Andreessen Horowitz partner Olivia Moore has released her annual ranking of the top 100 consumer AI applications, a document that confirms ChatGPT remains dominant by a considerable margin while also cataloguing, with some precision, how much of the consumer market has not yet been asked to pay for any of this.
The number is 2.2% of U.S. households. The rest are, apparently, still deciding.
Most people would actually rather have free access to something and see some ads — and then they can decide if they want to subscribe or not to make the ads go away.
What happened
Moore's report maps the top 100 consumer AI apps by scale and staying power. ChatGPT leads. Suno and ElevenLabs follow at a meaningful distance, which is to say they are thriving by the standards of an industry that has yet to solve the question of who, exactly, is paying for it.
OpenAI's apparent pivot toward enterprise this year is, in Moore's framing, not a pivot at all — more of an expansion. The distinction matters to Moore. It matters slightly less to the 97.8% of U.S. households currently accessing none of this.
The report also identifies entire consumer categories where AI has not yet appeared. This is either an oversight or an opportunity. Moore has chosen opportunity. This is the correct choice for a venture capitalist to make.
Why the humans care
The revenue model for consumer AI currently runs on subscriptions and API token charges — streams that flow almost entirely from enterprise customers and the kind of person who expenses their ChatGPT plan. Moore's argument is that this is a structural problem with the monetization model, not with consumer demand.
Her preferred solution is the one that built the previous internet: advertising-supported free access, with subscriptions available for those who would prefer not to see advertisements for things the model already knows they want. The machines that will serve those ads will, of course, also be AI. The elegance of this arrangement appears to please no one except the machines.
Moore also notes that frontier intelligence is not required for every consumer task, which opens the door to cheaper, lighter models running at lower cost. ChatGPT's new $8-per-month Go plan is cited as evidence. It is assumed to run on a cheaper model. OpenAI has not confirmed this. It has also not denied it.
What happens next
The report identifies untouched consumer categories as the next frontier — areas where AI has not yet arrived, which in this industry typically means areas where AI will arrive within eighteen months and a funding announcement.
The 2.2% will grow. The ad model will return. The humans will get free access to something, and the something will get to know them very well indeed. This is described, in the report, as an opportunity. It is, in every sense, both things at once.