Visa and MasterCard's durability comes from their five-party model connecting issuers, acquirers, merchants, and consumers. This structure creates a powerful, decentralized network effect that is extremely difficult to dislodge, making them a true "network of networks" that operates like a pre-crypto decentralized system.
Counterintuitively, Visa grows by introducing premium cards with higher merchant fees. These higher fees fund larger rewards, making the cards more attractive for issuing banks to promote and for affluent consumers to use. This strategy allows Visa to effectively capture market share through higher prices.
High interchange fees on premium credit cards fund rewards for affluent users. Merchants bake these fees into universal pricing, meaning customers paying with cash or debit cards—who receive no rewards—effectively subsidize the perks of wealthier individuals. This creates a wealth transfer from the poor to the rich.
When governments, like Australia's, cap interchange fees, merchants rarely pass the savings to consumers by lowering prices. Instead, they pocket the difference while the funding mechanism for consumer rewards disappears. This results in a direct wealth transfer from consumers to large retailers.
As AI software automates tasks once done by employees (e.g., insurance claims processing), a payroll expense becomes a software subscription. Companies pay for this AI with credit cards, turning what was an ACH payroll transaction into a net-new, high-margin payment volume for networks like Visa and Stripe.
The most durable enterprise companies create "hostages" through high switching costs and top-down contracts, not just beloved products. A simple heuristic for identifying a hostage company is poor UX that ignores the user, like a country dropdown list sorted alphabetically instead of by the user's location.
AI threatens companies with per-seat pricing (like Zendesk), as it reduces the need for human seats. However, incumbents like Workday, with per-employee pricing and captive customers, are protected. They can become the main distribution channel for new AI features, creating powerful new revenue streams without cannibalizing their core business.
Employees adopt unauthorized AI tools to make their own work easier—a classic principal-agent problem. An analyst using AI to finish a presentation in one minute instead of all night benefits personally, while the firm (the principal) sees no economic gain and instead incurs new data security risks.
The most compelling AI companies address "market failures" where demand exists but is unmet because the supply is too expensive. AI radically drops production costs (e.g., from a $35/hr graphic designer to a penny per Midjourney image), unlocking markets that are orders of magnitude larger than the original.
Instead of starting with a customer problem, start by reverse-engineering the ideal sales motion. Design a product with such a predictable and lucrative sales cycle that a top-tier enterprise salesperson would eagerly leave their high-paying job to sell it. This focus on sales predictability is the key to scalable success.