About the paper
PwC’s AI performance study examines why some large companies generate measurable ROI from AI while others remain stuck in pilot activity.
It is original proprietary survey research based on 1,217 senior executives, all director-level or above, across 25 sectors and across Africa, Asia, Europe, the Middle East, North America and South America; fieldwork took place in October and November 2025.
The sample is heavily weighted towards large publicly listed companies: 91% were publicly listed and 76% had revenue of US$1 billion or more.
Length: 30 pages
More information / download:
https://www.pwc.com/gx/en/issues/technology/ai-performance/want-ai-roi-go-for-growth.html#the-takeaways
Core Insights
1. What is the central argument of the report?
The report argues that AI ROI is not mainly a function of how many AI pilots a company launches. It comes from what PwC calls “AI fitness”: the ability to aim AI at important business outcomes, build the foundations that make AI scalable and reliable, and embed AI deeply across the enterprise.
PwC’s headline finding is that the most AI-fit companies achieve AI-driven financial performance that is 7.2 times higher than other companies. The study defines this performance as revenue and efficiency gains attributable to AI, adjusted against sector medians. According to the report, 20% of the surveyed companies capture 74% of AI-driven returns, showing that value is highly concentrated among a relatively small group of leaders.
The core message is therefore: companies should stop counting pilots and start building the organisational machinery that turns AI activity into measurable business impact. The strongest performers treat AI not as a tool for scattered productivity gains, but as a reinvention engine that can reshape products, business models, operating models and competitive positioning.
2. What does PwC mean by “AI fitness”?
PwC defines AI fitness as a combination of six foundational capabilities and three measures of AI use.
The six foundations are:
- Innovation — dedicated infrastructure, business-unit ownership and disciplined portfolio reviews.
- Governance and risk — Responsible AI frameworks, access controls, compliance processes and oversight bodies.
- Data and technology — modern platforms, trusted data, reusable components and redesigned workflows.
- Strategy — a clear connection between AI deployment and corporate strategy.
- Investment — sufficient funding and the ability to reallocate resources as priorities shift.
- Workforce — skills, incentives, collaboration models and trust in AI-generated insights.
The three measures of AI use are:
- Breadth and depth — how widely AI is used across the value chain and how deeply it is embedded into workflows.
- Sophistication — whether AI is used merely to assist and summarise, or whether it executes tasks, operates within guard rails or becomes autonomous.
- Capturing value from industry convergence — whether AI helps the company identify and exploit new value pools across traditional sector boundaries.
The important point is that PwC does not define AI leadership as “using more AI” in a generic sense. Leadership comes from combining AI use with the organisational conditions that make AI repeatable, trusted and commercially meaningful.
3. Why does the report say companies should aim AI at growth and reinvention, not just efficiency?
The report acknowledges that many companies use AI to improve efficiency: faster claims processing, code generation, customer support automation, productivity improvements and cost reduction. But PwC argues that the biggest returns come when AI changes what the company sells, how it creates value and where it competes.
AI leaders are 2.6 times as likely as other companies to say AI has improved their ability to reinvent their business model. They are also more likely to use AI to identify emerging value pools, respond to changing customer needs, collaborate across sectors and compete outside their traditional industry boundaries.
The strongest individual AI fitness factor in the study is the ability to capture growth from industry convergence. PwC’s logic is that AI allows companies to detect and act on new combinations of customer needs, data, services and partnerships that cut across old sector lines.
The John Deere example illustrates this argument. The company used AI-powered precision spraying not simply to make an existing machine smarter, but to support a more service-like model linked to verified outcomes. According to the report, See & Spray was used on more than 1 million acres in the 2024 growing season and saved an estimated 8 million gallons of herbicide mix, while also positioning John Deere for recurring services revenue rather than one-off hardware differentiation.
4. What foundations do AI leaders build differently from other companies?
PwC’s argument is that foundations matter because they improve the “conversion rate” from AI use to measurable results. Companies with stronger foundations reportedly see nearly double the improvement in AI-driven performance when they increase AI use, compared with companies with weaker foundations.
The report highlights five especially important foundation-building practices.
First, AI leaders fund and flex the portfolio like investors. They invest materially more in AI — PwC says leaders invest 2.5 times as much of revenue as other companies — and they are better at reallocating financial and human resources towards higher-value AI opportunities.
Second, they create infrastructure for innovation. They are more likely to provide sandbox environments and dedicated technical infrastructure for experimentation, and more likely to place accountable innovation owners inside business units.
Third, they build workforce trust. Employees in AI-leading organisations are 2.1 times as likely to trust AI-generated insights and act on them in daily decisions. The report links this trust to cross-functional co-creation, role-based learning, incentives to experiment and clear guard rails.
Fourth, they use governance to accelerate rather than obstruct. Leaders are more likely to have a documented Responsible AI framework and a cross-functional governance board, but the report frames governance as a way to move routine use cases faster while escalating only higher-risk cases.
Fifth, they remove data and technology friction. AI leaders are much more likely to have reusable centrally catalogued AI components, high-quality accessible data, modern cloud-based platforms, and redesigned workflows that incorporate AI rather than merely adding AI tools on top of existing processes.
The overall message is pragmatic: companies should not try to modernise everything at once. They should build only the foundations needed to scale the AI initiatives most closely tied to strategic outcomes.
5. What are the report’s main implications for executives?
The main implication is that executives need to move from an “AI activity” mindset to an “AI performance” mindset.
PwC repeatedly warns against treating AI pilots as proof of progress. The report suggests that many companies have visible AI activity but little evidence of revenue growth, cost reduction, better decisions or improved customer outcomes. The practical challenge is therefore to connect AI to specific business objectives, measure impact, and stop or scale initiatives based on evidence.
The report’s recommended executive agenda has four parts.
First, aim AI at strategic value, especially growth, reinvention and industry convergence, rather than only internal productivity.
Second, assign ownership and metrics. Every important AI initiative should have a business owner, success measures and accountability for outcomes.
Third, industrialise what works. Companies should take proven use cases and replicate them across functions, regions, workflows and decision points, rather than leaving value trapped in isolated pockets.
Fourth, move carefully towards greater autonomy. PwC argues that automation of routine, high-frequency decisions is strongly linked to AI-driven performance, but it should happen within explicit guard rails, with decision quality measured and trust thresholds met before expansion.
The conclusion is that AI advantage may compound. The leading companies are learning faster, redeploying solutions faster and safely automating more decisions. For laggards, the risk is not just missing short-term efficiency gains, but falling behind as AI-fit competitors turn their foundations into a widening performance premium.