transcend the actual surface … what we’re building towards is where you have your own personal agent that is capable of helping you … across everything in your life, be it personally or at work.” Source The interface stops being a fixed text box. The model is meant to work out what you need and act, on mobile, desktop, web, or in the car.

The redesign leans on the coding product Codex, whose weekly users have grown sixfold to more than five million since February, and steers people toward partner services such as Canva and Booking.com. Source Bets that do not fit are being cut: the Sora video app is gone and in-chat checkout has been shelved. Source The direction pulls OpenAI closer to Anthropic’s business-first stance and toward profitability ahead of an IPO. Source

The ad plan was built on the surface that just died

Two months ago the framing was different. Axios reported that OpenAI told investors to expect ad revenue of $2.5 billion in 2026, then $11 billion, $25 billion, and $53 billion, reaching $100 billion by 2030. Source The early signal was real: after ads went live in ChatGPT in February, the pilot crossed $100 million in annualized run rate within weeks and signed more than 600 advertisers, working with ad-tech firm Criteo and minimum budgets of $50,000 to $100,000. Source

Read the two numbers together. The $100 billion target assumes OpenAI reaches about 2.75 billion weekly users, up from roughly 900 million in February. Source Annualized run rate is a forward projection from a few weeks of pilot data, not money earned. So the ad business rests on two assumptions at once: that users keep multiplying, and that they keep meeting the product through a chat surface where a labeled ad has somewhere to sit. The super app pivot puts a question mark over the second one.

A personal agent that books your travel and edits your design does not present a feed of conversational replies with room for sponsored slots between them. If the primary interaction becomes the agent acting on your behalf, the obvious placement disappears. That does not mean the advertising disappears with it.

Where the money goes in an agent

It moves into the agent’s choices. Watch the detail in OpenAI’s own plan: the super app “steers users toward partner services such as Canva and Booking.com.” Source That is the commercial layer, and it has shifted from a slot you can see to a recommendation you cannot audit. When the agent picks which hotel to book or which tool to open, the question of who gets surfaced is worth more than any banner ever was, and it is far less visible to the person it influences.

This is the same gravity we described in AI agents don’t see your ads: the unit of value moves from the impression to the interaction. The super app makes it concrete and puts OpenAI’s own roadmap behind it. Google is building the same way, with sponsored listings and Direct Offers inside AI Mode and checkout embedded in Search. The placement that matters is inside the answer or the action, not around it.

For marketers, this rewrites the brief. Optimizing a sponsored conversational ad assumes there is a conversation to sponsor. Planning for an agent assumes something harder: being the option an agent selects when it acts. That favors clean, parseable content an agent can cite, structured product data it can act on, and partner integrations that put you on the shortlist before the agent decides. It looks less like buying media and more like the routing fight we covered in ASO after search — earning the pick inside a router you do not control.

One honest caveat. OpenAI has not said the ad business is cancelled, and a super app with 600-plus advertisers and partner deals has plenty of commercial surface left. The pivot does not kill the ad plan. It moves it somewhere harder to see, harder to label, and harder to measure, which is exactly the part a marketer should be planning for now.

TL;DR

• OpenAI is rebuilding ChatGPT into an agent-led “super app” and a senior employee told the FT that “chat is dead.” Source
• Its ad plan, reported by Axios, projects $2.5B in 2026 rising to $100B by 2030, and assumes about 2.75 billion weekly users. Source
• That plan grew out of the chat surface the company is now redesigning, which puts a question mark over where labeled ads sit. Source
• The commercial layer moves into the agent’s choices: the super app already steers users toward partners like Canva and Booking.com. Source
• For marketers the goal shifts from buying a slot to being the option an agent selects when it acts.