From AI experiments to business value: what matters in 2026
From curiosity to competitive instrument
Artificial intelligence has moved beyond experimentation. According to McKinsey’s 2025 State of AI report, 88% of organisations now use AI in at least one business function, up from 78% the previous year. Yet the majority remain stuck in piloting stages, with only about one-third having begun to scale their programmes. Global AI investment grew by over 71% between 2011 and 2016, and by 2018, marketing, commerce, CRM, and sales had attracted a combined AI investment of $753 million – signalling that AI was no longer a research curiosity. It had become a competitive instrument (McKinsey, 2025).
AI integration had evolved from traditional data collection into a dynamic, predictive system reshaping how organisations make decisions. Technologies spanning machine learning, deep learning, and natural language processing enabled businesses to analyse vast data in seconds, reduce human error, and access real-time market insights previously out of reach. The challenge today is no longer whether to adopt AI, but how to move from isolated initiatives to stable, value-creating systems embedded in everyday operations.
The data from 2025 makes this tension visible: enterprise AI spending grew more than sixfold to $13.8 billion, and three-quarters of knowledge workers were using AI tools daily, yet 42% of firms abandoned most of their AI projects before full deployment, up sharply from 17% the year before. The companies that broke through did so with clarity of purpose. Morgan Stanley’s GPT-powered assistant saved advisors 10–15 hours weekly. Stanford Health Care’s ChatEHR system delivered 30–40% faster chart reviews. Pfizer’s PACT initiative saved 16,000 hours of search time annually and cut infrastructure costs by 55%. What these cases share is not novel technology, it is that business value was defined before implementation, not discovered afterward (McKinsey, 2025), (AI Realized, 2025).
The BCG AI Value 2025 survey put a precise figure to the gap: only 4% of companies had achieved significant value from AI at scale, but those that did realised 1.5× revenue growth and 1.6× shareholder returns compared to peers. Weak leadership, not a lack of technology, as the top reason for failed adoption. Governance matters too: as the EU AI Act came into effect in 2025, ethical frameworks and data privacy standards are prerequisites for AI to deliver genuine organisational value, not constraints on it (AI Realized, 2025), (LXT / Gartner; World Economic Forum, 2025).
The gap between investment and impact
The prevailing challenge in AI adoption is not technological – it is structural. More than half of organisations (56%) spend between $1 million and $50 million on AI annually, and 15% spend $51 million or more. Yet Gartner’s research shows that only 53% of AI projects succeed in moving from prototype to production. Gartner’s AI Maturity Model spans five levels – Awareness, Active, Operational, Systemic, and Transformational – and only 6% of businesses have reached the Transformational level, where AI is inherent in the DNA of the business. The financial services and technology sectors lead, with 43% and 22% respectively reaching the highest maturity levels, sharing a common trait: AI embedded in core business processes rather than treated as an add-on (LXT / Gartner; World Economic Forum, 2025), (Brynjolfsson, Rock & Syverson, 2017).
The root cause of underperformance is rarely algorithmic. A BCG study found that roughly 70% of AI implementation challenges are related to people and processes, with purely technical issues accounting for only about 10% of barriers. This pattern reflects what academics call ‘implementation and restructuring lags, the consistent gap between recognising a technology’s potential and its measurable effects, a dynamic seen with electricity and computers before AI. An example: one organisation launched an AI-powered HR virtual assistant that saw almost no usage because employees were simply unaware it existed, with the only mention buried in an onboarding manual. Introducing technology without managing the organisational change around it leads to wasted investment and none of the hoped-for efficiency gains (Brynjolfsson, Rock & Syverson, 2017), (AI Realized, 2025), (Soni et al., 2020).
