Upskill or hire? Running the numbers on your AI skills gap
Every leadership team now faces the same uncomfortable question: when the AI skills gap shows up on a project plan, do you build the capability internally or buy it on the market? The instinct is often to hire a specialist rather than invest in executive education programmes for the people already on the team. But the numbers tell a more nuanced story, and getting this decision wrong is expensive in both directions.
The scale of the gap is not in dispute. IDC estimates that skills shortages, including the AI gap, will cost the global economy $5.5 trillion by 2026 through delayed launches, quality issues, and lost revenue. More than 90% of enterprises expect to face critical shortages this year, and only a third of leaders feel they have genuinely prepared their people for AI-driven roles. Business schools such as ESCP have seen executive demand for structured AI training rise accordingly. Demand keeps accelerating: job postings requiring AI skills have grown 144% year over year, nearly eight times faster than the overall job market, according to the Bipartisan Policy Center’s AI Skills Dashboard.
That demand shows up directly in pay. PwC’s Global AI Jobs Barometer puts the wage premium for AI-skilled workers at 62%, up from 57% last year and 25% in 2024 : and as high as 118% in some consumer-facing sectors. Half of employers say they struggle to fill AI-related positions at all. Put simply: hiring your way out of the gap is getting slower and pricier every quarter, and the candidates who can actually deliver are aware of their leverage.
So why not simply upskill everyone already on payroll? Because the internal picture is messier than it looks from outside. Skillsoft’s 2026 research found that only 24% of individual contributors strongly agree their employer has adequately prepared them to use AI, even though 77% of their managers believe otherwise – a perception gap that quietly erodes execution. Fewer than a quarter of employees receive AI training before new tools are rolled out, and 58% cite lack of time, not motivation, as the main barrier to building skills. Training exists; readiness often does not.
Running the actual numbers usually favors a blended approach, weighted by role criticality and time horizon. External hiring makes sense when you need a capability immediately, when the skill is narrow and specialized, or when no amount of internal training closes the gap fast enough for a live deadline. But hiring carries hidden costs beyond the wage premium: recruitment fees, onboarding time, cultural integration risk, and the near certainty that the person you hire today commands an even larger premium in twelve months as demand keeps compounding.
Upskilling wins on cost per capability built and on retention, since employees who are invested in tend to stay longer, but it is slower and requires real governance – something only 12% of leaders currently have in comprehensive form. It also spreads capability more evenly across a team rather than concentrating it in one new hire who can leave.
The most defensible path for most organizations is a portfolio one: hire selectively for the roles where AI fluency is now table stakes and speed is non-negotiable, while systematically upskilling the broader workforce for durable, organization-wide AI literacy. That second track is where structured, credentialed learning earns its budget line. Well-designed executive programmes give managers and specialists a shared, rigorous foundation in applying AI to real business decisions, rather than leaving capability-building to ad hoc tutorials that 58% of employees say they don’t have time for anyway.
Business schools with strong applied-AI curricula are increasingly the bridge between the two options : turning “upskill or hire” from an either/or bet into a calculated, ongoing capital allocation decision.