Strategy Is the Only Moat Left – Why the AI Tool Race Is Already Over
Why is everyone running after the one thing that doesn't really matter?
AI adoption and tools are nice, shiny things, but in a business context, they don’t really matter. The only moat is somewhere totally different. It’s within your company and your brain, and it’s called strategy.
In 2019, I wrote that digital transformation is 95% transformation and only 5% digital. Technology, I argued back then, is a trigger and an enabler – never the solution itself. Companies were buying software for organizations that had neither the knowledge nor the structures to do anything with it, and then wondering why 80% of their transformation projects failed.
Seven years later, we have replaced “digital” with “AI,” the software licenses with copilot seats, and the buzzwords with a fresh vocabulary of agents, orchestration, and reasoning models. The failure pattern has not changed at all. It has just become faster, more expensive, and – because AI is now a matter of national policy and not only corporate strategy – more consequential.
Here is the uncomfortable truth that executives and governments alike need to internalize: the AI tool race is already over. Not because someone won it, but because everyone did. When every competitor, every ministry, every mid-sized company has access to essentially the same models at essentially the same price, tools stop being a differentiator by definition. A capability that everyone possesses is not an advantage. It is the entry fee.
What remains – the only thing that remains – is the quality of your strategic decisions and the depth of your organizational transformation. Strategy is the only moat left. And most organizations are still digging in the wrong place.
Index
“We keep buying the noun and skipping the verb”
I love the quote in the title since I first heard it and it is so true. Every technology wave of the last two decades has followed the same choreography (and hype). A new (shiny object) aka capability appears. A new vocabulary forms around it. Vendors translate the vocabulary into products, consultants translate the products into roadmaps and promises, and leadership teams translate the roadmaps into procurement based on FOMO. Cloud, big data, blockchain, metaverse – each cycle, organizations bought the noun and skipped the verb. They acquired digital tools and avoided transformation. Same same but different.
I have watched this pattern from both sides of the table – inside companies and, over the past years, advising governments on their digitalization and AI agendas plus building whole economic development programs to support companies (which is unfortunately also mostly based on the noun). But its not just big corporates or governments that are running all in the same (mis)direction – the pattern is remarkably consistent across sectors, organization sizes and even borders. A national AI strategy that is, on close reading, a data center procurement plan for GPUs as everyone thinks that this is the new thing and I need to give it to Jensen – he is riding the FOMO wave well and even rephrasing everything into “AI Factory” so people think this is where AI is manufactured – very human reaction. A corporate AI transformation that is, on close reading, a license rollout with a training webinar attached and a big AI spending budget attached to it. In both cases the underlying assumption is the same: that acquiring the technology is the transformation. Nobody cares for the rest as long as procurement of AI works.
Lets be blunt here – This “just buy tech” was never the solution and it will never be. In my original article, I made the point that technology should be understood as a trigger for rethinking and an enabler of newly-rethought solutions – and that the actual work happens in culture, in structures, in processes, the strategy and above all in people. You first understand what is changing around you, then you decide what added value you want to create and for whom, and only at the end do you go looking for the technology that enables it. Most organizations run this sequence backwards. They start with the tool and then search, sometimes for years and with expensive consultants with pretty slides, for the problem it was supposed to solve.
AI has not repaired this backwards sequence. As it doesnt understand it and it is also not “intelligent” – it gives you the best sounding answer, but nothing more or less. But one thing it made – it made you hope for a universal solution and the industry has industrialized it. Because generative AI is so easy to deploy – no integration project, no multi-year ERP migration, just a login – the gap between “we have the technology” and “we have transformed” has never been easier to paper over. Everyone thinks by just getting a quick answer in a chat, they transformed the way the company workds. Adoption numbers go up (or the token consumption goes up akak “TokenMaxxing”), dashboards turn green, and the transformation quietly doesn’t happen but everyone keeps thinking they are more productive because they see more.
The evidence has caught up with the argument
For years, the idea that “it’s not about the tools” was something that one had to argue from experience and first principles. Honestly, I often felt alone, as everything felt more like the “Emperor’s New Clothes” story – everyone wants to be part of it so everyone just plays along and even becomes militant about their opinion (just look at your LinkedIn feed and you know what I mean). That’s why it’s refreshing to see that empirical evidence has also caught up. This June, BCG’s fourth annual AI at Work study (source) showed the results of a survey of nearly 12,000 employees, managers and leaders across more than a dozen markets. I would like to summarise the findings briefly because they are the trigger for this article and deserve to be recognised as a quantified confirmation of the 95/5 rule.
Three findings stand out.
First, adoption is essentially solved. Around three-quarters of frontline employees now use AI regularly – a jump of 23 percentage points in a single year – and among leaders the figure is above 90%. The “silicon ceiling” that kept frontline workers from the technology has broken. The tools work, too: 42% of frontline employees who use AI regularly report saving a full workday every week. Notably for a European readership, adoption is not a Western-led story – India and the Middle East lead, while France, Italy, and the US trail the average.
