What your Gym Membership Can Teach Us About Modern Finance

Across esoteric credit markets, asset-backed securities (ABS) have a collateral that is increasingly becoming something you cannot repossess. It is no longer simply backed by hard assets, but by a patient’s promise to repay for dental work, or a gym member’s monthly subscription. Investors are moving from underwriting the asset to underwriting the behaviour that generates its cash flows.

One of the clearest illustrations of this sits in the consumer health and fitness industry, where two bond deals are making completely opposite bets. One deal depends on people showing up, and the other depends on them staying home.

Cherry Technologies Inc. is a fintech that provides buy now pay later (BNPL) for health, medical aesthetics and wellness providers. Cherry wants you to show up, because the more appointments you schedule and finance, the more receivables it originates. By contrast, Planet Fitness is America’s leading gym provider, and well, they don’t really want you to show up. Planet Fitness is a low-cost gym with over 2,700 locations and 20.8 million members across America – that means that there are approximately 7,700 members per gym. They sell more memberships than they can physically accommodate, and the model works because most people keep paying without actually attending.

Meet the Collateral

Both companies have entered credit structures that bet on your habits. In June this year, Cherry issued $350 million across four tranches, collateralised by receivables financing elective medical procedures. These were transferred into a special purpose vehicle (SPV), which then issued notes to insurance companies, pension funds and asset managers. The deal operates with a 24-month revolving period, allowing the continual origination of new loans to replenish the collateral pool, whilst providing Cherry with ongoing access to funding. Effectively, investors, rather than a bank, are funding elective healthcare borrowing. It forms part of the broader evolution in securitisation, where technology platforms can originate specialised assets and access private capital markets directly.

Planet Fitness doesn’t securitise individual consumer receivables, rather, it relies on a whole business securitisation (WBS) structure, where the collateral is future operating cash flow. The bulk of that cash flow is franchise royalties, which are themselves a slice of the monthly dues members pay. In December 2025, Planet Fitness completed a $750 million refinancing through its existing WBS structure. Investors are not lending against the treadmills or gym equipment, but against the expectation that millions of people will keep paying their monthly subscriptions.

What a gym and Botox appointment can teach us about AI

Across very different industries and structures, credit markets have become more comfortable lending against cash flows generated by predictable human behaviour, Cherry depends on consumers seeking treatment; Planet Fitness depends on consumers continuing to pay membership. In both cases, the structure differs but economically, both begin with repeated human behaviour.

Meanwhile, AI has become one of the defining themes in credit markets. Morgan Stanley expects AI-related debt issuance to surpass $570 billion in 2026, most of it underwriting expectations of sustained demand for compute. That raises the natural question of whether structured credit has become increasingly comfortable lending against recurring human behaviour, and if so, could AI subscription revenues eventually support similar structures?

To answer that, it is worth understanding why AI subscriptions are fundamentally different.

Many AI companies, such as Anthropic, OpenAI, and Google, rely on recurring subscription revenues. At first glance, it may seem like a gym membership where customers pay a fixed monthly fee in exchange for access to the service, but economically, it is the opposite. Suppose you and I both have Claude Pro subscriptions for $20 a month. I use it every day, generating code, building websites to publish fixed income articles, whereas you only log in once a week. We pay the same fee, but I am far costlier for Anthropic to serve as every additional prompt requires compute capacity, GPUs and electricity. In other words, although revenue is fixed, costs increase with usage.

This distinction matters for structured credit because investors are not simply underwriting recurring revenue but the cash flow available to service this debt. In AI, revenue may become highly predictable, but the cost of generating that revenue moves alongside demand. Therefore, the cash flow generated becomes far harder to size at underwriting.

Does that mean AI subscription securitisations will never exist? I do not think so. If pricing continues to evolve towards committed spending or tiered usage, and margins become sufficiently stable, there is no obvious reason why recurring AI subscription revenues could not plausibly become financeable in the same way gym memberships or healthcare receivables are today. Credit markets are already lending against the infrastructure powering AI, yet they seem less comfortable lending against the subscription revenues that pay for that infrastructure. The most valuable collateral is no longer necessarily the asset that can be repossessed after default, but the cash flows generated by behaviour that proves to be persistent.

Credit markets have always been in the business of predicting the future. Traditionally, that meant estimating the resale value of a house, a car, or a factory. Today, it is judging whether the behaviour underpinning those cash flows is predictable. Investors have become comfortable financing gym memberships, cosmetic dentistry, and other cash flows rooted in human habits rather than hard assets. If AI companies can one day demonstrate the same stability, not just in revenue but also in margins, they may find that the next asset to be securitised is not a server or a GPU, but the subscription habits of millions of users. The future of collateral may not be something you can repossess, but the behaviour people cannot stop repeating. In modern credit markets, the habit is becoming more valuable than the asset.