I wrote this week about the murky economics of consumer AI, which some see as a crisis and others as an opportunity.
Olivia Moore, a partner in consumer AI at Andreessen Horowitz, released a report on the top 100 consumer AI apps on Monday. ChatGPT remains the biggest player, but smaller players like Suno and Celebrities are also showing real staying power. What’s even more interesting is what we don’t see. Moore’s report details six consumer categories that appear to be completely unaffected by AI.
Moore believes there is a huge opportunity for consumer AI, especially if the industry can tap into revenue streams beyond just subscriptions and API fees. On this week’s call, I spoke with her about the unusual economics of consumer AI and why the category is still in its infancy.
This interview has been edited for length and clarity.
It’s an interesting moment for reporting on consumer AI, as we’ve seen OpenAI pivot to the enterprise this year. This has led to a lot of pessimism, including mine, about the overall revenue situation for consumer AI. But do you think you’re more optimistic?
I agree that OpenAI is back for the enterprise. To me, it’s not necessarily a pivot because they’re still releasing a lot of products for consumers. It’s even more of an extension. But as much as I love consumers, I don’t blame them. This is because almost all of AI revenue to date has come from subscriptions and subsequent token usage. And it’s much more concentrated on the enterprise and prosumer side.
How can consumer AI products change that revenue problem? The State of Markets report reminded us that only 2.2% of US households pay for AI. Does that number need to go up before the market becomes attractive?
Actually, I don’t know if I want the number of users to increase. I’m more interested in returning to a world where consumers can monetize instead of paying for subscriptions out of pocket.
I think it’s very easy in Silicon Valley to say, “Well, I want to pay for the product,” because there are people with high incomes who have corporate cards that pay for the tools. But I think most people actually want to access something for free, see ads, and then decide whether to subscribe or not and hide the ads.
Another challenge is the marginal cost of AI services. This is still much higher than traditional internet services such as Facebook and Google Search. Do many of these services, whether open source or other lightweight models, find ways to keep costs down?
Yes, I think you will see an improvement. Currently, ChatGPT also has a Go plan, which costs just $8 per month, but I think it works on a cheaper model. Not all tasks necessarily require frontier intelligence, so cheaper models may be needed, especially for some consumer use cases. It could also mean more consumer companies building on an open source model. I’ve been hearing this a lot from founders lately, and I feel like that’s starting to change.
The problem is that the users who are making money today are doing coding and other technical automation and probably need frontier intelligence. As more people build for consumers where the models are not products, I think lower priced models will be used more often.
It’s unusual to talk about technical automation in consumer services, and this feels like another unique feature of AI. Are the lines between consumers and businesses starting to blur?
I wrote an article about this, probably about a year ago, trying to figure this out called “The Great Expansion.” Before AI, we had consumer-first companies like Canva that often took six to seven years to add a team or enterprise plan. We are now seeing companies such as Gamma, Celebrities, and Cursor start out as consumer-facing businesses and become majority businesses within 18 months.
I would argue that most of what we think of as consumer AI is prosumer AI. This really shows up in the revenue list, and there are basically about three categories that power users spend their money on. There are also product building apps like Lovable, Replit, and Fal. There’s Product Marketing, which is an AI ad generator like Higgsfield and HeyGen. Then there are common work managements like Manus, Fireflies AI, and Granola. These are things that consumers are paying for in the first place, but I would argue that they are not consumers in the sense that we think of them pre-AI.
There are also all blank categories like social apps, dating apps, marketplaces, retail, travel, finance, health, and more. There are no participants in the top 100 list for these categories. This is quite surprising. And I think that’s what we need to see over the next six months.
So maybe we haven’t really seen true consumer AI yet?
It’s certainly very fast.
If you make a purchase through links in our articles, we may earn a small commission. This does not affect editorial independence.