This structural gap is compounded by a workforce paradox. A BCG global survey covering over 10,600 leaders, managers, and frontline employees across 11 countries reveals that while more than three-quarters of leaders and managers use generative AI several times a week, regular use among frontline employees has stalled at 51%. In a separate survey of 1,010 C-suite executives, 92% reported up to 20% workforce overcapacity in legacy roles, with nearly half expecting more than 30% excess capacity by 2028, while 94% simultaneously face AI-critical skill shortages in areas such as AI governance, prompt engineering, agentic workflow design, and human-AI collaboration. Nearly 52% of leaders rank job redesign as their top workforce priority, yet only 46% currently integrate workforce planning into their AI roadmaps (McKinsey, 2025), (LXT / Gartner; World Economic Forum, 2025), (Aiimi, 2024).
BMW‘s ‘AIconic’ multi-agent system in its purchasing function illustrates what good looks like: already supporting 1,800 active users and handling over 10,000 searches, it was paired with digital training and dedicated AI innovation spaces for employees at all levels. Their experience shows that AI is not just efficiency – it is about re-architecting roles, workflows, and accountabilities to enable humans and AI agents to co-create value (LXT / Gartner; World Economic Forum, 2025), (Engineering Science & Technology Journal, 2023).
What you measure defines what you lead
Measuring the value of AI is not optional, it is the mechanism by which organisations sustain investment, build confidence, and scale responsibly. The main objectives for AI in the workplace are consistent across the literature: to deliver insights that drive value, to automate business processes and decisions, and to increase productivity and efficiency. Efficiency and productivity gains are seen as the most dominant problems AI can solve (65% of respondents), followed by improved analytics (50%) and business model expansion (48%). Empirical results across industries confirm the potential: AI-enabled predictive maintenance decreased machine downtime by 30% in manufacturing, AI-driven inventory management cut overstock and out-of-stock situations by 25% in retail; and in finance, AI reduced error rates by 20% and accelerated risk management decision-making by up to 40% (Aiimi, 2024), (Engineering Science & Technology Journal, 2023), (Jimenez Castillo, 2023).
GenAI also enables business model innovation: firms can now simulate future demand, generate synthetic personas for market testing, and craft hyper-personalised services, contributing directly to strategic agility and long-term resilience. Maturing organisations treat quality training data as essential, ranking it ahead of quality controls and algorithms, and the highest-maturity companies indicate the strongest need to increase training data budgets over the next five years. Companies actively reshaping workflows with AI save significantly more employee time, sharpen decision-making, and shift work toward more strategic tasks (LXT / Gartner; World Economic Forum, 2025), (Soni et al., 2020), (Moro-Visconti, 2026).
The organisations that capture sustained AI value treat it as an operational and cultural transformation, not a technology project (Zouaghi & Wamba, 2026).
Five priorities stand out for leaders:
Invest in training: Regular AI usage is sharply higher for employees who receive at least five hours of training with access to in-person coaching. Only one-third of employees say they have been properly trained (LXT / Gartner; World Economic Forum, 2025), (O’Brien & Downie, n.d.), (Boyd, 2025).
Secure visible leadership support: The share of employees who feel positive about generative AI rises from 15% to 55% with strong leadership support – yet only about one-quarter of frontline employees currently receive it (Boyd, 2025), (Boston Consulting Group, 2025).
Embed workforce planning into AI strategy: Scenario planning over a five-year horizon should become standard practice. Without aligning skill forecasts and AI adoption plans, companies risk stalled transformations (Zouaghi & Wamba, 2026), (Boyd, 2025).
Track value relentlessly: Track improvements in productivity, quality, and employee satisfaction. Invest in people to reshape workflows and build upskilling and reskilling capabilities (LXT / Gartner; World Economic Forum, 2025), (Boston Consulting Group, 2025).
Build governance frameworks early: Robust governance should include continuous red-teaming of GenAI applications, privacy-preserving machine learning techniques, and real-time anomaly detection (Boston Consulting Group, 2025), (TechClass, 2025).
Conclusion
In 2026, the question is no longer whether AI delivers business value – the evidence confirms that it can and does. The question is whether organisations are structured, led, and culturally prepared to capture that value sustainably. By focusing on people, educating them, empowering them, and easing their path through change, business leaders can unlock the true power of these technologies. The organisations that will lead in the decade ahead are those that move now from pilots to production, from experimentation to integration, and from isolated AI initiatives to enterprise-wide value creation.
