AI is breaking one of software’s most attractive assumptions: once a product is built, serving one more customer used to cost almost nothing. In generative AI, every prompt, every generated answer, and every agentic task must be produced and paid for anew.
That shift changes the business model. Traditional software was built around low marginal cost after the upfront investment. AI is different, because usage itself creates an operating cost. As Stripe regional head and MD of Southeast Asia, Greater China and South Korea, Sarita Singh argues, made-to-order software also means made-to-order costs.
The internet economy was designed around human intent. People searched, clicked, subscribed, and checked out. AI agents now 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 also increase consumption far beyond what companies expect.
The result is a new focus on margins. Singh says 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. Stripe research found that more than one in six sign-ups at AI companies are linked to multi-account abuse.
The numbers come first. In Asia, AI adoption is moving 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. Bain said 85% of Southeast Asian shoppers are already using or considering AI tools to guide their shopping decisions.
That is pushing AI companies toward usage-based or hybrid pricing, which better reflects the cost of consumption and allows companies to track usage and collect payments in real time. Lovable, an AI software creation platform that uses Stripe, started with subscriptions and later shifted to usage-based billing. Once customers exceed the free allowance in their plan, they are charged based on AI token consumption.
But the same model creates a new vulnerability: token theft. According to Singh, attackers create multiple accounts, abuse free trials, and consume AI tokens without ever intending to pay. Every token an abuser consumes creates an immediate operating cost the company will not recoup.
Stripe says AI companies cannot wait until a payment request fails, because by then the user has already moved on. The company says abuse needs to be detected at sign-up, before the first token is used. Stripe Radar, its fraud prevention AI tool trained on data from millions of businesses worldwide, prevented $1.3 billion in fraud in Singapore last year.
Capital rewards clear rules. The market has already voted on AI as a growth story, but the balance sheet is now asking a harder question: who pays for the usage? In the AI era, free enterprise works only if companies can meter demand, stop freeloaders, and protect margins before the cost shows up in the red ink.
