I had an epiphany while hosting an AI conference recently. It’s quite common for CEOs to talk about AI transformation in one way, and Chief AI Officers to speak about different issues.
For many CEOs, AI transformation could mean rolling out licences, redesigning decision-making processes, developing new business models or a combination of all three.
Heads of AI, meanwhile, often talk about governance, security, data issues and problem-plagued pilots.
So, what’s going on? Nearly three in four CEOs say they're the person in charge of AI, so getting on the same page is critical.1
- ~3 in 4CEOs say they own AI1
- ~9 in 10organisations use AI3
- 37%see any earnings impact3
- 6%are high performers3
I spent time combing through surveys, consultancy research and trade press to understand what leaders really mean, where they're coming from and where they say they're going.
One phrase, four different jobs
When a CEO says "AI transformation", they could actually be talking about one of four very different jobs. I’ve outlined them below from least to most effort.
1. Equip the workforce
Licences, tools, training.
Worker access to sanctioned AI jumped from under 40 percent to around 60 percent in a year.2 And eight in ten people say it makes them personally more productive.3 This job's largely under way.
Largely under way2. Redesign the work
Rebuild workflows around AI rather than bolting a tool onto the old process.
McKinsey tested 25 attributes and found workflow redesign had the biggest effect on whether a company saw a financial return.4 Yet only 30 percent of organisations are redesigning key processes, while 37 percent stay at the surface.2 This is the current frontier.
Current frontier3. Rewire the organisation
Roles, decision rights, even the C-suite.
The share of organisations with a chief AI officer went from 26 percent to 76 percent in a year.5 In Australia it's just over 2 in 5.6 Leaders expect the share of operational decisions made by AI to nearly double, to 48 percent.5 It's starting, and it's contested.
Starting, and contested4. Reinvent the business
New products, channels, pricing.
The few companies getting real returns pursue growth and innovation alongside efficiency.3 IBM reports CEO agendas shifting from productivity gains towards business model change.5 For most, it's still ambition.
Still ambition
One of the key distinctions is to note the first two types of AI transformation are funded from CEO budgets, and the third one is a point of contention.
Where leaders are coming from, and where they say they're going
A year ago, AI transformation stories were largely focused on shifting from pilots to live programs.
Just a third of CEOs called themselves the main decision-maker.1 Spend averaged around 0.8 percent of revenue.1 What’s interesting is they have been borrowing a legacy mindset from digital transformation: AI transformation is a program with a start date and an end date.
Today, that legacy mindset is fading fast. McKinsey says AI transformation is a permanent condition.7 Accenture talks about AI-driven reinvention as continuous, not episodic.9
The destination is also becoming clearer as time goes on. In one common scenario, a company has embedded AI inside end-to-end workflows. Agents run the routine tasks and people (try to) keep up with approvals, also known as ‘human-in-the-loop’. People also manage exceptions and remain accountable for AI’s work.8 In scenarios like this, AI spend within an organisation roughly doubles to 1.7 percent of revenue.1
The C-suite also changes as AI maturity develops. Around 59 percent of CEOs are expecting the CHRO's influence to grow.5
Where they're coming from
- AI as an IT project
- About a third of CEOs call themselves the main decision-maker
- Spend about 0.8% of revenue
- A program with an end date
Where they say they're heading
- AI inside end-to-end workflows
- Agents on routine paths, people on approvals and accountability
- Spend about 1.7% of revenue
- A redrawn C-suite
Of course, how this happens in reality is where things get messy. I’ve counted five stages that leaders describe, and there’s a different story for each one.
Stage 1: Experiment
The pitch"Prove it works."
The realityRoughly 75 percent of consumer goods companies are still in pilot mode.10
Stage 2: Equip
The pitchPut tools in every hand.
The realityAccess is near 60 percent of workers, but most firms are building fluency rather than redesigning work.2
Stage 3: Redesign
The pitchRebuild workflows around AI.
The realityNearly three-quarters of high performers did it, against about a quarter of everyone else.3 Only 25 percent of organisations have moved 40 percent or more of their pilots into production.2
Stage 4: Orchestrate
The pitchRun fleets of agents under proper governance.
The realitySome 71 percent of technology leaders expect agentic deployment within 12 months. Just 21 percent have the governance ready.11 Meanwhile, 62 percent of large US organisations are already building or deploying agents.8
Stage 5: Reinvent
The pitchNew growth, new models.
The realityOnly 12 percent of CEOs report both cost and revenue benefits.12 And 6 percent qualify as high performers, meaning AI delivers at least 5 percent of their earnings.3
Most organisations sit between stages two and three. Call it the valley. Individual productivity gains are real, but they're not turning into enterprise returns, because the work around the tool hasn't changed.
