Cross-industry
The AI Value Index was established to explore, report on and judge claims about AI trends, use cases, case studies and signals. Today we find IBM sending a signal about a core factor impacting how AI is used in business. The article recognizes business "operating models that were not designed for how AI now produces work." This is a structural problem, according to the article, and the result is a tax on outcomes.
The tax, which IBM calls the "AI dividend tax" is an interesting way of characterizing "a structural condition in which AI-driven productivity gains are generated but they systematically fail to convert into enterprise value." Value creation is taxed by organizational structure, resulting in leakage that prevents productivity from turning into cost savings, capacity growth, efficiency and other positive business outcomes, at the organization level.
IBM argues that capturing this value requires changes to how work moves through the business. Requirements, development, testing and deployment may each become faster without the handoffs between them changing. Governance applied after work is completed can become another obstacle; IBM proposes embedding it in delivery instead. The article also questions commercial arrangements built around hours and staffing, where doing the same work with less effort may not reward the provider. Its proposed alternative combines tools, controls, skills and teams organized around business outcomes. The opportunity is to make productivity gains useful across the organization, rather than leave them isolated within individual tasks.
For the AI Value Index, this signal opens a useful line of inquiry: what changes when a business reorganizes work around the capabilities AI provides? IBM offers a framework for pursuing that opportunity, not a measured demonstration of its return. Evidence would come from businesses showing how specific changes to workflows and accountability translate into shorter delivery times, greater capacity or lower costs, while maintaining quality. The important comparison would connect those changes to results across the business. That would help distinguish faster activity from a larger dividend: more useful work completed and more value delivered at the organization level.
The AI Dividend Tax
IBM argues that enterprise operating models can prevent AI productivity gains from becoming business value.
