AI is getting cheaper, faster, and more powerful.
Is it getting more valuable?

Few leaders expected 2026 to move this quickly. Models have become faster, cheaper and more capable at a pace that has outrun most planning cycles. According to Stanford’s 2025 AI Index, the cost of running a system performing at GPT-3.5 level fell more than 280-fold between November 2022 and October 2024, a shift that has rewritten the economics of deployment.

The frontier is also leaving the screen. World models, systems that learn how physical environments behave, and physical AI are moving from research labs into commercial deployment. Tesla, Boston Dynamics and a broader wave of robotics players are progressing at a pace that would have seemed implausible three years ago. That matters well beyond the technology sector. Leaders in manufacturing, logistics, healthcare and infrastructure should be watching now, and not after competitors have moved. When machines can perceive, plan and act in the real world, the implications for operations, safety and workforce planning extend far beyond software.

The technical narrative is remarkable. The harder question is whether the business narrative is keeping up. Cheaper models and faster progress are inputs, not outcomes. Value is what organizations do with them.

From boardroom conversation to boardroom discipline

AI has migrated from IT budgets to the board's agenda, and that shift brings something valuable, i.e., financial discipline. The question is no longer about adopting AI - it has matured, and now it's about “What are we actually getting for it?” McKinsey’s 2025 global survey found that 88% of organizations use AI in at least one business function, yet only 39% report any significant impact on EBIT, and for most of those enterprises, the contribution is less than 5%. On the other hand, nearly 66% of respondents say their organizations have not yet begun scaling AI across the enterprise.

Enterprises are beginning to weigh the full cost of adoption, in time, capital, change management and organizational disruption, against measurable returns. Gartner predicted that at least 30% of generative AI projects would be abandoned after proof of concept by the end of 2025, citing poor data quality, escalating costs and unclear business value. This scrutiny is healthy and overdue.

But talking about AI in the boardroom is not the same as governing it there. Moving from conversation to discipline, with clear investment criteria, measurable return expectations and honest reporting, is the step most organizations have not yet made. The companies with real AI advantage are those with the clearest view of what each tool delivers, and why. Simply plugging sophisticated tools isn't enough.

Capability, not seniority, will decide who wins

Future roles will be AI-assisted. That is already happening in many knowledge-work environments, and the scope continues to expand. The World Economic Forum’s Future of Jobs Report 2025 expects 39% of workers’ core skills to change by 2030. Training budgets alone will not close that gap. The practice of embedding AI in real work has to be deliberate and directional.

The real argument is not that AI will take jobs. It is more specific and more urgent: people who use AI well will hold a structural advantage over those who don’t, regardless of seniority or technical background. Those who learn to use AI to amplify their judgement, domain knowledge and decision-making will build careers. Whereas others who wait for the tools to do the work will face unmet expectations.

Building that capability is leadership's primary responsibility, and HR can only enable it. It means redesigning roles, making time for learning and rewarding people who apply AI with rigor rather than enthusiasm alone. Organizations that make this investment will outcompete those that don’t.

Regulation is the scaffolding for scale

The regulatory environment is maturing alongside adoption, and for good reason. According to Stanford’s AI Index, legislative mentions of AI rose 21.3% across 75 countries in 2024. AI introduces threats to privacy, security, labor markets and the integrity of information that existing frameworks were never built to address.

The emerging response is not a brake on AI’s potential. It is the structure that allows adoption to scale without accumulating liabilities that eventually outweigh the benefits. Organizations that treat regulation as a compliance cost rather than a governance framework are misjudging the stakes, and will find themselves retrofitting controls onto systems already embedded in critical operations.

While the technical progress continues with great momentum, what remains open is what organizations, leaders and individuals do with it. The challenge for every executive team is direct.

The organizations that can answer yes to all three will be best placed to capture what AI is actually making possible, and to do it responsibly. Technical progress is not the same as business value.

Our obligation is to make it so.