McKinsey & Company CFO Yuval Atsmon said artificial intelligence offers significant productivity gains for firms but introduces new operational costs.
This balance is critical as companies move from experimenting with AI to integrating it into core business models. The shift requires chief financial officers to manage not only the potential for job displacement but also the tangible costs of scaling technology.
Atsmon said the emergence of "token-bill management" is a new financial challenge. As businesses increase their reliance on large language models, the cumulative cost of processing data—measured in tokens—becomes a significant line item in corporate budgets.
Beyond direct costs, the integration of AI is altering the demand for specific human skill sets. Atsmon said that while some technical roles may shift, there is a renewed value in generalist skills. These workers are better equipped to oversee AI outputs, and integrate them across different business functions.
To capture the full benefit of the technology, firms must redesign their workflows. Atsmon said that simply layering AI over existing processes is insufficient. Companies must fundamentally change how work is structured to avoid inefficiency.
These efficiency gains are expected to be substantial on a national scale. For example, McKinsey research indicates that AI could unlock €15 billion [1] in productivity gains in Hungary by 2030.
Atsmon said that McKinsey is applying these same principles internally. The firm uses AI to streamline its own operations while simultaneously advising clients on how to navigate the rollout of these tools without compromising financial stability.
“AI offers significant upside for firms but also brings costs such as higher token-bill expenses.”
The transition from AI experimentation to implementation is shifting the conversation from theoretical capability to fiscal management. The mention of token-bill management suggests that the 'hidden costs' of AI—specifically API and compute expenses—are becoming a primary concern for corporate finance leaders, potentially offsetting some of the labor savings gained from automation.
