KPMG 2026 CEO Outlook: Why CEOs Must Lead AI Adoption From the Front

Why CEOs Must Lead AI Adoption From the Front

KPMG 2026 CEO Outlook: A CEO Perspective

I have experienced both sides of this report.

For years I ran a business with the full weight of decisions, much like the ones it describes. These days I spend my time advising leaders through exactly this kind of change, which means I read the 2026 KPMG US CEO Outlook Pulse Survey with the particular scepticism of someone who has been handed a great many confident forecasts over the years. I have to admit that it held up better than most.

Drawn from 100 leaders at large American firms, it captures something I recognise straight away: a strange mix of real uncertainty and unshakeable commitment. Executives are facing genuine turbulence on trade, tariffs and regulation, and they are still backing AI without flinching. For 52% of them there is now enough certainty to make significant investment decisions, and even the sceptics are pressing on. AI has stopped being a line item for experiments. It has become a required investment, and the person expected to own that call is the chief executive. On that central point, the report and I are in complete agreement.

What follows is my reading of the report, offered as an endorsement with a few honest caveats. The numbers are sound and the direction is right. The harder questions, in my experience, sit in the spaces the data cannot quite reach, and I will come to those.

Policy uncertainty is making agility the CEO's defining skill.

The backdrop to every decision in the survey is uncertainty. Trade, tariffs, interest rates and regulation are all in flux, and 52% of CEOs name policy uncertainty as the single biggest pressure on their short-term decision-making. Anyone who has led through a downturn knows the reflex this triggers, which is to freeze and wait for the fog to clear. What struck me about the data is how few of these leaders are doing that. They are using AI as a tool for resilience instead, and nowhere is that clearer than in the supply chain.

A striking 71% of companies are now deploying AI to manage and optimise trade compliance, and 48% are actively modelling or deploying tariff mitigation strategies. These operate as front-line defences against a volatile trading environment rather than back-office tidying, and they only work when a leader has decided that scenario planning belongs on the strategy agenda. That decision has to come from the top. In my own experience, the tools were never the constraint. The constraint was whether the person in charge treated volatility as a reason to hide or a reason to get sharper. The CEOs in this report have largely chosen the second, and it is the right choice.

High AI usage, lagging AI value

Here is the first caveat, and it is one I see almost every week. Usage of AI is close to universal, yet AI-driven value is still lagging well behind the money that has gone into it. Plenty of firms have deployed the technology and are still waiting for the returns to show up anywhere that matters.

The reason is rarely the technology itself. More often it is the way the technology has been introduced. Too many organisations have bolted AI on as a scatter of disconnected tools, each solving a narrow task, none of them changing how the business actually works. The survey is clear that transformative value comes from a different approach, one that embeds AI into new business models and new ways of working rather than layering it over the old ones. I've watched capable teams pour budget into pilots that impressed everyone in the demo and changed nothing in the operation, which is a pattern I wrote about in Beyond the Pilot Trap. The report puts hard numbers behind a lesson a lot of us learned the expensive way.

It also explains a contradiction in the data that I happen to think is entirely rational. While 77% of CEOs believe generative AI may have been overhyped in the short term, they argue its disruptive potential over the next five to ten years is actually underhyped. A quarter of them suspect there is an investment bubble. None of that is slowing their spending, and quite right too, because the short-term froth and the long-term significance are two separate questions. Serious leaders are answering the second one and letting the first look after itself.

Why CEOs think GenAI is underhyped for the next decade

It is worth understanding why executives who freely admit to the hype are still betting so heavily, because the logic is sound. Their case for the long term rests on three ideas.

The first is business model transformation, where the real gains arrive once firms stop deploying AI agents in isolation and start embedding them into how the enterprise is structured. The second is the pace of innovation itself, which AI appears to be compressing, giving firms the agility to move through uncertainty faster than before.

The third is resilience, the idea that by handling routine work AI creates more room for people to invest in the relationships, judgement and contextual knowledge that keep an organisation standing when conditions turn.

I would add a note of caution to the optimism. A leader can believe the technology is overhyped this quarter and underhyped this decade at the same time and hold both views without any contradiction. The danger is using the long horizon as an excuse to avoid the hard, unglamorous work of change today. The firms treating AI as a decade-long re-architecture will pull away, provided they are also doing the difficult short-term work of shifting how people actually operate. The same instinct sits behind the move towards autonomous systems I explored in The Agentic Shift.

The talent gap is being solved from the inside.

An AI-augmented business needs people who can work alongside the technology, and here the survey reveals a preference I strongly endorse. Faced with a shortage of specialists, 61% of organisations plan to close the gap by upskilling their existing employees, well ahead of the 46% focused on hiring from outside.

Some of that is strategic and some of it is a necessity, with around 34% of leaders worried they simply will not be able to find or hire the technical AI talent they need on the open market. I have a bias here, and I will own it. The people who already understand your business, your customers and your particular flavour of chaos are usually a better bet than a new hire who understands the model but not the mission. When the external pipeline is that tight, developing your own people stops being a nice-to-have and becomes the plan. That maps closely onto the staged approach I set out in The 5 Stages of AI Adoption.

