OpenAI's plan to profitability: Ads

So… back in the long gone early days of 2019, Sam Altman (head of sales at OpenAI) said that his plan to turn profit was to build an AI so good that he could just ask it how to turn profit. Since then, one of two things must have happened:

• AI is now good enough to make OpenAI profitable
• Sam has fallen prey to the internet’s original sin: ads

The evidence

First: new feeds

Two new features built on OpenAI’s LLMs point toward ads as a source of revenue with the possibility of turning OpenAI into a profitable company that isn’t dependent on closing ever-larger rounds of investment money. The new features are the Sora app and the Pulse thing that will give you a morning update based on what’s important to you.

Both are feeds (i.e., algorithmically curated lists of things a system wants you to see), and feeds are great places to put ads in front of people. They are also personalized feeds which allow for personalized ads, which is how Google and Meta make all their money.

Second: new hires, new roles

On December 4 2024, OpenAI made Kate Rouch its CMO. Kate has previously worked at Meta and Coinbase, and at the time, some people saw this as a first step towards an ad rollout.

In May 2025 OpenAI announced that Fidji Simo would become CEO of Applications, and given her history at Meta, where she oversaw a bunch of feeds, basically, this also points towards ads in feeds.

Finally, in September 2025, OpenAI started looking for someone to be “[...] part of a new ChatGPT growth team tasked with building the backbone of OpenAI’s in-house paid marketing platform.”

Third: where’s the money, Lebowski?

Third evidence is of the negative kind. So far, OpenAI’s spend far outstrips its revenue from API customers, and they have gotten record smashing rounds of funding more times than any other company in the history of the world.

So they might struggle to get more money in by asking for it, and companies aren’t seeing good enough returns (if any) on the APIs for OpenAI to hike the prices up to cover the costs of compute, let alone make a profit.

So ads it is. And the good news is that they will be amazing.

We’re entering the second stage of enshittification

A quick summary of the three stages of enshittification:

• Stage one: users are having an amazing time
• Stage two: advertisers are having an amazing time <= “AI” is about to enter this stage
• Stage three: no one is having an amazing time, but the platform is making hay

In the second stage, the platform owner wants to attract advertisers, and the best way to do that is to lower prices compared to the competition. They can also allow for super fine-grained targeting. There will be inflated data, of course (rememeber the pivot to video from Meta), but people placing ads in OpenAI's feeds will get a lot of clickthroughs. OpenAI will probably help generate targeted ads using their own algorithms and do so at very low prices.

My guess is that global ad spend is a zero sum game at this time, i.e., a dollar spent on OpenAI is a loss for the Google Meta ad duopoly. And if any of you reading this work in a marketing department that has an expanded budget, please let me know. This means it’s not unlikely that we will see a price war, since Meta and Google ("Moogle") don’t want to lose customers now that they need money to build data centers the size of Manhattan. We will definitely see an increase in offers to try AI ad features from all three players.

Other things to expect

Any successful product built on ChatGPT will be copied by OpenA

This is what's colloquially known as a service being Sherlocked, named after the time Apple copied a popular independent app that did what Sherlock does nowadays.

It’s a classic play for those who manage to build a platform. Look at what becomes successful, and then copy it and/or block the original. Once more, Apple provides a perfect example where they first banned third-party apps from using the volume buttons to control camera apps, then implemented that feature in their own app.

Rapid decline in user trust

This goes for both Google and OpenAI, but mainly OpenAI. The Google SERP is already filled with irrelevant links that companies have paid for to get displayed near the top. So far, the only trust issue with ChatGPT is that it gets things wrong about 30% of the time. But we have all learned to ask for the “top five results, put in a table with links to the source next to the title and an excerpt” in our prompts as a way to mitigate this.

All of a sudden a chunk of those results will be irrelevant crap that companies have paid for to get displayed near the top. Soon, every question about healthy friday snacks will be a list of how to best prepare some branded chocolate cookies.

Bubble goes POP

If ads don't work out, there’s a very real chance that OpenAI could drag the entire world economy into a crash. The big tech companies that are propping up the US economy can’t keep passing the same hundred billion dollar bill around and pretending that it generates new revenue each lap it makes.

So let’s all hope that OpenAI's ad rollout is smooth and that there will all of a sudden be a trillion more dollars in ad spend over the coming year.

The good news

The good news is that anyone who spends money on targeted ads will likely get better targeting for less money, sometime around Q2 2026. The ads will automatically rewrite themselves to align with the tastes of increasingly well defined demographics. And the demographics will be automatically “twinned,” split, and merged based on data collected not only from the “users” but also from your competitors.

I have no guess as to how long this will last though.