Second, the value largely evaporates. Two-thirds of those same employees receive limited or no guidance on what to do with the time they save, and more than half do not redirect it into more strategic work. The hours the technology liberates simply leak out of the organization. Only a minority of companies have moved beyond deploying tools toward actually redesigning workflows end-to-end or building new business models – and those that have made that move outperform the tool-deployers on essentially every measure BCG tracked, from value captured to employee trust and job satisfaction.
Third, and this is the finding I would frame and hang on every executive floor and in every ministry – strategic clarity beats tool access, even when the tools are worse. BCG compared organizations along two axes: strength of AI strategy and strength of access to AI tools. Employees in organizations with a clear strategy but limited tools reported measurable business impact at 80%. Employees with strong tools but no clear strategy: 60%. Upgrading your tools while your strategy stays vague buys you five percentage points. Upgrading your strategy while your tools stay mediocre buys you twenty-five.
And the final verdict isnt surprising. On the numbers, strategy is worth roughly five times more than tools. The under-equipped organization that knows where it is going beats the fully-equipped organization that doesn’t – and not narrowly, but decisively. That is the 95/5 rule, rendered as a payoff matrix by one of the world’s largest strategy consultancies.
I take no (ok, a little) satisfaction in being right about this for seven years and being the “unsexy person” in the board rooms about boring old things like transformation while everyone wants to get their “AI expert” badge from some kind of LinkedIn post. The pattern was always visible to anyone who looked at transformation as an organizational discipline rather than a procurement exercise. What has changed is that the excuse is gone. Nobody can claim anymore that this is a matter of opinion.
Why intelligent leaders keep optimizing the 5%
If the evidence is this clear, why does the pattern persist? Why do boards, cabinets, and leadership teams full of intelligent people keep pouring attention and budget into the 5% while the 95% goes unmanaged?
The answer is not stupidity (in most cases). As mentioned, it is always a new weave of the “Emperor’s New Clothes” story. It is incentives – and it is worth naming them precisely, because you cannot correct a bias you have not diagnosed.
- Tools are legible and easy to buy; transformation is not. A tool is a line item. It has a price, a vendor, a go-live date, and a logo for the annual report. A transformed decision-making process has none of these things. Boards, budget committees, and parliaments approve what they can see, and transformation is structurally invisible until long after the decisions that produced it.
- Tools offer attribution; transformation offers ambiguity. A leader who signs an AI platform deal can point to it next quarter. A leader who spends two years redesigning how the organization allocates freed-up capacity will see the results emerge slowly, entangled with a hundred other variables, and probably credited to their successor. Rational careerists optimize for what they can be credited with. The system rewards buying the 5%.
- The entire supply side sells the 5%. Vendors sell licenses. Integrators sell implementations. A great deal of the consulting industry sells maturity assessments that count technologies rather than decisions. Almost nobody’s revenue model is built around helping you make one genuinely hard organizational choice – which processes to rebuild from zero, which management layers no longer make sense, what your people should do with the day per week the machine just handed back to them. The market supplies what it can invoice.
- Technology demos; transformation doesn’t. A tool wins the meeting because it can be shown. Thirty seconds of a model drafting a policy brief is more persuasive in a boardroom than any argument about operating models – even though the second determines whether the first ever creates value.
- Vocabulary substitutes for work. Adopting the season’s terminology – this year it is agents and agentic workflows – feels like progress and signals modernity to peers, boards, markets, and voters. The word becomes a stand-in for the work. I have sat in enough strategy sessions, corporate and governmental, to know that fluency in AI vocabulary and depth of AI transformation are almost entirely uncorrelated. Some of the most jargon-fluent organizations I have encountered are among the least transformed.
None of these forces is new. They are the same forces that produced two decades of failed digital transformation programs. What is new is the speed. AI compresses the cycle from buzzword to procurement to disappointment from years into quarters. Organizations no longer have the luxury of failing slowly because they perceived the speed and they could talk to AI in a chat and everyone just thought it is the end of the economy, the world as we know it, their job etc. – so everyone was caught in the craze earlier as it was more personal than the e.g. Metaverse or Bitcoin hype-cycles before – so also the hype will be bigger and the fall/losses from it will be also larger.
The 95% has not changed – but the stakes have
So what actually constitutes the 95% in the AI era? Structurally, it is the same work I described years ago because the hard part of transformation was never technology-specific. But three elements deserve particular attention now, because AI raises their stakes dramatically.
First: deciding what freed capacity is for. This is the most concrete strategic failure in organizations today. The technology delivers exactly what it promises – it automates things which can free up time. And then nothing happens, because nobody made a decision about what those hours are for. Capacity without direction is not leverage; it is just “hidden cost”. If AI gives your organization back the equivalent of 10% of its working time and you have no answer to “and now what? we cut jobs!”, you have not transformed anything. Most of the time you just have subsidized slack. The decision about where reclaimed capacity flows – into deeper client work, into innovation, into faster policy cycles, into growth – is a pure strategy decision. No technology will make it for you.
