AI is breaking one of software’s most attractive assumptions: Once you build a product, serving one more customer costs almost nothing.
Traditional software may have required heavy upfront investment, but once the product was created, developers could distribute the result at very low marginal cost. Generative AI is different. Every prompt, every generated answer, every agentic task must be produced and paid for afresh.
Earlier this year, Stripe co-founder Patrick Collison argued that, in the AI era, “software should be like pizza,” made to order at the moment of use. But made-to-order software comes with made-to-order costs.
The internet economy was designed largely around human intent. People searched, clicked, subscribed, and checked out. Businesses optimized around that behavior: acquire a customer, convert them into a subscriber, and serve them at low marginal cost.
But now, AI agents can research, write code, call APIs, execute tasks and even pay for things on behalf of people or businesses. That promises speed and scale, but automated activity can very quickly increase consumption—and costs—far beyond what’s expected and what businesses are used to managing.
Everyone is watching AI inflate compute bills; very few are paying attention to who’s freeloading. Free-trial and multi-account abuse are the quiet killers of AI unit economics, and businesses often discover the damage only after their margins are gone.
The challenge is no longer about making compute cheaper—but ensuring that usage translates into revenue, rather than allowing unexpected or abusive consumption to erode businesses’ margins.
In Asia, AI adoption is shifting from experimentation into action. In 2025, a Stripe survey found that 82% of business leaders were already implementing, or planning to implement, agentic AI. Half of those surveyed expect a larger share of sales to come through AI-driven channels by 2030. According to Bain, 85% of Southeast Asian shoppers are already using or considering AI tools to guide their shopping decisions.
Metering the new unit of commerce
In the AI era, two different users can create vastly different workloads. Every interaction with an AI carries a real cost; as such, tokens are now the key unit for measuring and metering consumption.
Therefore, that pushes AI companies toward usage-based or hybrid pricing, which better reflects the cost of consumption, allowing companies to track usage and collect payments in real-time.
Take Lovable, the AI software creation platform that uses Stripe. Lovable started by offering subscriptions, but has since shifted to usage-based billing. Once customers exceed the free allowance in their plan, they are charged based on AI token consumption.
This new model introduces a new vulnerability: token theft. The actors behind this abuse rely on gaming the system. Attackers create multiple accounts, abusing free trials and consuming AI tokens without ever intending to pay for them.
Every token an abuser consumes creates an immediate operating cost the company will never recoup.
Stripe research found that more than one in six sign-ups at AI companies are linked to multi-account abuse.
AI companies can’t wait to act when a payment request fails; by then, the user has already moved on. Instead, they need to detect abuse at sign up, before the first token is ever used. Stripe, through solutions like Stripe Radar, a fraud prevention AI tool trained on data from millions of businesses worldwide, prevented $1.3 billion in fraud in Singapore last year.
Protecting margins
The challenge isn’t simply detecting token theft or fraud, but rather deciding how to respond.
Businesses need greater visibility into where their margins are exposed, along with the flexibility to set their own risk tolerance and tailor their response to different behaviors. For example, someone repeatedly creating new accounts to steal free AI tokens should be treated differently from a paying casual user whose usage suddenly spikes.
As AI changes the cost of serving customers, protecting margins will depend not only on cheaper compute, but also on making sure every token consumed counts.
The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.
