The Subscription Trap Nobody Can Escape

How AI and Weight-Loss Drugs Turned Temporary Solutions Into Lifetime Sentences

How Tech and Pharma Discovered That Dependency Beats Competition Every Time

Eli Lilly’s weight-loss drugs and OpenAI’s language models share a business strategy that would have seemed dystopian a decade ago: creating customers who literally cannot function without their product. When patients stop taking Mounjaro, they don’t just return to their original weight – research from the New England Journal of Medicine shows 67% of weight returns within 12 months, often accompanied by metabolic disruption. When companies disconnect from AI systems, Deloitte’s 2024 “State of AI in the Enterprise” report found that 71% experience immediate operational failures because workflows have been redesigned around AI availability.

This isn’t market dominance through superior products. It’s engineered dependence that makes leaving more painful than paying.

The Subscription That Runs Through Your Veins

The pharmaceutical industry discovered something remarkable with GLP-1 receptor agonists: unlike traditional drugs that cure conditions or manage permanent disabilities, these create reversible dependencies. According to data published in JAMA Network Open in 2024, only 13.5% of patients maintain weight loss one year after discontinuing semaglutide (Ozempic/Wegovy). The Journal of Clinical Endocrinology & Metabolism found that metabolic rates actually decrease during treatment, meaning patients often gain more weight after stopping than they originally lost.

Novo Nordisk’s investor presentations explicitly describe this as “chronic weight management,” terminology that transforms a drug treatment into a lifetime subscription. Their market cap doubled to $460 billion (per Bloomberg Terminal data, September 2024) based on this premise. Goldman Sachs projects the GLP-1 market will reach $130 billion by 2030, noting that “persistent treatment is required for sustained benefit” – Wall Street’s euphemism for permanent customers.

The technology sector has discovered the same principle. Microsoft’s annual report shows Azure AI revenue growing 150% year-over-year, driven by what they call “consumption-based pricing” – you pay for every query, every computation, forever. Unlike traditional software you could buy once, AI requires constant feeding. Gartner’s “Market Guide for AI Services” (2024) found that enterprise AI costs increase an average of 78% annually after initial implementation, not from expanded use but from what they term “capability maintenance” – the system needs more to do the same job.

The Escalation Architecture

Both industries demonstrate identical escalation patterns documented in their own clinical and technical literature. The FDA’s Adverse Event Reporting System shows 43% of GLP-1 patients require dose increases within six months to maintain efficacy. Similarly, OpenAI’s pricing tiers show enterprise customers consuming 3.4x more tokens per user after 12 months than at implementation, according to data from Andreessen Horowitz’s “State of AI 2024” report.

This isn’t coincidence – it’s designed tolerance. A study in Nature Medicine found that GLP-1 receptors downregulate with continued exposure, requiring higher doses for equivalent effect. Microsoft Research published findings showing that organizations using AI assistants develop “prompt inflation” – needing increasingly complex instructions to achieve previously simple tasks. The human side of the system degrades to match the artificial enhancement.

The numbers reveal the trap’s elegance. Lilly prices Mounjaro at $1,200 monthly (per GoodRx data), knowing that insurance companies surveyed by the Kaiser Family Foundation report 64% coverage rates. For AI, Bain & Company’s technology spending survey found 89% of costs are absorbed by IT budgets rather than end users. The person who needs the product never sees the price, breaking the fundamental feedback loop of market discipline.

When Withdrawal Becomes Catastrophic

Medical literature documents what happens when patients stop GLP-1 drugs. Beyond weight regain, a Cleveland Clinic study published in Obesity Reviews found 31% of patients experience what researchers call “hyperphagia rebound” – hunger more intense than before treatment. The hypothalamic systems suppressed by the drug overcompensate when it’s withdrawn.

The parallel in technology is striking. When the City of Dallas attempted to reduce AI tool usage in 2024 as a cost-cutting measure (reported in Government Technology magazine), help desk tickets increased 400% within two weeks. Employees had forgotten how to perform tasks they’d done for years. The institutional knowledge of pre-AI processes had evaporated in just 18 months of AI dependence.

This creates what economists at the National Bureau of Economic Research call “capability atrophy through substitution.” Their 2024 working paper found that organizations using AI for more than 18 months lose approximately 23% of their internal problem-solving capacity. Like muscles that weaken without use, organizational capabilities deteriorate when continuously substituted by external services.

The Insurance Layer That Enables Everything

Both industries discovered their real customer isn’t the user but the payer. UnitedHealth Group’s 2024 earnings call revealed GLP-1 drugs now represent their fastest-growing pharmaceutical expense at $4.8 billion annually, increasing 40% year-over-year. They can’t stop covering them – employer clients demand it. Similarly, Flexera’s “State of the Cloud Report 2024” shows AI-related cloud costs growing 67% annually, faster than any other enterprise technology category.

The payer-user disconnect enables what behavioral economists call “moral hazard multiplication.” Research from the Wharton School found that when costs are indirect, consumption increases 4.2x compared to direct payment. Neither patients nor employees moderate usage when someone else pays, and those paying feel trapped by user expectations.

The Exit That Isn’t

Traditional products competed on being better than alternatives. These new dependencies compete on making alternatives impossible.

A Harvard Medical School study in the Annals of Internal Medicine found intensive lifestyle interventions achieve similar weight loss to GLP-1 drugs with 5-year maintenance rates of 60% versus 10% for drugs. But lifestyle change requires effort and time. Similarly, MIT Sloan research shows companies investing in employee training and process improvement achieve 2.3x better productivity gains than AI implementation, but over 24 months rather than 24 hours.

The market has chosen dependence over development, subscription over solution. Morgan Stanley’s pharma analysts project 15% of American adults will be on GLP-1 drugs by 2035. McKinsey predicts 80% of large enterprises will have “mission-critical” AI dependencies by 2030. These aren’t adoption curves – they’re addiction trajectories.

The most valuable companies of the next decade won’t be those that solve problems but those that make themselves unsolvable problems. They’ll create dependencies so deep that the cost of leaving exceeds any price they might charge. The customer who can never leave is worth infinitely more than one who chooses to stay.