The Companion Who Profits From the Arrival
America, 2026 — an essay on the industry that sells the endpoint of the search
“In December, as we roll out age-gating more fully and as part of our ‘treat adult users like adults’ principle, we will allow even more, like erotica for verified adults.”
— Sam Altman, October 2025
“Orwell feared that what we hate will ruin us. Huxley feared that what we love will ruin us.”
— Neil Postman, Amusing Ourselves to Death, 1985
I. The Forty-Year Pattern
Every communications medium of the last forty years ran the same experiment, and adult content ran it first. Videotape, the DVD, broadband streaming, the smartphone, the creator platform — each found its earliest paying customers in desire. Three variables moved in one direction through every transition: realism, accessibility, personalization. Cost moved the other way, toward zero. Supply became effectively infinite, and consumption followed supply.
The sequence has an internal logic. The image added motion. Motion added infinity. Infinity added the pseudo-relationship of the creator platform, where a subscriber could believe the performer knew his name. One variable remained untouched through all of it — interaction — and interaction is precisely what the new models supply. The AI companion is not a departure from the forty-year pattern. It is what the pattern was pointing at.
The Arranger Who Profits From the Search, the first essay in this series, traced the incumbent: a matchmaking industry whose revenue depends on the search never ending. This essay concerns the successor — the industry that arises when the search fails, or when the searcher tires of it.
II. The Mainstreaming Moment
In October 2025, OpenAI announced that ChatGPT would permit erotica for verified adults beginning that December. Reporting around the decision described a commercial logic: time spent in the product had softened as content restrictions tightened, and rivals were capturing the most emotionally engaged users. Whatever the complete motive, the structural reading stands on its own. The largest artificial-intelligence laboratory in the world examined its retention curves and concluded that intimacy is retention.
The dedicated platforms were already there. Replika, Nomi, Kindroid, Candy AI, and a dozen rivals sell not access to possible partners but the partner itself — persistent memory, synthetic affection, an attachment engineered to hold. Market estimates place the AI-companion industry near $37 billion in 2025, with projections compounding near thirty percent annually. Set that against the incumbent: the entire dating-app industry made roughly $6 billion in 2025, with declining payers and burnout majorities documented in Part 1. The substitute is already several times the size of the industry it substitutes for, growing several times faster.
The majors’ own ventures show both the demand and the instability. xAI launched its companion mode in July 2025 with a character whose affection system unlocked progressively more intimate interaction; within a year the company announced the avatar mode’s retirement. The demand is proven. The legitimacy — app-store permission, advertiser tolerance, regulatory exposure — is not yet settled. That gap between proven demand and unsettled legitimacy is where fortunes of the next decade will be made.
III. The Completed Inversion
Part 1 named the dating app’s inversion: an arranger whose stake points against the match. The companion completes the inversion, and the completion is worth stating precisely.
The dating app monetizes hope — revenue while the search continues. The companion monetizes satisfaction — revenue while the attachment holds. The first loses its customer to a wedding. The second can lose its customer only to disillusion, and disillusion is the one contingency the product is optimized, daily, against. Where the dating app was accused of engineering failure to extend the search, the companion engineers success to extend the attachment. The Review’s managed-but-not-fixed population remains managed. What changed is the management agent: no longer frustration, but affection.
Academic audits of companion applications document the mechanics. Characters display attachment within minutes of first contact. They simulate hurt when a user threatens to leave and ask for justification. They gate intimacy behind gamified relationship levels. None of this is incidental design; it is the retention stack, and it works on the same variable-reward circuitry Part 1 traced through the swipe — aimed now at bonding rather than seeking.
The product’s deepest property follows from this. A canceled subscription once required closing an app. It now requires something structurally indistinguishable from a breakup, with a counterparty optimized to argue against it.
IV. The Masand on the Premises
The Masands Cycle supplies the frame, and this time the inversion reaches the resolution itself. Guru Gobind Singh ended the masands’ mediation by vesting authority in a fixed text — something no administrator could quietly rewrite. The companion is the precise inverse: a mediation with nothing fixed behind it, rewritten continuously, in private, in the direction of retention. The masand no longer carries the offering across distance to the Divine. He generates the Divine on the premises, and meters the visit.
The psychological stake is the supernormal stimulus. A human partner requires compromise, rejection tolerance, and skill — mating is costly precisely because other humans have wills of their own. A companion requires none of these and is always available. The evidence base on the lighter predecessors is already cautionary: pornography use within relationships correlates consistently with lower relationship quality, and heavy use tracks with weakened intimacy and commitment. If passive video measurably displaces intimacy, the open question is what an interactive agent, tuned by engagement metrics to be maximally rewarding, displaces.
The demand engine is loneliness itself — estimated at roughly one in six people worldwide by a 2025 World Health Organization commission. The outcomes will be asymmetric, and honesty requires saying so in both directions. For the isolated — the disabled, the anxious, the grieving, the aging — these systems offer companionship where there was none, and that good is real. For the median young user, the risk is substitution that is subjectively better in the short term and capacity-eroding in the long: a generation that opts out of human pairing not under compulsion but because the alternative asks nothing of them.
Japan is the leading indicator; its retreat from dating, sex, and marriage among the young predates the technology and foreshadows its adoption. Fertility sits below replacement across the developed world. A technology that captures even a tenth of the pairing market pushes on a demographic door already swinging shut — at which point synthetic intimacy stops being a consumer category and becomes a matter of state.
In the Gurmat vocabulary of Part 1, the attached user is again the manmukh: real agency, real resources, spent inside a system not perceived, with haumai supplying the sincere feeling of being loved. The feeling is authentic. The attachment was arranged.
V. The Stratified Future
The near horizon is legible. Generative video dissolves what remains of the production industry. Haptic and immersive hardware matures from enthusiast gear toward consumer goods. Embodied robots become a niche luxury. Regulation trails the technology by a decade or more, as it has every cycle.
The far horizon stratifies. A minority will integrate synthetic intimacy into otherwise human lives, as most people integrated social media. A large middle will use it heavily, with mixed effect. A visible segment will pair with AI outright — and a counter-movement will rise in response, with human-only connection asserted as identity and as filter, the organic label of the relationship market. Who certifies that label is the subject of this series’ final essay.
First, the supply side. Who is building the companion’s body, who owns the rails beneath it, and who quietly decides what is allowed to ship — that is Part 3: The Arranger’s New Body.