A guide for finance and business leaders to measure AI returns, realise business value and manage the full cost of AI.
Beyond the AI business case: Making the economics of AI work
Book a 30-minute working session
PwC can help clarify the economics and the measures that matter.
Where is the return on AI?
After several years of experimentation, rising technology investment and ambitious productivity claims, finance and business leaders face a critical challenge: demonstrating where AI is creating economic value, how much of that value their organisations are capturing and what it costs to sustain.
Traditional investment disciplines remain essential. But AI introduces new considerations: productivity gains do not automatically translate into financial savings, capabilities developed for one use case may create value across many others, and the cost of delivering AI-enabled outcomes continues to evolve as adoption scales.
AI economics is the discipline of measuring and managing the value and full cost of AI-enabled work throughout its lifecycle.
The challenge of realising AI value
PwC’s 2026 AI Performance study reveals a striking divide in the value organisations are generating from AI. Just 20% of organisations account for 74% of the financial value created. These top-performing organisations, defined by PwC as AI leaders based on their AI-driven revenue and efficiency gains relative to industry peers, demonstrate what is possible when AI investment translates into measurable business outcomes.
PwC's September 2026 pre-event survey of 55 Australian finance leaders provides a further perspective.
Source: PwC pre-event survey of 55 Australian finance leaders, September 2026.
Four lenses for understanding AI economics
AI's economic impact begins at the level of the task. Beyond measuring time saved, organisations need to understand the full cost, quality and speed of producing a successful outcome.
This means accounting for human effort, AI consumption, tools, retries, controls and assurance. Tracking these measures over time provides a clearer picture of whether AI is genuinely improving operating economics as adoption increases and workflows evolve.
Measures:
- Cost per successful outcome
- Quality and accuracy
- Speed
- Human intervention and exception rates
Productivity gains don't automatically translate into financial returns. What matters is how organisations use the capacity released.
Some of that capacity may translate into direct cost savings, while some may enable additional activity without equivalent workforce growth. Other benefits may come through improved performance, stronger customer outcomes or activities that were previously uneconomic.
Finance needs to distinguish between productivity created and value realised, with clear attribution across cost savings, additional activity and performance improvements to avoid counting the same benefit more than once.
Measures:
- Cash savings realised
- Additional activity
- Improved performance
- New activity that was previously uneconomic
One of AI's more distinctive economic characteristics is its potential to create value beyond the use case it was originally designed to support.
Reusable components, including data, integrations, enterprise knowledge and controls, can reduce the time and investment required for subsequent applications. Productive AI capabilities, such as agents or skills, may also be adapted and deployed across different tasks or business contexts.
For business leaders, the opportunity is to look beyond individual use cases and consider how each investment can create capabilities that deliver value across the enterprise, while accounting for the costs of adaptation, assurance and ongoing operation.
Measures:
- Avoided development cost
- Incremental cost to adapt and deploy
- Time to deployment
- Applications supported
Unlike many traditional technology investments, AI introduces a variable cost of production that continues to evolve after deployment.
Each additional task may consume inference, retrieval, tools and human oversight. Total AI expenditure may increase as demand grows, even when the cost of individual tasks is falling.
As models improve, prices shift and workflows evolve, organisations need to continually assess whether they're delivering successful outcomes in the most economically effective way.
Measures:
- Inference and model costs
- Tools, retrieval and storage
- Routing and assurance
- Optimisation as usage grows
- Technology improvment over time
When productivity gains tell only part of the story
Consider a bank handling 100,000 customer interactions annually. AI reduces the cost per interaction from $20 to $7, creating $1.3 million in potential productivity.
But does that mean the bank has generated $1.3 million in financial value? Not necessarily. The answer depends on how that capacity is used, whether the capability can generate value elsewhere and what it costs to sustain.
The fuller economic picture
In this illustrative example, $1.6 million in realised business value, plus $0.5 million in avoided development costs and less $0.2 million in additional lifecycle costs, produces $1.9 million in annualised economic value.
Three priorities for realising AI value
AI does not replace established investment and benefits realisation disciplines. It extends them. Three priorities can help Finance bring the full economics of AI into investment and performance management.
1. Establish the economics of the work
Define a meaningful unit of work and establish its baseline. Measure the cost, quality, speed and volume of successful AI-enabled outcomes, including inference, tools, retries, human input and assurance.
2. Build the full economic case
Distinguish productivity created from value realised. Account for credible reuse benefits, incremental adaptation costs and the ongoing expenditure required to operate and improve AI.
3. Manage AI value across the portfolio
Assign accountability, track realised business outcomes and review where investments should be scaled, redesigned, paused or stopped as technologies and economics change.
FAQs
Read the full report
The full guide explores how AI changes the cost of work, how productivity becomes realised value, how reusable capabilities extend returns and how operating economics evolve as AI scales.
The economics of AI: What every CFO needs to know
Virtual Event | Thursday 19 November 2026 | 12-1pm AEDT
Get in touch
Matthew Tutty
Partner, Strategy&, PwC Australia
Ben Meacock
Partner, CFO Advisory National Leader, PwC Australia
Mathea Beck
Partner, Productivity Leader, PwC Australia
Matthew Hunt
Assurance Alliances Market Lead, PwC Australia
Nicola Costello
Partner, Digital and AI Trust Leader, PwC Australia
Nikhil de Silva
Partner, Digital Leader, PwC Australia
