For a week, consumer AI looked like it was back. Meta’s Muse assistant and its plush mascot Jolly became an unexpected hit; OpenAI launched Dots, a cartoony personal agent chasing the same idea; and the startup Instinct reportedly reached a $10 billion valuation on the strength of agentic errand-running — booking travel, reserving tables, cancelling subscriptions. The bull case writes itself: agents have finally become reliable enough to be worth paying for.
The numbers say something else. In its semiannual State of Markets report, Andreessen Horowitz reproduced two charts from a PNC research note: as of May, just 2.2% of consumers were paying for AI, at an average of $31 a month. Bank of America found a similar figure in March — roughly 3% of US consumers paid for AI, up 40% year over year. A Menlo survey in September was sunnier, finding that a quarter of adults use AI daily and half of those users pay. Andreessen spins the data positively — “it’s still so early” — but on both charts the growth line looks stubbornly linear. The leap from GPT-5.2 to Astra is barely visible in it.
The problem is less revenue than cost. AI is unusually expensive to operate compared with lightweight predecessors like social networking or cloud computing. TechCrunch’s arithmetic: take Netflix as the benchmark for a saturated online service (325 million subscribers); $34 per customer would yield just $11 billion a year — less than a third of OpenAI’s operating costs. Even hundreds of millions of paying users do not guarantee breaking even.
That is why frontier labs have gone “gunshy” about consumer AI and pivoted toward the enterprise playbook. OpenAI’s enterprise bookings have reportedly doubled since July, and even the Dots launch had a strong enterprise angle, pitched for software engineers and agency creatives. Muse can lean on Meta’s ad-targeting machine and more patience; Instinct can take a cut of purchases made through its agent and skip the cost of training a frontier model.
But the underlying economics have not improved, and the conclusion is blunt: the ugly math of consumer AI puts a hard cap on how large a consumer-only company can grow without enterprise revenue. It is one of the few things in the industry that is not changing.
