Capacity-based pricing is emerging as the preferred model for software companies introducing artificial intelligence (AI), with around four in five vendors opting for fixed commitments rather than charging customers purely on actual consumption, according to a Bain & Company report.
The report, based on an analysis of publicly available pricing across around 200 B2B software-as-a-service (SaaS) companies, said capacity models offer customers more predictable budgets while allowing vendors to maintain more stable revenue.
"Capacity, rather than consumption, is emerging as the preferred model, as it offers predictable budgets for customers and stable revenue for vendors," Bain said.
Under capacity-based pricing, customers commit to a fixed amount of AI usage for a set period and generally do not receive a refund or carry forward unused capacity. This gives customers more predictable budgets while allowing vendors to secure revenue from committed capacity.
Bain said about 80 per cent of vendors introducing AI pricing are choosing capacity models. The report noted that such models retain some of the economic advantages of traditional seat-based software pricing, as vendors receive revenue from committed entitlements even when customers do not fully use them.
For customers, the model can also make spending easier to plan, as procurement teams and chief financial officers can approve a defined commitment instead of facing potentially open-ended variable costs.
The report found that AI pricing is increasingly moving beyond traditional per-seat licensing, but this does not mean seats are disappearing. About one in five AI-native software companies still relies mainly on per-seat licensing, often supplemented with usage entitlements. Most companies expanding beyond seat-based pricing are adding new pricing meters rather than replacing seats altogether.
Among companies introducing hybrid AI pricing, output-based models account for about 55 per cent, effort-based models for 35 per cent and outcome-based models for only 10 per cent.
Bain said outcome-based pricing, where vendors are paid for a business result rather than the work produced, has a more limited role. It is most suitable when an outcome is observable, uniquely attributable to AI and can be clearly agreed upon between buyer and seller.
"Outcome-based pricing is likely to remain compelling but relatively narrow in scope over the next several years," the report said.
The report said direct usage pricing will continue to have a role, particularly in infrastructure software and products sold to technical buyers. However, capacity is likely to remain the favoured approach for most enterprise applications.
"In AI pricing, 'outcome-based' or 'usage-based' are the headlines. The specific meter is the strategy," Bain said.
