Consumer AI has a money problem, and one of Andreessen Horowitz’s top consumer investors thinks the answer is to stop charging consumers. In an interview with TechCrunch AI, a16z partner Olivia Moore talked through her new report on the top 100 consumer AI apps. Her view runs against most of the industry: the real consumer AI market hasn’t started yet, and when it does, it probably won’t run on subscriptions.
That’s a bold call to make right now. OpenAI has moved back toward enterprise this year, and a State of Markets report cited by TechCrunch AI found that just 2.2% of U.S. households pay for AI. A lot of people are asking whether consumer AI can ever pay for itself.
What the Top 100 List Shows
ChatGPT still leads “by a mile,” according to the report. Smaller players like Suno and ElevenLabs are proving they can last. What stands out more is who’s paying. Moore says the spending clusters in three groups:
- Product building: Lovable, Replit, Fal
- Product marketing: AI ad generators like Higgsfield and HeyGen
- Work management: Manus, Fireflies AI, Granola
Individuals pay for these at first. But Moore argues they aren’t consumer products in the pre-AI sense. “Almost all of what we think of as consumer AI, I would argue, is prosumer AI,” she told TechCrunch AI.
The 18-Month Jump to Enterprise
This is where the business model has really changed. Moore points out that Canva took six or seven years to add team and enterprise plans. Now Gamma, ElevenLabs and Cursor start as consumer tools and become majority-enterprise businesses within about 18 months. She calls this “The Great Expansion.”
This matters because it explains why consumer AI revenue looks weak even as some “consumer” apps grow quickly. The ones that make money stop being consumer apps. The money moves to work budgets and corporate cards.
Ads Over Subscriptions
Moore’s most interesting point is that she doesn’t necessarily want more people paying out of pocket. “I’m more interested in getting back to a world where the consumer is monetized in a way that isn’t subscription dollars out of their own pocket,” she said. Her argument is that most people would take free access with ads and upgrade later, and that high-income Silicon Valley users, who often expense their tools, miss this.
The obstacle is cost. Serving an AI query still costs much more than serving a Facebook feed or a Google search. Moore sees that gap closing. She cites ChatGPT’s $8 Go plan, which she assumes runs on cheaper models, and says a16z founders are building on open source models more and more. “We don’t necessarily need frontier intelligence for every task,” she said.
There’s a catch, though. Today’s paying users do coding and technical automation, which probably does need frontier models. Cheaper models only win once more products target everyday consumers, where, as Moore puts it, “the model is not the product.”
The Empty Categories
The biggest finding is what’s missing from the list. Social, dating, marketplaces, retail, travel, finance and health have zero entries in the top 100. These are the categories that made the last generation of consumer internet giants, and AI hasn’t produced a breakout in any of them. Moore expects that to change over the next six months.
Looking Ahead: 2027 to 2029
Keep in mind who’s making these predictions. Moore is a venture investor, and her firm has a stake in consumer AI working out. a16z has published this list several times since 2023, and every edition has shown ChatGPT far ahead. Her six-month window for whitespace breakouts is aggressive. Still, her diagnosis lines up with the data.
If she’s right, here’s what the next few years could look like:
- Ad-supported AI goes mainstream. Free tiers paid for by ads, with subscriptions as an optional upgrade.
- Model choice becomes a cost strategy. Consumer apps send most tasks to small or open source models and save frontier models for the hard ones.
- Vertical consumer apps finally break through. Travel, health and finance are huge markets with no clear AI leader yet.
What Builders Should Do Now
- Plan for enterprise early. If your prosumer tool works, teams will show up quickly. Have team plans and admin controls ready.
- Know your cost per user. You can’t run an ad model if each free user costs more to serve than they bring in.
- Look at the empty categories. No incumbent AI app is in your way there, though that probably won’t last long.
The first wave of consumer AI was mostly professionals paying for productivity. The next one, if it shows up, will need to make money from people who don’t expense anything. The full interview is available at TechCrunch AI.