AI companies are rethinking how they charge to win a bigger slice of business spending

In a significant transformation, software companies are rethinking their pricing strategies for AI tools, a move that could substantially alter how businesses allocate their budgets. Rather than relying on a per-employee fee structure, many AI firms are now exploring pricing based on the actual work their products accomplish. This insight comes from a recent Goldman Sachs report, which highlights discussions with approximately 40 firms in the software and internet domains. According to Goldman Sachs analysts, businesses are increasingly marketing their AI capabilities as units of productivity or labor. This strategic shift enables these companies to secure larger contracts and access new budget resources. By aligning pricing with the value delivered, firms can decouple profits from the operational costs associated with AI, thereby preserving healthy margins. This new pricing paradigm is already manifesting in product offerings. For instance, Salesforce has launched “agentic work units,” while Workday offers credits linked to specific “units of work.” This evolution in pricing models is a response to the rising expenses tied to developing and operating AI technologies, prompting a reevaluation of software spending predictability for customers. In a related context, OpenAI CEO Sam Altman suggested that the future of AI could resemble utility services. He proposed that AI might eventually be sold in a manner similar to electricity or water, with clients purchasing 'tokens' for usage. Altman emphasized a vision where intelligence becomes a utility, accessible and metered for diverse applications. The broader software industry is also pivoting from traditional monthly 'per seat' licenses to more flexible, usage-based pricing structures. This trend, highlighted by Business Insider's Alistair Barr last year, signifies a growing shift towards pay-as-you-go models, aligning costs more closely with actual usage and value derived by businesses.