New technologies commoditize old sources of advantage. Consider the invention of the camera: Portrait painters could not adapt by simply deploying the new technology–using the camera to make their workflows faster and more efficient–because photography had commoditized their offering wholesale, reducing customers’ willingness to pay and drying up their value pool.

Our research shows a similar pattern emerging with generative AI. We mapped job roles to work activities for 800 U.S. public firms and then scored each activity’s potential for augmentation or automation by gen AI, based on factors such as the need for physical presence, digital availability of inputs, and reliance on task-based logic. We find no correlation between the resulting measure of gen AI automation potential per sector and the development of sector-level profit margins since the launch of ChatGPT. In fact, sectors with the highest potential for automation—finance, tech, and media—saw margins stagnate or fall.

We also found that this relationship is not mediated by differences in gen AI adoption levels across sectors. To approximate this, we used publicly available data indicating the extent to which firms are actively building or integrating gen AI capabilities–such as the share of employees with relevant skills and the number of third-party AI technology vendors firms used.

In sum, even as companies pour resources into gen AI, the resulting productivity gains appear to be competed away, passed on to customers or suppliers rather than retained by firms–an early sign of commoditization. As such, the data suggests that the benefits of gen AI will not lie in using the technology simply to perform existing activities faster or more efficiently.

Rather, business leaders seeking to deploy gen AI for advantage should look toward another impact of new technologies: the unlocking of new value pools.