The part the report cannot measure: culture and the voices outside the room

This is where my endorsement turns into a challenge, and it is the part I care about most. The report measures budgets, adoption rates and risk registers with real rigour. What it cannot measure is the thing that decides whether any of it works, which is culture. Every figure in this survey depends on people choosing to change how they work, and people do not change their habits because a board approved a capital allocation. They change when they trust the reason, when they are involved in the how, and when the change is done with them rather than to them.

The evidence that this is the soft spot is sitting in the report's own data. While 67% of CEOs say they have an initial perspective on how AI will change roles, only 29% have well-defined roles that genuinely account for it. That gap is not a technology problem. It is a culture and communication problem, and it is exactly the sort of thing that never shows up in a capital plan until it quietly derails one.

I will be blunt about the boardroom, having spent enough time in one. It is the easiest place in any organisation to be wrong with confidence. The people around that table are typically the furthest from the daily work the AI will touch, and they tend to agree with each other more than is healthy. Some of the most useful conversations I have now are nowhere near a boardroom. They are with the person on the service desk who can tell you in thirty seconds why the shiny new tool will be ignored, the junior analyst who spots the flaw everyone senior has learned to stop seeing, and the customer who experiences your process rather than your org chart. Those voices outside the meeting room are the ones that tell you whether a strategy survives contact with reality. The CEOs who lead AI adoption well are not just the ones who set the budget. They are the ones who go and listen where the work actually happens, and who build a culture safe enough for the awkward truths to travel upward.

The real risk is to the leadership pipeline

The most thoughtful concern in the entire report is not technical, and it is one that keeps me up at night more than any cyber statistic. CEOs are worried that automating the tasks traditionally done by junior staff will quietly remove the ground where judgement is built. If entry-level work disappears into an AI agent, younger employees lose their exposure to ambiguity, to failure and to the slow business of learning through getting things wrong. That is precisely the experience that produces good senior judgement a decade later, and you cannot download it.

Alongside it sits a related worry, that an overreliance on AI for decision-making could erode critical thinking across the whole organisation. The pull to accept a confident machine answer without scrutiny is very real when the machine is fast, fluent and usually right. The response so far is early. Around 44% of firms are giving managers guidance on supervising autonomous systems, while only 31% have reached the point of piloting formal programmes to upskill managers on AI oversight. The firms that come through this well will treat AI as a way to free people up for the higher-order thinking that defines good leadership, not as a reason to stop developing it in the first place.

Governance is where CEOs turn risk into trust

Scaling AI raises the stakes on risk, and the survey shows CEOs taking that seriously. Asked what will matter most to their prosperity over the next three years, 60% point to the pace of AI innovation and risk management, ahead of every other factor. I find that framing encouraging, because it treats moving fast and staying safe as the same challenge rather than competing ones, which is how it should be treated.

The specific worries are sharp. Some 91% of CEOs are concerned about data and privacy risks tied to AI agents, along with AI-assisted malware. Fears about AI-assisted phishing sit at 89%, concern about AI agents acting as insider threats at 80%, and 58% are already thinking about quantum computing attacks on encryption. In response, 67% are increasing their cybersecurity investment. The point I would stress to any leader is that responsible innovation is what converts a technological risk into organisational trust, and trust, once you have earned it, is a genuine competitive asset.

This is the part of AI adoption that cannot be delegated. Governance, ethics and the tone set at the top are leadership responsibilities by definition, and getting them right is what lets a business move quickly without moving recklessly. It is the core of how 360 Strategy provides AI consulting in Scotland, because I have come to see strategy and governance as two halves of the same job rather than separate exercises.

Ready to lead your firm's AI adoption with governance and culture built in from the start? Book a Clarity Call and we will map where you actually stand and where the awkward truths are hiding.

What the AI budgets reveal about conviction

If you want to know how seriously a boardroom takes something, follow the capital. On that measure the conviction behind AI is hard to argue with. Some 37% of CEOs are allocating between five and ten percent of their capital expenditure to AI, another 35% are dedicating eleven to twenty percent, and 6 percent are pushing past that into the 21-30% range. For a technology whose returns many of them openly admit are still emerging, that is a remarkable act of faith, and I mean that as a compliment rather than a warning.

They are getting more disciplined about measuring it, too, with 64% having already updated their frameworks to capture the specific value AI generates and 57% reporting that returns from their generative AI investments have met expectations so far. That is a leadership class that is neither starry-eyed nor cynical, treating AI as a serious long-term investment and holding it to account like one. I respect that posture, because it is the one I try to instil in the leaders I work with.

If I could leave a chief executive with a single line from all of this, it would be that leading AI adoption from the front is less about buying software and more about architecting the future shape of the organisation. The KPMG report gets the strategy right. My one addition, earned the hard way, is this: lead it from the front, and listen hardest to the people furthest from the front. That is the difference between a transformation that shows up in the results and one that only ever shows up in the budget.

Mark Evans is the founder of 360 Strategy, an AI strategy and adoption consultancy based in Scotland, and a former chief executive who now advises leaders through AI adoption and change.

Source: 2026 KPMG US CEO Outlook Pulse Survey.