AI firms shift from seat pricing to usage based revenue models
Artificial intelligence companies are moving away from traditional per user pricing toward models that charge based on actual usage, according to a new analysis from Goldman Sachs. The shift reflects how AI is changing the nature of work, with software increasingly performing tasks instead of human users.
The report, based on discussions with dozens of software and internet companies, shows that firms are adopting pricing tied to productivity units, completed tasks, or compute usage. This approach allows vendors to expand deal sizes and tap into new corporate budgets linked to operations rather than headcount.
Several major software providers are already implementing this model. Salesforce introduced Agentic Work Units, which measure tasks completed by AI systems instead of the number of employees using the software. Workday launched Flex Credits, a system that tracks AI usage within subscriptions. SAP has also moved toward consumption based pricing, arguing that automation reduces the relevance of seat based billing.
The shift is driven by a structural change in how software delivers value. As AI agents automate workflows, fewer employees need direct access to applications. This undermines the seat licensing model that has defined the software as a service industry for more than a decade.
Executives across the industry are outlining a future where AI is treated as a metered utility. Sam Altman said intelligence could be sold like electricity, with customers paying for what they consume. OpenAI already uses token based pricing tied to data processing. Satya Nadella has indicated that Microsoft is shifting from per user pricing to per agent billing as AI becomes embedded in workflows.
The transition carries risks. Usage based revenue can fluctuate, and customers may reduce consumption when costs become transparent. Many companies are adopting hybrid models that combine subscriptions with consumption metrics to stabilize income.
Analysts expect AI driven pricing to dominate the sector over time. Goldman Sachs estimates that AI agents could represent more than 60 percent of software economics by 2030, signaling a long term decline of the per seat model and a fundamental reshaping of how the industry generates revenue.
-
17:00
-
16:45
-
16:40
-
16:30
-
16:20
-
16:15
-
16:01
-
16:00
-
15:45
-
15:40
-
15:30
-
15:20
-
15:15
-
15:00
-
15:00
-
14:45
-
14:40
-
14:30
-
14:20
-
14:15
-
14:00
-
13:50
-
13:45
-
13:30
-
13:15
-
13:05
-
12:30
-
12:20
-
12:15
-
12:00
-
12:00
-
11:45
-
11:40
-
11:30
-
11:20
-
11:15
-
11:00
-
11:00
-
10:45
-
10:40
-
10:30
-
10:20
-
10:15
-
10:00
-
10:00
-
09:45
-
09:40
-
09:30
-
09:20
-
09:15
-
09:00
-
09:00
-
08:45
-
08:40
-
08:30
-
08:20
-
08:15
-
08:00
-
07:50
-
07:45
-
07:30
-
07:15
-
07:00
-
06:20
-
17:30
-
17:20
