2030: The Real Problem Isn’t the Skills Gap. It’s a Training Model That No Longer Works.

Let’s start by clearing something up.

The figure is well known. The World Economic Forum estimates that 39% of current skills will be transformed or obsolete by 2030. You’ve read it. You may well have put it in a presentation. It circulates through leadership committees, HR seminars and annual reports.

And precisely because it circulates so widely, it no longer produces the effect it should. Familiarity with an alarming statistic eventually numbs rather than galvanises — not through indifference, but through a very human tendency to normalise.

That figure, however, is not what concerns us here.

What concerns us is what it conceals. Behind the statistic lies a signal that most organisations have yet to genuinely address: the growing dissonance between existing training architectures and the true nature of the transformation required.

This is not a question of speed or budget. It is a question of paradigm.

Training as a Response — and Its Structural Limits

In most organisations, the dominant response to the skills gap looks something like this: train more, train better, train faster. Invest further in development programmes, broaden catalogues, multiply digital learning pathways, roll out more ambitious development plans.

These efforts are real. They reflect a genuine awareness of the challenge. But they are, to an increasing degree, insufficient — not for want of will, but because of a structural mismatch.

Here is why.

Training rests on an implicit assumption: we know what needs to be learnt. A need is identified, named, categorised, funded and delivered. This model works remarkably well in a world where roles evolve slowly, where competence is a destination — you acquire it, certify it, bank it — and where obsolescence is predictable and sector-specific.

That world is closing.

What is now taking shape, structurally, is a world where competence has become a continuous trajectory rather than a fixed asset; where obsolescence cuts across sectors and is partly unpredictable; and where value lies not in holding knowledge, but in the capacity to recombine, unlearn and relearn.

A training plan is, by its very nature, behind the curve. It catalogues needs that have already been identified, named and budgeted. That process takes time — sometimes six months, sometimes a year. And in the meantime, the world keeps moving.

The Half-Life of Skills: A Concept That Demands Urgent Attention

Physicists use the concept of half-life to describe the time it takes for half of a radioactive element to decay. Borrowed from the sciences, the concept now applies to professional skills with striking accuracy.

The half-life of a technical skill in software development used to be measured in years. Today, it is measured in months. In domains directly exposed to generative artificial intelligence, know-how acquired in 2022 has already lost a significant share of its operational relevance by 2026.

This acceleration is not uniform — it varies by sector, by role, by function — but the direction is clear, and several converging forces are amplifying it.

Generative AI is compressing cycles of cognitive obsolescence. Skills once considered stable — writing, data analysis, certain forms of diagnosis — are being partially substituted or augmented at an unprecedented pace. What is changing is not only what we do, but the relative value of what we know how to do.

Evolving organisational models are fragmenting traditional learning communities. Project-based structures, hybrid working and networked organisations are dissolving the informal spaces where tacit knowledge was once passed on. People learn less through osmosis, less through mentorship — and formal training programmes do not always compensate for that loss.

The pressure for immediate productivity creates a structural tension. Performing today and transforming for tomorrow are two demands that managers and their teams must reconcile every day. In that conflict, the short term almost always wins — pushing investment in transformation back to later, which often means too late.

The French regulatory framework remains built around training as a discrete, fundable act. The CPF, the skills development plan, certification schemes — all are designed around the logic of an identifiable, measurable, attributable training event. This legal architecture makes it difficult, sometimes impossible, to fund what would often be most strategically valuable: continuous learning, integrated into the flow of work, and impossible to formalise in hours and certificates.

Three Forward-Looking Hypotheses for Leaders

Hypothesis 1: The Obsolescence of Three-Year Plans

If the pace of obsolescence accelerates further with the widespread adoption of agentic AI between 2026 and 2028, today’s three-year training plans will be structurally outdated by the time they are executed. Organisations that have not shifted towards a model of continuous learning embedded in work will find themselves having invested significant resources in training that is costly and largely irrelevant.

The direction of this hypothesis is near-certain. Its intensity is less so.