How many actually profit from AI?
This is where things really get interesting. Nearly 9 in 10 organisations are using AI in at least one function.3 And around 44 percent are scaling it across the enterprise.3 Yet just 37 percent report any earnings impact.3
Meanwhile, 30 percent are redesigning key processes.2 Twenty-five percent have moved a meaningful share of pilots into production.2 Just 12 percent of CEOs have identified both cost and revenue benefits.12 Six percent are high performers.3
| Measure | Share | Notes |
|---|---|---|
| Use AI in at least one function | Nearly 9 in 10 | 3 |
| Scaling across the enterprise | ~44% | 3 |
| Any earnings impact | 37% | 3 |
| Redesigning key processes | 30% | 2 |
| Moved a meaningful share of pilots to production (40%+) | 25% | 2 |
| Both cost and revenue benefits (CEOs) | 12% | 1,12 |
| High performers (5%+ of earnings from AI) | 6% | 3 |
Now, an important qualifier: these stats come from different surveys that define things differently, so don't think about this as one clean funnel. We’re more interested in the overall shape, or trends. And to that point, the takeaway is that almost everyone's using AI, but very few are actually making money from it.
Sound like your organisation?
Words you hear, and what they really mean
Meanwhile, six phrases keep turning up in executive commentary about AI transformation. I’ve taken a moment to simplify things.
- You hear:
AI is a CEO-level priority.
- What it really means:
Personal accountability has moved up the chain faster than capability. Nearly three-quarters of CEOs call themselves the main decision-maker, and half think their job depends on getting AI right.1 Meanwhile, 61 percent say their boards are rushing the transformation.13 The pressure arrives from above. The capability sits below.
- You hear:
We're becoming AI-first.
- What it really means:
Decision rights are being redistributed between people and machines. That's an organisation-design project, not a software one.5
- You hear:
We're deploying agents.
- What it really means:
Usually that means supervised task automation, not autonomy. Some 62 percent of large US organisations are building or deploying agents.8 But an earlier Gartner survey found only 15 percent of IT application leaders were even considering fully autonomous ones.18
- You hear:
AI will lift productivity.
- What it really means:
This one hides a split. Mercer reports virtually all of 825 C-suite leaders expect some AI-driven headcount reduction within two years.14 Yet KPMG finds only 9 percent of large US CEOs plan AI-driven cuts this year, while 55 percent expect to hire more.15 In Australia, 1 in 5 CEOs rank cutting hiring among their top workforce priorities.16 Same word. Two very different plans.
- You hear:
We're investing in our people.
- What it really means:
Intent is running ahead of funding. Only 18 percent of Australian CEOs say productivity gains are paying for reskilling.16 And in one vendor survey, 92 percent of C-suite leaders admit they're cultivating a favoured tier of AI-fluent staff.17
- You hear:
Transformation never ends.
- What it really means:
There's no finish line, so there's no project plan. Budgets, governance and change capacity have to run as standing functions.7
Six things leaders expect to overcome
Pilots that never become workflows
MIT's widely quoted study found most pilots delivered no measurable profit-and-loss impact.19
Planned responseRedesign end to end, measure the financial impact and lock in senior commitment.3
Data and legacy systems
Only 10 percent of CFOs fully trust their enterprise data, and 86 percent say legacy systems limit AI readiness.20 Half of firms still run more than half their core systems on legacy platforms.20
Planned responseTreat data as an operational asset with a named owner.21
Skills and change capacity
Some 93 percent of firms say workforce challenges are slowing adoption.20 In one vendor survey, 54 percent of C-suite leaders said adoption was tearing their company apart.17
Planned responseHire for AI, data and digital roles – a top priority for 30 percent of Australian CEOs.16
Governance lagging the agents
Only 30 percent say their risk and governance are highly prepared.2
Planned responseGovernance that works across vendors to curb agent sprawl,18 and a head of AI governance. Forrester expects 60 percent of the Fortune 100 to appoint one this year.22
Proving the return
Fifty-six percent of CEOs report no significant financial benefit yet.12 And 42 percent of Australian CEOs can't confidently assess their return.16 Gartner finds just 13 percent of CIOs have seen significant value, even though spending has risen 18-fold in two years.11
Planned responseShow measurable value before spending more.
Alignment above and below
Only 25 percent of CIOs align technology risk with business risk.11
Planned responseCEO-led upskilling sessions for boards.13
What could go wrong? Lots of things
The good news is leaders have identified most of the risks. The bad news is it's a long list.
