Meta and OpenAI are picking up the pace again in targeting the consumer AI market with personal AI assistants, but critics say the limits to profitability remain.

A dilemma of operating costs that are higher than demand is a key issue in consumer AI businesses, TechCrunch reported on Sept. 30 local time.

Consumer AI has recently been drawing attention again. Meta's personal AI assistant "Muse" and mascot "Jolly" recorded an unexpected hit, and OpenAI's "Dots," released on Sept. 29, is also built around the same personal assistant concept. Instinct, which handles tasks such as travel bookings, restaurant reservations and subscription cancellations, was valued at $10 billion.

But the pace of growth in users paying to use these AI agents is gradual. According to PNC Research, consumers paying for AI services accounted for 2.2 percent as of May, and average monthly spending was $31. The assessment is that even as model performance improves sharply, the number of paying users and the scale of spending have not risen clearly.

The break-even point also remains far off. Even if an online service with 325 million subscribers, like Netflix, is used as a benchmark for market saturation, $34 per customer would limit annual revenue to $11 billion. That is less than one-third of OpenAI's operating costs.

Other surveys show a similar trend. Bank of America estimated in March that about 3 percent of U.S. consumers pay for AI. That was a 40 percent increase from the previous year. A Menlo survey in September produced more optimistic results, showing that one-quarter of adults use AI daily and half of them are paying users. But TechCrunch reported that AI operating costs are far higher than those of social networks or cloud services, making it hard to guarantee profits even with hundreds of millions of paying users.

Major AI companies shifting their focus from consumers to corporate customers is not unrelated to this. Dots was also launched as a personal assistant, but OpenAI also presented use cases for software engineers and agency creators.

Meta and Instinct are looking for separate revenue models. Meta has a personalised advertising base, easing pressure to monetise, and is also considering entering the enterprise market. Instinct is pursuing a plan to take commissions on purchases handled by its agent. But TechCrunch reported that the challenge remains that it is still not easy to scale a consumer AI business significantly without enterprise revenue.