Hypothesis 2: The Social Tension Around the Demand for Adaptability

If this signal combines with growing cultural resistance around the meaning of work and a rejection of permanent availability, an explosive contradiction emerges: organisations will require continuous skills transformation, while a significant portion of employees will refuse the model of an individual perpetually reconfiguring themselves to meet business needs.

The expectation to “become a learner” risks being experienced, in certain contexts, as a new form of alienation — placing responsibility for obsolescence on individuals who did not cause it. This tension is already perceptible. It could become a meaningful source of social conflict.

A high degree of uncertainty, but the signal is there.

Hypothesis 3: The Risk of the Training Illusion

This is the most understated hypothesis — and probably the most dangerous. Organisations will accumulate training hours, report rising metrics, mobilise growing budgets, and arrive at 2030 having trained their people in skills that bear little relation to the actual roles of tomorrow. Not through negligence, but because no one had the courage to do the harder work: identifying what truly needs to change — and therefore what needs to be let go.

Training can become a ritual. A signal sent inward and outward: “we are taking action.” A mechanism of collective reassurance in the face of a transformation that no one truly controls. There is nothing shameful in that. It is a deeply human response to uncertainty that is genuinely difficult to sit with — but it needs to be named before it can be moved beyond.

What “Skills Transformation” Actually Means

Moving from a training logic to a skills transformation logic is not a question of pedagogical format or digital tooling. It means accepting three genuinely difficult shifts.

First shift: unlearning is as strategic as learning.

The most strategically significant meta-skill between now and 2030 is not a technical competency. It is what is often called unlearning: the capacity to actively dismantle patterns, reflexes and areas of expertise that have defined someone’s value for years — in order to make room for what is coming.

This capacity is almost entirely absent from current competency frameworks and training plans. We know how to measure what we acquire. We do not know how to manage what we let go. And yet the transformation ahead depends as much on what we stop doing as on what we start.

Second shift: HR timescales and transformation timescales are incompatible.

HR cycles — annual appraisals, succession planning, training budgets, workforce planning — are structured around annual or three-year rhythms. The skills transformation underway demands near-continuous adjustment, responsive to emerging signals on a quarterly basis at most.

These two timescales cannot coexist without friction. As long as the HR cycle itself is not called into question, organisations will be managing the appearance of transformation — not the transformation itself.

Third shift: the lever is not training. It is the organisation.

Training individuals within an organisation that does not create the conditions for real learning is like pouring water into a sieve. How many organisations are genuinely prepared to protect time for learning, to tolerate mistakes during transitions, to accept that people will not perform at full capacity while they are transforming? What is most often missing today is managers capable of holding uncertainty around skills — not just performance.

The debate about skills for 2030 remains overwhelmingly focused on what individuals need to learn. The structural blind spot is what organisations need to change in order to make that learning possible. Shifting the lens — from the individual to the system — is the essential precondition for a response equal to the challenge.

The Questions Leaders Should Be Asking Right Now

This analysis points to three questions that we believe every leadership team should be addressing before the end of 2026.

Your training plans cover what you already know how to name. But the transformation ahead will require capabilities that no one in your organisation can yet quite articulate — because the working situations that will make them necessary have not yet stabilised. That is not a failure of foresight. It is simply the reality. The strategic question to ask is: does your organisation create the conditions for these capabilities to emerge and be progressively named — or does it wait for them to be catalogued elsewhere before taking them on board?

Post-training satisfaction scores measure comfort. What matters is the actual transformation of professional behaviours at 90 days. Organisations that begin to track this indicator are making a genuine paradigm shift. Others are accumulating data that reassures without informing. The strategic question to ask is: do we measure what actually changes in day-to-day working practices after our training programmes — or only how participants felt about them?

Funding training and enabling learning do not automatically go hand in hand. The former requires a budget and a catalogue. The latter requires an organisational architecture deliberately built for continuous learning: spaces to experiment, to make mistakes, to integrate. The strategic question to ask is: are we an organisation that genuinely enables people to learn — or one that simply funds training?

The real question for 2030 is not “do you have the right skills?” It is: “are you a learning organisation?”