AND Digital says many enterprises are still struggling to turn AI enthusiasm into measurable results, with weak understanding, fragmented implementation, and poor data discipline continuing to block scaled adoption.
The warning follows fresh Forrester analysis pointing to a familiar pattern in the market: widespread experimentation, rising investment, and a much smaller group of businesses that can point to meaningful impact. The businesses performing best, Forrester argues, are the ones focusing less on internal productivity alone and more on how AI creates value for customers. That aligns with a wider shift in enterprise technology spending, where boards are becoming less interested in pilot counts and more interested in whether AI improves growth, service, and operating performance.
Catherine Rousseau, Technical Solutions Director at AND Digital, said: “AI investment is accelerating, but the real determinant of success is whether organizations are ready to deploy AI at scale, and that readiness depends entirely on data quality and governance. When 58 per cent of organizations describe their data as ‘chaos’, it’s clear why many struggle to move from pilots to production.”
AND Digital also pointed to separate research from MIT suggesting many AI initiatives still fail to meet expectations, reinforcing the broader concern around return on investment. That is one reason data governance has become such a central operational issue. Businesses that can standardise access, improve data quality, and create reliable controls are more likely to move beyond siloed proofs of concept. Those that cannot often find themselves with brittle workloads, inconsistent results, and growing compliance questions.
Rousseau added: “The leaders in AI are the ones investing in high-quality data solutions and without governance and reliable data contracts, AI workloads become brittle, costly, and difficult to audit. This year’s surge in AI infrastructure spending will only deliver value for organizations that pair it with disciplined data readiness, turning raw datasets into governed, high-trust assets that can power reproducible and compliant AI pipelines.”
AI adoption is no longer being judged only on whether a model can be deployed. It is being judged on whether data, controls, ownership, and commercial purpose are coherent enough to support repeatable outcomes. In that environment, readiness is turning from a technical detail into a board-level condition for value creation.
