Every major piece of AI research published in the last eighteen months lands on the same conclusion. Not roughly the same — the same. The gap between AI adoption and AI value is not a technology problem. It is an organizational design problem. And the companies that close the gap don't spend more on AI. They spend differently.
I work in the Gulf, where I see this pattern first-hand in nearly every conversation with senior leadership. The appetite for AI is real and growing. But appetite alone doesn't produce results. What follows is what the research says about why most AI investments stall — and what the minority who succeed are doing differently.
The scale of the gap
MIT's CISR research group surveyed 721 companies and found that enterprises in the first two stages of AI maturity — experimenting and piloting — had financial performance below their industry average. Only those in stages three and four, where AI is scaled across operations, performed above it. Just 7% of companies had reached that final stage.
BCG's 2024 research tells a similar story from a different angle: out of 1,000 executives surveyed across 59 countries, only 4% had built cutting-edge AI capabilities that consistently generate significant value. Another 22% were beginning to see returns. The remaining 74% were still struggling to move beyond proofs of concept.
95% of enterprise generative AI pilots produced no measurable return on the $30–40 billion invested.
— MIT GenAI Divide Report
These aren't fringe studies. This is MIT and BCG, independently, repeatedly, arriving at the same finding.
Why the gap exists: the 70/20/10 pattern
The BCG data reveals something that should change how every executive thinks about AI investment. Among the companies that successfully scaled AI, the resource allocation follows a consistent pattern:
RESOURCE ALLOCATION IN SUCCESSFUL AI COMPANIES
Seventy percent. Not on better models. Not on more compute. On change management, workflow redesign, product development processes, talent, and governance.
Most companies allocate in the opposite direction — they over-invest in the technology and under-invest in the organizational work needed to make that technology produce results. BCG found that around 70% of AI implementation challenges stem from people and process issues, while only about 10% involve the AI models themselves. The technology is rarely the bottleneck. The bottleneck is the organizational architecture around it.
MIT Sloan's research on manufacturing firms adds another dimension: companies that adopted AI saw an initial drop in productivity — a J-curve effect — because AI requires systemic change that creates friction before it creates value. Older, more established firms struggled most, with legacy routines and hierarchies resisting adaptation. In those firms, the decline in structured management practices after AI adoption accounted for nearly a third of their productivity losses.
This is not a technology failure. It is an integration and change management challenge, and it's entirely predictable.
What the top performers do differently
Three patterns distinguish the companies that extract real value from AI.
They focus fewer initiatives, not more.
BCG found that AI leaders pursue roughly half as many AI opportunities as their less advanced peers — but they expect more than twice the return on each one. They scale more than twice as many of those initiatives into production. Discipline, not ambition, drives their results.
They buy from specialists rather than building internally.
The MIT GenAI Divide data is striking here: externally procured AI solutions succeeded at nearly twice the rate of internally built systems. Companies that partnered with specialist providers consistently outperformed those attempting to build proprietary AI from scratch. This holds especially in regulated industries where the temptation to build internally is strongest.
They connect AI to enterprise data and workflows.
MIT Technology Review found that companies with enterprise-wide integration platforms were five times more likely to use diverse data sources in their AI workflows. They also had more multi-departmental AI deployment and greater confidence in assigning autonomy to AI systems. The integration layer — connecting AI to operational data, customer systems, and business processes — is where the value compounds.
What this means in the Gulf
The Gulf market has specific characteristics that make these findings particularly relevant. Vision 2040 in Oman, and similar national transformation programs across the GCC, have created strong executive interest in AI. The willingness is there. The budgets are often available. What's frequently missing is the integration architecture — the connection between AI capabilities and operational business data.
Add to that the data sovereignty requirements: many Gulf organizations need AI systems that keep sensitive data within compliant infrastructure. This makes the build-versus-buy decision even more consequential. Choosing the right implementation partner — one who understands both the technology and the local regulatory environment — is not a vendor selection exercise. It's a strategic decision that directly affects whether an AI investment produces measurable returns or joins the 95%.
The pattern across all this research is clear: the companies that succeed with AI don't treat it as a technology purchase. They treat it as an operational transformation — one that requires fewer, better-chosen initiatives, deep integration with existing business data, and a serious investment in the people and processes that surround the technology.
The gap between the 94% and the 14% isn't about access to AI. It's about the architecture to make AI operational.
If you're evaluating AI investments and want to understand where you stand — or how to close the gap — let's have a conversation.
Book a conversation