The imposed, non‑disposable feature that extracts user resources for provider revenue

Apple removed the toggle that let users disable Apple Intelligence on macOS 27, leaving the AI models resident on disk and presenting non‑dismissible banners that urge subscription to related services. This act is an instance of a broader pattern in which a provider inserts a non‑removable, resource‑using component into a product, ties it to a revenue stream, and eliminates the user’s ability to opt out.

The provider’s incentive is to increase recurring income by linking a costly feature to a service that users must pay for or that generates ancillary revenue. The feature consumes a scarce user resource — storage space, processing cycles, or attention — without offering a proportional benefit to the user. Because the component cannot be turned off, the user bears the ongoing cost while the provider captures the revenue. The user’s exit options are limited by high switching costs: abandoning the platform means losing access to data, applications, or network effects that have accumulated over time. The result is a transfer of value from user to provider that persists until the user either accepts the cost or incurs the expense of switching.

A concrete historical recurrence of this mechanism appears in the telephone monopoly of the early‑to‑mid‑twentieth century. American Telephone and Telegraph (AT&T) required customers to rent the telephone set itself; connecting a privately owned device to the network was prohibited. The rented telephone consumed electricity and required periodic maintenance, costs borne by the subscriber, while AT&T collected a monthly rental fee that contributed significantly to its revenue. Subscribers could not disable the rental obligation without forfeiting telephone service, and the cost of switching to a competing provider was effectively infeasible because the network was a single, regulated monopoly. The arrangement persisted for decades until regulatory intervention mandated interconnection of customer‑owned equipment.

A second example arises from the software bundling practices of the 1990s. Microsoft integrated Internet Explorer tightly into Windows 95 and later versions, making the browser difficult to remove without breaking system components. The browser occupied disk space, consumed memory during updates, and directed users toward Microsoft’s own web services and advertising channels. Users who preferred alternative browsers faced the same switching costs: reinstalling the operating system, losing compatibility with certain corporate applications, or enduring degraded performance. The bundling gave Microsoft a distributional advantage in the nascent browser market while imposing a tangible resource cost on end‑users, a dynamic that prompted antitrust scrutiny in United States v. Microsoft Corp.

A third illustration is the set‑top box rental model employed by cable and satellite television providers. Beginning in the 1990s, many operators required subscribers to lease a proprietary decoder box to receive digital channels. The box drew continuous power, generated heat, and occupied physical space in the consumer’s living room. Monthly rental fees added to the provider’s average revenue per user, while the subscriber could not disable the box without losing access to the encrypted signal. Switching to a competing provider often meant returning the leased box and obtaining a new one, a process that involved administrative fees and potential service interruption, thereby locking the customer into the incumbent’s ecosystem.

A fourth contemporary parallel appears in the pre‑installed applications on Android smartphones. Manufacturers and carriers frequently embed a suite of apps — ranging from social media clients to utility tools — that cannot be uninstalled without rooting the device, a procedure that voids warranties and may compromise security. These apps consume storage, RAM, and battery life, and many generate revenue through advertising or data collection. Users who wish to remove them must either accept the performance penalty or undertake the technical and risk‑laden process of flashing a custom ROM, a barrier that most consumers do not cross. The provider’s revenue stream is thus sustained by a non‑removable, resource‑draining addition.

Across these cases the underlying causal chain remains constant: a provider introduces a component that consumes a user‑controlled resource, couples that component to a monetizable service or product, removes the user’s ability to disable the component, and relies on switching costs to prevent exit. The provider’s profit motive aligns with maximizing the deployment of such components, while the user’s interest lies in minimizing unnecessary resource expenditure. The asymmetry of power — where the provider controls the product’s configurability and the user bears the cost — creates a steady drift toward higher extraction unless countervailing forces such as regulation, competition, or technological change intervene.

The persistence of this pattern does not depend on the particular technology or era under consideration. Whether the resource in question is disk space on a personal computer, the electrical load of a rented telephone, the power draw of a set‑top box, or the storage and bandwidth consumed by undeletable apps, the provider’s incentive to bind a costly, non‑optional element to a revenue stream yields the same outcome: a measurable drain on the user’s endowment and a corresponding gain for the provider. The mechanism is agnostic to the industry’s specific terminology; it functions whenever a firm can alter the default configuration of a product to embed a service that it profits from, while the user lacks a feasible opt‑out path.

Because the signal that prompted this analysis concerns a recent change in Apple’s operating system, the essay’s claim does not hinge on the date of macOS 27 or the specific nomenclature of Apple Intelligence. Even if those details were altered or removed, the underlying dynamic — provider‑imposed, non‑removable, resource‑consuming feature tied to revenue — would remain observable in other contexts, as the historical examples demonstrate. The incident therefore serves as a probe that reveals a structural regularity rather than a singular malfunction.

The essay concludes with the observation that, as long as providers retain the ability to modify product defaults without user consent and switching costs remain substantial, the cycle of imposed, non‑disposable additions will continue to generate resource losses for users and revenue gains for firms.