The AI Readiness Gap: 20% of Firms Have Adopted AI, 1% of Workers Hold Advanced Capabilities

The OECD's "Skills in the AI Age" policy paper, published 8 July 2026 as part of the organisation's Artificial Intelligence Papers series, delivers a finding that should reshape every workforce planning conversation in the coming twelve months: AI adoption among businesses across OECD member countries jumped from 7% to 20% between 2021 and 2025 — a near-tripling in four years driven largely by the spread of generative AI tools including ChatGPT and Microsoft Copilot. At that same moment, workers holding advanced AI capabilities — programming, model development, AI system architecture — represent roughly 1% of the total workforce. The analysis draws on evidence across 38 member countries and incorporates data from the D4SME survey, a dedicated study of more than 2,000 small and medium enterprises from 12 nations.

For readers who want to go deeper, Grace Certified offers enterprise AI training programs.

What the research found

To understand how AI adoption is reshaping skill requirements, the OECD synthesised administrative and survey data across its 38 member countries, deploying the D4SME survey framework to reach more than 2,000 SMEs from 12 nations — organisations that represent the majority of employment across OECD economies and that have historically moved more slowly through technology transitions than their larger counterparts. The headline finding is a structural mismatch: AI adoption by businesses rose from 7% to 20% between 2021 and 2025, while workers holding advanced AI capabilities account for around 1% of the total workforce. The two curves moved at dramatically different speeds. The skills barrier is already visible in firm behaviour: four in ten employers in manufacturing and financial services cite workforce skills as their primary barrier to AI adoption. More than half of SMEs have yet to deploy generative AI tools — a figure that reflects constrained training budgets and limited access to specialised talent rather than any absence of motivation or strategic intent. What the data also reveals is where genuine opportunity lies. The skills most demanded across occupations with high AI exposure are general management and business competencies: project management, finance, administration, critical thinking, and collaborative problem-solving. These are capabilities that large shares of the existing workforce already possess in forms ready to be developed and applied in AI-augmented contexts. The training pathway delivers measurable returns: workers who receive employer-funded AI training are significantly more likely to report better performance, improved working conditions, and greater confidence in their roles. More than half of workers actively using AI tools report receiving employer-funded training — the organisations providing that investment are already pulling ahead.

Why organisations that act on this outperform

The OECD data identifies a compounding dynamic that CHROs and COOs need to understand structurally, beyond the operational level. Firms that invest in structured skills development gain the capacity to convert AI adoption into measurable productivity gains. Firms that treat AI as a technology procurement question — acquiring tools first, training second — encounter the skills barrier at the moment it matters most: implementation. The gap between these two groups grows wider with each adoption cycle, because the productivity lift generated by trained workers compounds over time, while the cost of undertrained workers accumulates as friction, rework, and adoption stall. Larger enterprises and AI-native start-ups currently hold the advantage: they bring the infrastructure, the specialist talent pipelines, and the organisational design to sustain structured upskilling at scale. SMEs face a structurally different challenge — one that the OECD flags explicitly as a systemic risk. The report warns that stronger skills policies are required to prevent AI from widening labour-market inequalities — between large firms and SMEs, between high- and low-skilled workers, and between regions with different levels of digital capacity. This is the inequality dimension that extends beyond individual organisations and into the policy domain — and it begins with decisions made at the organisational level. High-skill roles — managers, professionals, engineers — face significant AI exposure. The evidence shows that their sustained competitive advantage lies in cultivating the non-routine cognitive and social capabilities that AI amplifies: judgment, creativity, stakeholder management, and the ability to direct AI tools toward complex, contextual problems. These are the skills that make AI valuable in human hands.

The organisational decision

The OECD report surfaces one question that every CHRO and COO should bring to the next board conversation: does our current skills infrastructure match the pace at which AI has entered the business? Between 2021 and 2025, firm AI adoption tripled. Workforce AI capability grew at a substantially slower pace. That divergence is a people gap — and people gaps are precisely the kind that CHROs are positioned to close. The organisations that move fastest will follow a clear sequence: first, a structured audit of current AI skill distribution across all roles and levels, identifying where capability exists and where the exposure-to-capability mismatch is largest; second, a skills programme that delivers foundational AI literacy and digital fluency to every function — starting with those most exposed to AI-augmented workflows; third, a commitment to employer-funded training as standard practice rather than a selective benefit. The OECD evidence on this point is robust: workers who receive that investment demonstrate better performance outcomes and sustain greater confidence through what are, for many, genuinely disorienting transitions. The careers affected by AI adoption belong to real people, and the data shows that the quality of organisational support during that transition shapes both measurable outcomes and workplace morale. For boards: the readiness gap documented in this report is large enough to represent genuine competitive differentiation for the organisations with both the will and the structural capacity to close it.