Second: redesigning work collectively, not equipping individuals. Handing every employee a copilot is the AI-era equivalent of buying software for an organization that has no structures to use it. The value does not sit in any individual’s productivity; it sits in how tasks flow between people, functions, and now machines/systems. That means selecting a small number of core processes and rebuilding them end-to-end around what the technology makes possible – not layering AI on top of workflows designed for a world without it. As I wrote seven years ago: there is no way that a new system can be handled with old structures. The organizations pulling ahead are not the ones with more tools. They are the ones willing to break their own structures.
Third: governing a moving target. AI more of a general-purpose technology (if you include to the LLM hype, also ML, alrogithms etc.). That makes transformation a permanent condition rather than a program with a finish line or some “implementation sprint” – and it makes governance the new bottleneck. The most urgent question is not “should we use tools?” but “who is accountable for the outcomes?” Most organizations, private and public, currently have no answer. Funnily also half of employees in BCG’s survey say their company has no clear guidance for managing mixed human-AI teams – also because its driven by marketing promises and FOMO and not by real business need in most cases, but that is another topic.
The interesting thing is that governments are also trapped in this short term hype-sprint, despite the fact that each of these points has a multiplier effect at a national level. A Technology First approach at sovereign scale involves redesigning administrative processes without planning for workforce transitions or establishing accountability frameworks for algorithmic decision-making. The states that convert AI into productivity, better services and genuine strategic autonomy will not be the ones that procure the most infrastructure, but the ones that use the smartest solutions, as technology makes it possible to rethink centuries-old frameworks. Sovereignty over technology means little without sovereignty over the decisions that technology facilitates, so focus on frameworks for decision-making rather than following the GPU buildout hype, as none of these solve real problems for 99% of businesses and government.
You cannot transform what you refuse to measure (our you measure wrong)
There is a final trap, and it may be the most dangerous one, because it wears the costume of rigor: measuring the wrong things.
Most AI dashboards today measure the 5%. Seats provisioned. Active users. Prompts per day or the worst “tokens” (would you measure how much you spend and increased spending as a success metrics? – some of the stupidest thing I ever heard tbh). Adoption curves created by Microsoft, OpenAI etc. – all just there to make you spend more. These numbers are seductive precisely because they always go up – and they tell you almost nothing or in the worst case they are destructive. Adoption tells you that people use the technology. It does not tell you whether anything of value happens as a result. An organization can score brilliantly on every adoption metric while its strategy remains vague, its saved time leaks away, its processes stay untouched, and its governance doesn’t exist. The dashboard is green; the transformation is nowhere.
If strategy is the only moat left, then the scoreboard has to measure strategy – and the organizational conditions that let it convert into results. In my experience, that comes down to a handful of dimensions that any leadership team, board, or ministry can honestly assess:
- Strategic clarity. Is there an explicit answer to why and where – and does the frontline actually know it?
- Reinvestment discipline. Is freed capacity tracked, and is there a deliberate decision about where it flows? What happens after you optimized/automated/streamlined the first flow(s)?
- Redesign depth. Are you deploying tools onto old structures, or rebuilding core processes end-to-end? How many, and how far?
- People readiness. Is reskilling real or rhetorical? Are employees shaping the redesign, or being presented with it? (Nearly nine in ten people expect they will need major upskilling; barely a third feel properly trained. That gap is a leadership failure, not a skills failure.)
- Governance maturity. Is there clear accountability for human-machine decisions, and a standing mechanism that re-evaluates as the technology moves?
- Say-do alignment. Do the organization’s actions match its AI messaging? Employees (and customers) notice the gap before leadership does – and trust, once lost there, takes the whole transformation down with it.
None of these dimensions asks which AI-model you licensed or how many tokens you burn through for writing an email. All of them predict whether the license will ever pay off. This is exactly the philosophy we have built into our own business diagnostics work at MoreThanDigital: measuring transformation readiness and decision quality rather than tool inventory – scoring the 95% that determines whether the 5% ever creates value. Because what gets measured gets managed, and as long as organizations measure adoption, they will keep managing adoption while the transformation goes unmanaged.
Same lesson, higher stakes
Technologies come and go and currently its the AI-hype where everyone is running after. The tool question is settled – settled for you, and settled for every one of your competitors simultaneously because everyone is now desperately trying to get a toy. This time its just a little faster and more FOMO-fears are coming up in this wave – but its important to know one thing in every technological hypecycle: When everyone owns (or simply can get) the same tools, the tool cannot be the strategy.
What is left is everything the tools cannot buy: clarity about where you are going, the courage to break structures and even break your company (Innovators Dilemma) that no longer fit, the discipline to decide what freed capacity is for, people who are equipped and included rather than surprised, and institutions capable of governing a technology that will not stand still.
In other words: 95% transformation, 5% digital. Seven years after my publication on this, one technological revolution after another, and roughly twelve thousand survey respondents later, the ratio has survived intact. Only the names and topics have changed around it – also AI is just another technology that needs to find its use case and be implemented or not.
So finally a last time – AI did not make strategy less important. Strategy and change management are the only moat in the market and it will also be in 10 years (or earlier when I write the next article).
Comments are closed.