Olivia Moore sees consumer AI’s next test beyond subscriptions

Revenue beyond subscriptions

Moore is not convinced that consumer AI needs a larger pool of paying subscribers to become an attractive market. She would rather see products earn revenue in more ways than charging users directly.

That matters because AI companies have so far relied heavily on subscriptions and token usage, while serving enterprise customers and advanced users brings much of the cost. OpenAI has become more active with businesses again this year, but Moore does not see that as a full retreat from consumers: the company continues to release consumer products. To her, it looks more like an expansion.

For many people, Moore argued, an ad-supported service may be more appealing than paying upfront. Users could then choose to subscribe to remove ads. The model is familiar from the internet, but AI’s higher serving costs make it harder to copy.

Cheaper models, different products

The cost of serving AI products remains much higher than for established internet services such as Facebook or Google Search. Moore said those costs are already falling. ChatGPT’s $8-a-month Go plan, she suggested, may use cheaper models. Not every consumer task needs a frontier model, and more companies can build products on open models.

For now, much of the revenue comes from people using AI for coding and other technical automation, work that may genuinely require advanced models. Moore expects cheaper models to spread as more companies build consumer products where the model itself is not the product.

The consumer label can also be misleading. Moore’s view is that many products described as consumer AI are really tools for advanced users. The companies those users pay for cluster around three jobs:

These tools may start with individual users, but that does not make them consumer products in the traditional sense.

The categories still missing

Moore described the shift from consumer to business as faster than it used to be. Before AI, companies such as Canva could start with consumers and wait six or seven years before adding team or enterprise plans. Gamma, ElevenLabs and Cursor, by contrast, began as consumer products and became predominantly enterprise businesses in less than 18 months.

That compression makes the gaps in Moore’s list more interesting. Social apps, dating, marketplaces, retail, travel, finance and health have no representation among the 100 leading apps. Moore expects products in those categories to appear within the next six months.

I think that forecast is the more useful test of the consumer AI market than another count of subscriptions. The current winners largely help people make or manage work; the next wave would have to make AI useful in ordinary consumer services, where the model is less visible and the revenue model may matter as much as the technology.