AI Spend Is Up. Returns Aren't-Yet
CEOs poured money into AI across 2024. Most are still waiting for it to show up on the P&L.
According to PwC's latest Global CEO Survey of 4,454 leaders across 95 markets, 56% say AI hasn't produced revenue or cost benefits so far. Some progress exists-around a third reported higher revenue and 26% saw lower costs-but only 12% achieved both in the last 12 months.
Confidence is sliding, too. Just 30% are very or extremely confident about revenue growth over the next year, down from 38% last year and 56% in 2022. You can read the source survey here: PwC Global CEO Survey.
Where returns are showing up
Recent reads on S&P 500 performance suggest returns are uneven by sector. Technology, communication services, and financials are leading on measurable AI lift, with energy moving up fast.
As one leader put it: "A small group of companies are already turning AI into measurable financial returns, while many others are still struggling to move beyond pilots. That gap is starting to show up in confidence and competitiveness-and it will widen quickly for those that don't act."
What the top 12% are doing differently
PwC found the small cohort achieving both cost and revenue gains were 2-3x more likely to have a strong AI foundation. That means AI is embedded across products and services, demand generation, and strategic decision-making-not just in pilot projects.
The pattern is clear: firms that treat AI as a business system (strategy, data, operating model, talent) see results. Those that treat it as isolated experiments don't.
The AI strategy that actually pays
- Start with a value thesis: Tie every use case to a clear revenue, cost, or risk metric. Prioritize 2-3 high-velocity use cases; stop the rest.
- Fix data before models: Standardize definitions, clean pipelines, and build shared features. Automate data quality checks and access controls.
- Ship with a product model: Cross-functional pods (product, data, engineering, legal, domain experts) with a budget, a deadline, and a KPI.
- Win adoption, not headlines: Train managers and ICs, build incentives into workflows, and support with co-pilots. Measure feature usage weekly.
- Instrument outcomes: Baseline KPIs and track monthly: revenue lift, conversion, cost per ticket, time-to-quote, claim cycle time, fraud hit rate.
- Balance the portfolio: Allocate 70/20/10 across run/grow/transform. Reallocate budget quickly to winners; retire weak signals.
- Govern without gridlock: Clear policies for data privacy, model risk, and audit trails. Pre-approved patterns to speed delivery.
- Vendor strategy with discipline: Mix build and buy. Demand ROI proof, not demos. Own your data and integration layer.
How to build confidence in the next 12 months
- Embed AI where revenue moves fastest: Demand gen, pricing, personalization, sales support, and service deflection.
- Upgrade forecasting and planning: Scenario models for supply, demand, and margin-then link them to decision rights.
- Make bold bets where you have edge: Data advantage, distribution, or adjacent products. Exit pilots that don't clear the bar.
- Use deals to buy speed: Acquire data, capabilities, or distribution. Partner where build-time kills ROI.
PwC also found a strong link between more revenue from new sectors, higher margins, and higher CEO confidence. Reinvention beats waiting.
The takeaway for executives
AI isn't a silver bullet. It's a system. The companies pulling ahead are treating it that way-anchored to P&L, powered by clean data, delivered by product teams, and measured relentlessly.
If your plan depends on people using new tools well, invest in skills. A focused upskilling path can close a big share of the gap. Curated options by role are here: Complete AI Training: Courses by Job.
Your membership also unlocks:
