Companies Are Buying AI Fast — But Most Aren't Using It Well Yet, Says McKinsey
McKinsey, one of the world's largest consulting firms, has published its 2026 annual report on how organizations are using artificial intelligence. The report, released August 25, 2026, follows a year in which the share of organizations using generative AI jumped from 33% to 71%. Generative AI is the kind of artificial intelligence that creates new content — text, images, or code — rather than just analyzing existing information. Global private investment in this technology reached $33.9 billion over the same period. (McKinsey Greater China, LinkedIn)
The report's main finding sounds simple but carries real weight: the value of AI comes from changing how a company runs, not from picking the best AI tool. McKinsey tested 25 different factors across organizations of all sizes and found that companies willing to restructure their operations — not just buy technology — are the ones seeing returns. (McKinsey QuantumBlack)
Think of it like buying a high-end kitchen. If you install professional-grade appliances but keep cooking the same meals the same way, you have spent a lot of money for very little change. The equipment matters, but only if you also change how you cook.
McKinsey has tracked this pattern across multiple years of surveys: companies buy the technology faster than they redesign their processes, and that gap is where projects stall.
A second McKinsey report, "State of AI Trust in 2026: Shifting to the Agentic Era," published March 25, 2026, looks specifically at trust. It identifies ten insights across three themes: the current state of AI trust, emerging risks, and governance. The average score for how mature organizations are at managing AI responsibly increased to 2.3 in 2026, up from 2.0 in 2025 — a small improvement on whatever scale McKinsey uses. (McKinsey Tech and AI)
The report's subtitle points to where things are heading: from AI that generates content toward "agentic" systems. These are AI programs that can take multi-step actions on their own with less human supervision — for instance, an AI that doesn't just draft an email but independently researches a topic, writes the message, schedules a meeting, and follows up afterward. That raises new questions about trust and control that current safeguards were not designed to handle.
Separate McKinsey findings from "The State of Organizations 2026," based on responses from 10,018 people, identify the top barriers to AI adoption as concerns about AI, ethical concerns, and organizational challenges. (McKinsey) What is notable is what does not top the list: cost and technical capability. The friction is human, organizational, and ethical.
External data supports the adoption trend. Deloitte's 2026 State of AI in the Enterprise report found that worker access to AI rose by 50% in 2025. (Deloitte) Combined with McKinsey's 71% adoption figure, this paints a picture of AI tools moving quickly from experimental labs into the hands of everyday workers. QuantumBlack, McKinsey's AI division, publishes ongoing research on how organizations can use AI responsibly, and the State of AI series is its flagship annual report. (McKinsey QuantumBlack)
The broader context here is the gap between how widely AI is being deployed and how deeply organizations have actually changed. Adoption numbers like 71% and 50% growth in worker access look impressive, but McKinsey's own conclusion — that value comes from rewiring how companies run — quietly acknowledges that most organizations have not yet done that rewiring. The responsibility maturity score of 2.3, while improving, sits closer to the bottom of the scale than the top. And the barriers identified in the State of Organizations data point to leadership, ethics, and change management — not a shortage of computing power or model quality — as the real constraints.
For anyone working in technology, the practical takeaway is this: the infrastructure and the AI models themselves are increasingly solved problems. The hard part has shifted to redesigning processes, building governance that can handle AI systems acting on their own, and helping workforces adapt. Organizations that treat AI as a simple purchasing decision rather than a transformation of how they operate are the ones the data suggests will spend heavily and gain little.
The agentic-era framing in the trust report is the thread to watch going forward. AI systems that take multi-step actions independently introduce questions of delegation and oversight that current governance frameworks, built for earlier generations of AI, were not designed to address. McKinsey's ten insights on emerging risks presumably begin to map that territory, though the low maturity score suggests most organizations are still catching up to today's generative AI, let alone preparing for systems that act on their own.
The overall picture from McKinsey's 2026 research, alongside Deloitte's enterprise data, is one of technology spreading faster than organizations can adapt to it. Investment is flowing at $33.9 billion, tools are reaching workers, and adoption has more than doubled in a year. But trust, ethical governance, and operational change lag behind. The value, as McKinsey puts it, is in the rewiring — and most organizations are still in the wiring-up phase.
