Meta is adding a soft paywall to its smart glasses, and it starts with a number: Conversation Focus will be limited to three hours per month for non-subscribers. The company says this is coming, and it is tying that cap to whether you pay $19.99 a month for a Meta One Premium subscription.

Meta even tries to frame it as not-a-paywall. In a help article, the company insists it will not require a subscription to use your glasses, period. It is calling the change “rate limits” for certain AI features, while still making the practical outcome obvious: if you want more Conversation Focus time, you pay.

Here is the key detail that makes this more than a minor product tweak: even premium subscribers will only get 15 hours of Conversation Focus per month under that “rate limit,” according to the same framing. In other words, Meta is not selling “unlimited AI.” It is selling a higher allowance, with tighter caps still in place for everyone. That is a big difference from the fantasy most users have when they buy hardware and assume the core experience will be theirs.

Why would Meta do this? Because metering AI features is one of the easiest ways to connect costs to revenue in a market where AI usage can spike unpredictably. Conversation Focus is the kind of feature that can keep running in the background or be used repeatedly, which matters to anyone budgeting inference, compute, and backend services. If the AI experience is delivered through server-side processing, every extra minute can carry a real cost. Rate limits are the blunt instrument companies use when costs show up faster than subscription plans can scale.

There is also a second, more tactical reason: subscriptions are a cleaner business lever than one-time hardware upsells. Meta’s help article says it will not require a subscription to use the glasses, but the structure is still monetization-by-throttling. A user can technically keep using the device, but if the “AI feature” is a core part of the value proposition, cutting the usage hours pressures behavior. That is how “soft paywalls” work in practice: the door stays unlocked, but the most relevant room has a metered occupancy sign.

This matters to decision-makers because it changes what “ownership” means in consumer tech. Smartphones taught the market to accept carrier plans and app purchases. Streaming taught the market to accept throttling, bundling, and tiers. Now smart glasses are walking into the same culture, but with a twist: the subscription is not just for content. It is for access to AI functionality on a device you already own. That combination creates churn risk and customer trust risk, especially if competitors offer different terms or if users interpret the cap as a bait-and-switch.

Regulatory and policy pressure tends to follow these narratives. Even if the company labels the change as “rate limits,” regulators and lawmakers will still look at whether the user is being charged to access features they expected. The wording “rate limit” may be chosen because it reduces the claim that the company is blocking basic access. But executives should watch how regulators view the actual user experience: limited feature time unless you pay is functionally paywalled access, just with friendlier language.

For Meta specifically, the move is also a test case for how hardware ecosystems will monetize AI over time. If Meta can attach recurring revenue to an AI feature and keep the baseline device usable, it sets a template other companies will notice. For boards and senior product leaders, the question becomes: are you designing for stable usage expectations, or designing for controlled, monetized consumption? This isn’t only about smart glasses. It is about the broader AI product pattern that is emerging across consumer devices, where “the app” is partly software and partly a service, and the service has costs and leverage.

Strategically, the stakes are straightforward. If you are building or investing in consumer AI hardware, your commercial model will likely face the same tension: users want the premium AI experience consistently, while providers need predictable spending. Meta’s answer is a compromise that looks like a compromise on paper and feels like a paywall in the daily experience. The risk is that trust erodes faster than subscriptions grow. The opportunity is that, when managed well, metering turns variable AI costs into durable revenue.

For executives in adjacent roles, the takeaway is hard to ignore: AI features are becoming the new battleground for subscription strategy, and “we are not paywalling, we are rate limiting” is likely to become a common line. The numbers Meta has picked, three hours for non-premium and 15 hours even for subscribers at $19.99 per month, suggest how quickly “access” can be converted into “allowance.”