error
Subscriptions are not available for this site while you are logged into your current account.
close
Skip to main content
Submit RFP

      Artificial intelligence is transforming how organisations operate, with investment accelerating as businesses embed AI into products, services and everyday operations. Unlike traditional software, many AI services are consumption-based, meaning costs are driven by usage rather than fixed licence fees.

      As AI adoption scales, organisations must look beyond implementation and focus on value. Success is no longer measured by how many AI solutions have been deployed, but by whether they deliver measurable business outcomes and a return on investment.

      KPMG's AI Value Management framework helps organisations maximise value per token by optimising AI spend, strengthening governance and connecting AI investment to measurable business value. The result is a structured approach that enables organisations to scale AI with confidence, control and accountability.

      su-crighton
      Su Crighton

      Partner, Technology and Data

      KPMG in the UK

      Managing AI costs and business value

      As organisations move from AI experimentation to enterprise-scale adoption, controlling AI costs becomes significantly more complex. AI spend is influenced by factors including token consumption, model selection, infrastructure choices and application design, making it significantly less predictable than traditional software licensing models.

      The challenge is intensified by AI agents, which can make multiple model calls autonomously, and by adoption across business units outside the technology function. This can create fragmented ownership, limited visibility and unpredictable cost increases.

      Managing spend alone is not enough. Organisations also need to understand whether AI investment is improving productivity, revenue, customer experience, risk management or other strategic outcomes. Traditional IT budgeting was not designed for this combination of variable consumption and distributed usage. Effective AI Value Management therefore requires organisations to connect cost, usage and performance data, so that investment decisions are based on measurable value rather than adoption alone.

      Understanding AI tokenomics

      AI tokenomics describes how AI consumption translates into cost and business value. A token is a unit of text processed or generated by a large language model, and many AI services charge according to the number of tokens used.

      However, total cost is shaped by more than volume alone. Different models carry different price and performance profiles, while long or inefficient prompts can increase consumption without improving outcomes. Intelligent model routing helps direct each task to the most appropriate model, avoiding the unnecessary use of higher-cost options. Caching can also reduce spend by reusing previously generated information rather than processing the same request repeatedly.

      Together, these techniques help organisations improve efficiency, control inference costs and maximise the value delivered from every token consumed.

      The KPMG AI Value Management Framework

      KPMG's AI Value Management framework provides a structured approach that helps organisations optimise AI consumption, strengthen governance and maximise business value. Rather than focusing solely on reducing costs, the framework enables organisations to make informed decisions about where AI investment delivers the greatest return while maintaining visibility and control as adoption scales.

      Built around five complementary capabilities, the framework helps organisations understand how AI is used, optimise consumption, improve operational efficiency and establish the governance needed to scale AI confidently and responsibly.

      • Observe and Define

        Establish end-to-end telemetry across AI services to understand how AI is being used. Define meaningful KPIs that measure token consumption, cost, usage and business value, creating a reliable baseline for decision-making.

      • Classify and Route

        Implement intelligent model routing so that requests are directed to the most appropriate AI model based on their complexity. This helps balance performance, cost and response quality while avoiding unnecessary use of premium models.

      • Cache and Retrieve

        Reduce duplicate processing through multi-level caching, enabling frequently used information and responses to be reused where appropriate. This lowers token consumption, improves efficiency and helps control AI costs at scale.

      • Engineer and Optimise

        Apply structured prompt engineering techniques to improve the quality and consistency of AI outputs. Developing reusable prompt libraries helps reduce unnecessary token usage while delivering more reliable results across users and use cases.

      • Govern and Control

        Establish clear organisational governance for AI consumption through policies, guardrails and automated monitoring. By improving accountability and providing greater oversight of AI usage, organisations can better manage risk, optimise spend and support sustainable AI adoption.

      Together, these five capabilities form a practical framework for connecting AI consumption with measurable business outcomes. By combining operational visibility, technical optimisation and effective governance, organisations can maximise value per token and scale AI with confidence.

      The Benefits of AI Value Management

      AI Value Management helps organisations make more informed decisions about where and how they invest in AI. By connecting AI consumption with measurable business outcomes, organisations can prioritise high-value use cases, optimise spending and demonstrate a clearer return on investment.

      The framework helps create more predictable AI budgets, reducing the risk of unexpected cost increases as adoption scales. It also provides leaders with greater confidence that AI investment is aligned to business priorities, supported by consistent measurement and evidence-based decision-making.

      The result is a more sustainable approach to AI adoption—one that enables organisations to scale AI responsibly, maximise value from every token consumed and balance innovation with long-term commercial value.

      AI Value Optimisation FAQs

      AI Value Management is the process of tracking, governing and optimising AI investments to ensure they deliver measurable business value, not just technology adoption. Organisations are increasingly focusing on AI economics, accountability and value realisation.

      A token is the unit of text an AI model processes (words, parts of words or characters). AI providers typically charge based on the number of input and output tokens used, making tokens a key driver of AI costs.

      AI costs depend on user adoption, token consumption, model choice and evolving use cases. KPMG found organisations with full visibility of AI operating costs are 5x more likely to achieve established ROI (15% vs 3%).

      Traditional IT cost management focuses on controlling technology spend. AI Value Management focuses on both cost and business outcomes, such as productivity gains, revenue growth and ROI.

      By comparing AI costs against measurable benefits such as cost savings, productivity improvements, revenue growth, risk reduction and customer outcomes. Cost visibility and clear accountability are critical to demonstrating ROI.

      It should be a shared responsibility across business, technology and finance teams, with executive accountability. KPMG found 75% of CEOs own AI as a strategic priority, highlighting the need for leadership oversight.

      Our AI insights

      Thought-leadership, expert analysis, and guidance on all things AI.

      KPMG helps you design, build and scale AI solutions that accelerate performance, unlock value and manage risk and complexity with confidence.
      Read more

      Accelerating the Power of AI with confidence.

      Read more

      Insights from a panel of over 2,000 global business leaders, including more than 100 in the UK, for their views on AI topics including ROI, productivity and the shifting workforce and talent pipeline.
      Read more

      Something went wrong

      Oops!! Something went wrong, please try again
      MTD

      Get in touch

      Discover why organisations across the UK trust KPMG to make the difference and how we can help you to do the same.

      Contact us