Risks
- Risk compounding faster than value. Some 86 percent of CIOs say AI risk is growing faster than AI value.11
- Cancelled projects. Gartner forecasts 40 percent of agentic AI projects will be scrapped by 2027 over cost, unclear value or weak risk controls.18
- Cyber exposure. Thirty-one percent of CEOs now name cyber risk a major concern, up from 24 percent.12 About 9 in 10 large US CEOs worry about AI-powered malware and phishing.15
- Workforce trust. Forty percent of workers fear losing their job to AI.14 And 60 percent of companies in one vendor survey plan to lay off non-adopters.17
- Cuts that don't pay. Gartner reportedly finds workforce reductions are common but not a reliable driver of AI return.23
- Regulatory fog. Only 13 percent of Australian CEOs say AI regulation gives them clear guidance.16 And 83 percent of leaders say AI sovereignty is essential to strategy.5
Opportunities
Now the upside.
- Growth, not just efficiency. The winners use proprietary data to open new products, channels and commercial models.9
- Back-office returns. MIT found the biggest return in back-office automation, even though most budget goes to sales and marketing.19
- Redeploy rather than replace. IKEA reskilled 8,500 call centre workers to sell interior design while chatbots took the routine calls.11 McKinsey Global Institute sizes the prize at about US$2.9 trillion a year in the US by 2030, if work is redesigned.24
- Australian momentum. Every Australian CEO surveyed reports measurable results, and 80 percent plan to lift AI investment this year.16
Four patterns worth pushing on
Now let’s look at four patterns we’ve identified from multiple sources.
- Scope drift hides the real bet. The same phrase covers a licence rollout and a business-model rebuild. Ask which of the four jobs is actually funded.
- Accountability is outrunning capability. CEO ownership has roughly doubled,1 yet only 13 percent of CIOs report significant value and 21 percent have agent governance.11 The pressure's arriving before the machinery.
- Personal gains aren't being banked. Eight in ten report better individual productivity. Just 37 percent see an earnings impact.3 Time saved isn't turning into margin, capacity or growth, because the surrounding work hasn't changed.
- The unresolved question is a human one. Leaders haven't settled whether productivity means fewer people or more capacity. And employees can read that ambiguity.
Point four is notable.
AI transformation isn't a technology problem – it's a human one.
The stories people tell themselves about what AI means for them and their jobs are shaping what happens next.
Which of the four jobs are you really doing?
If you're leading an AI program and you're not sure which job you're actually funding, I'd love to help you work it out. It's the conversation at the heart of my Humans Required keynote. Get in touch and let's have a crack at it.
Notes
- BCG, AI Radar 2026, January 2026, and World Economic Forum summary. ↩
- Deloitte, State of AI in the Enterprise 2026, surveyed August to September 2025, and press release. ↩
- McKinsey, The state of AI in 2026: On the road to ROI, August 2026. ↩
- ServiceNow, Your workflows weren't built for the AI era, May 2026, citing McKinsey. ↩
- IBM Institute for Business Value, 2026 CEO Study, May 2026. ↩
- Datacom research, via AAP. ↩
- McKinsey, State of Organizations 2026, March 2026, via Unleash. ↩
- KPMG, AI Quarterly Pulse Survey Q3 2026, September 2026. ↩
- Accenture, via World Economic Forum, January 2026. ↩
- BCG and The Consumer Goods Forum, June 2026, via the BCG newsroom. ↩
- Gartner, AI risk is rising faster than AI value, 19 October 2026, with InformationWeek. ↩
- PwC, 29th Global CEO Survey, January 2026. ↩
- BCG, Split Decision: The CEOs and Boards Survey, via Let's Data Science. ↩
- Mercer survey of 825 C-suite leaders and Global Talent Trends 2026, as reported by International Business Times. ↩
- KPMG US CEO survey, via Axios, March 2026. ↩
- EY-Parthenon CEO Outlook, Australia, May 2026 and October 2026. ↩
- Writer, 2026 AI adoption in the enterprise. Vendor survey. ↩
- Gartner agentic AI survey of IT application leaders, 2025, and cancellation forecast via OODA Loop. ↩
- MIT NANDA, The GenAI Divide, 2025, via Fortune. ↩
- RGP, CFO survey, December 2025, and Slalom AI Insights Survey, via Digit. ↩
- Entrepreneur UK, How AI is redesigning business models, July 2026. ↩
- CIO Dive, five CIO predictions for AI in 2026, citing Forrester. ↩
- AI Agent Store, weekly signal, May 2026, summarising Gartner. Secondary report. ↩
- McKinsey Global Institute, Redesign work for people and AI. ↩
