Meta is piloting a paid consumer AI product, and the fundamental problem is obvious to everyone except, apparently, Meta.
The company has spent two decades conditioning its users to expect everything for free, subsidised by advertising. Now it wants them to pay for a chatbot.
The pricing problem
Reports suggest the subscription could land around $20 a month, with a cheaper ad-supported tier. At that price point, Meta is competing directly with ChatGPT Plus, which has a substantial head start in consumer AI and a brand that is synonymous with the category.
The question is not whether Meta's AI is capable. It is whether it is capable enough for users to pay for it when they already have free access to Meta AI across Instagram, WhatsApp and Facebook.
To justify a subscription, Meta would need to offer functionality that is genuinely superior to the free version.
The alternative is worse: degrading the free tier to push users towards paying, which risks alienating the user base the entire advertising business depends on.
The neocloud temptation
Mark Zuckerberg has floated the idea of Meta becoming a neocloud, renting out surplus compute capacity to other companies. This should worry investors more than it excites them.
If Meta's best use for its enormous AI infrastructure is renting it to others, it means the company has overbuilt and cannot generate enough internal demand to justify the spend. That is not a growth story. It is a write-down waiting to happen.
The real test
Meta's AI subscription will be worth watching over the next six months, but the bar is high. Even converting a small percentage of its billions of users would generate meaningful revenue. The question is whether the product is compelling enough to convert anyone at all.
Free is a hard habit to break, especially when the company asking you to pay is the one that made it free in the first place.