Skip to main content

      Companies are stepping up their use of artificial intelligence. However, there is often a lack of clarity regarding the exact extent of usage, the resulting costs and the measurable added value.

      The solution lies in a simple unit: the token. It is the smallest processing unit of a language model and makes AI usage measurable, just as a kilowatt-hour measures electricity consumption. The token makes it possible to view the scope of use, costs and added value of AI in context. 

      Download whitepaper now (in Germany only)

      picture_as_pdf

      Whitepaper

      KI-Kosten im Blick: Wie Token Verbrauch, Kosten und Nutzen messbar machen

      Wie Unternehmen mit Token KI-Verbrauch messen, Kosten bis 2031 abschätzen und den Einsatz von KI-Agenten nach ihrem Geschäftsnutzen priorisieren.



      Falling prices, higher usage

      In our white paper ‘Managing AI Value in Organisations: The Token as the Pacing Mechanism”, we use the example of a manufacturing company to demonstrate the financial impact of the increasing use of AI. Although token prices are set to fall by a factor of around 1.9 by 2031, the volume will grow by a factor of between 5.7 and 8.5 over the same period. As a result, annual AI operating costs will rise from around six million euros to between 19 and 28 million euros. 

      To provide a robust overview of the consumption, costs and benefits of AI deployment, we have developed an integrated AI management model, which we present in the white paper. It combines three perspectives: 

      • Token model (usage):

        Shows how much AI is used within the organisation, in which areas and by whom: employees, agents or machines. A comparison with best-practice figures highlights over-consumption and untapped potential.

      • Pricing model (costs):

        Forecasts token price trends over five years and across different providers. This yields reliable cost scenarios.

      • Agent model (benefits):

        Identifies the AI agents with the greatest value contribution and derives a prioritised implementation roadmap within the available budget.

      The control model links AI consumption and AI costs to the achievable added value. This makes it possible to identify where excessive consumption can be reduced and in which areas AI usage should be expanded. It also highlights which AI agents promise the highest value contribution and how future cost trends can be factored into budget planning at an early stage. 


      Ready to talk?

      More KPMG Insights for you

      Technology, Media & Telecommunications (TMT)

      Hardly any other industry is at the centre of the digital transformation like the TMT sectors.
      Man with laptop in the data center

      Your contacts