With subscription tiers starting at $7.99 a month and a free tier preserved, Zuckerberg is betting that price will be the deciding factor in AI adoption
The AI tools market has a new pricing problem. Most serious players charge a premium for access to their best models, betting that users will pay up for quality. Mark Zuckerberg has a different theory.
Meta is testing a tiered subscription model called Meta One, with a Plus tier at $7.99 per month and a Premium tier at $19.99 per month. The company is also keeping a persistent free tier alive.
What Meta is actually building here #
The subscription testing, which began in May 2026, is one half of a two-pronged strategy. The other half targets businesses directly.
Meta launched AI business agents in early June 2026, designed to operate on platforms like WhatsApp. These use a consumption-based pricing model, meaning enterprise clients pay based on token usage rather than a flat fee.
Zooming out, Meta is essentially building two parallel businesses on top of its AI stack: a consumer subscription product competing with ChatGPT and Google’s Gemini, and an enterprise agent business competing with Microsoft Copilot and Salesforce’s AI tools. Both are designed to undercut the market on price.
The infrastructure bet behind the strategy #
Meta is projecting capital expenditures on AI infrastructure between $115 billion and $145 billion in 2026 alone. That figure is nearly double what the company invested the prior year.
To put that in perspective, $130 billion, the midpoint of that range, is roughly the GDP of Hungary.
Meta’s open-source Llama models are another piece of this. By releasing Llama openly, Meta encourages developers to build on its stack, creating an ecosystem that feeds back into platform adoption. It is the inverse of OpenAI’s proprietary approach.
What investors and competitors should be watching #
Meta’s social platforms already generate the cash to fund this expansion, which means the AI business can operate at lower margins, or even at a loss, for longer than a standalone AI company could sustain.
The risk is execution. The metaverse push, which consumed enormous resources earlier in the decade, did not produce the consumer adoption the company projected. AI is a more natural fit for Meta’s existing surfaces, particularly because the AI features are embedded in apps people already use daily rather than requiring them to strap on a headset.
Competitors should watch the WhatsApp enterprise play most closely. WhatsApp has over two billion users globally, and turning that distribution into an enterprise AI sales channel is a structural advantage that Microsoft and Salesforce would have to spend years trying to replicate.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our