AI Investment Is Doubling. Most Companies Still Can't Prove the Return.
IN ONE MINUTE
AI spending is rising fast, but most CEOs still can't point to a financial return. In PwC's 2026 Global CEO Survey, 56% of 4,454 CEOs reported no significant financial benefit from AI so far, and only 12% saw gains in both revenue and cost. The research points the same way every time: the gap is rarely the technology. It is how the investment is organised, owned and measured. Before approving the next initiative, leaders can ask three questions: is success defined at production scale, is there one named owner, and are we funding the people as well as the platform?
THE OBJECTIVE
This issue looks at why AI budgets are growing faster than AI results, and offers a short checklist any executive team can use before its next AI steering committee.
WHAT THE RESEARCH SHOWS
The investment is real. Gartner now forecasts worldwide AI spending of about $2.7 trillion in 2026, up almost 50% on last year. BCG's AI Radar 2026, a survey of 2,360 executives including 640 CEOs, found companies plan to double their AI spend to around 1.7% of revenue, and 72% of CEOs now say they are the main decision-maker on AI, twice last year's share.
The confidence is real too. In the same BCG survey, 90% of CEOs said they expect AI agents to produce measurable returns this year.
The results are harder to find:
- PwC: 56% of CEOs report no significant financial benefit yet; 12% report both revenue and cost gains.
- MIT's NANDA initiative (2025): around 95% of generative AI pilots studied showed no measurable impact on the P&L.
- S&P Global (2025): 42% of companies abandoned most of their AI initiatives.
PwC adds a useful detail: the CEOs seeing both cost and revenue gains were two to three times more likely to have embedded AI extensively across their operations. The difference sits in how the work is organised, not which model is used.
WHY THE MONEY ISN'T SHOWING UP YET
Three patterns come up again and again in the research, and in my own experience.
- AI is run as a technology project. A tool is selected, a pilot works, the demo impresses. But the workflow around it stays the same. A pilot proves the technology works. It doesn't prove the organisation can change around it.
- Returns are judged on the wrong clock. The BCG "future-built" companies, about 5% of the sample, reported 1.7x the revenue growth and 3.6x the three-year shareholder return of laggards. Those gains build over years, not quarters. A 90-day review can end a programme just before it pays back.
- The budget goes to the platform, not the people. Deloitte's CTO Bill Briggs has described companies spending roughly 93% of their AI budgets on technology and 7% on people. Adoption follows the money.
THE FRAMEWORK: THE SCALE-READY TEST
Three questions to ask before approving, or reporting on, any AI initiative.
| Question | What a good answer looks like |
|---|---|
| 1. Is success defined at production scale? | Metrics describe changed workflows and business outcomes, not model accuracy in a demo. |
| 2. Is there one named owner? | A business leader whose objectives depend on the result, not a committee that "oversees" it. |
| 3. Are we funding the people as well as the platform? | Budget covers role redesign, training and incentives, so using AI becomes the easiest way to do the job. |
If any answer is unclear, the initiative may be a good pilot, but it isn't yet ready to scale.
Questions leaders ask
Why do most AI pilots fail to scale?
Because the pilot tested the technology, not the change. A pilot proves the tool works; it doesn't prove the organisation can work differently around it. Before scaling, define success as changed workflows and business results, name one business owner, and fund training and incentives as well as the platform.
How long should an AI programme take to show ROI?
It depends on the bet. Quick automation wins can pay back within a year. Redesigning how a function works takes longer: BCG's leading companies built their gains over years, not quarters. Judge larger programmes on staged milestones, or a 90-day review may end one just before it pays back.
Who should own AI transformation?
The CEO sets direction, but each outcome should sit with the business leader who owns the P&L the AI is meant to improve. A committee that only oversees rarely delivers. Before approving an initiative, check there is one named owner whose own objectives depend on the result.
THE LAST WORD
AI budgets buy technology. Only leadership buys the change that makes it pay.
Your organisation is probably spending more on AI this year than last. Where has that money changed how people actually work, not just which tools they have?
—
Until next week,
Artificial Ian
Tell your network: if a colleague is preparing an AI business case this quarter, this checklist may save them a meeting.
→ Found this useful? Forward it to a leader who chairs an AI steering committee.
→ Subscribe to The AI Executive Briefing, weekly in your LinkedIn inbox.
Sources: PwC 29th Global CEO Survey (Jan 2026); Gartner AI spending forecast (Sept 2026); BCG AI Radar 2026 (Jan 2026); BCG, The Widening AI Value Gap (Sept 2025); MIT NANDA, The GenAI Divide (Aug 2025); S&P Global Market Intelligence (2025); Fortune interview with Bill Briggs, Deloitte (Dec 2025).