Gulf Cooperation Council (GCC) countries are laying the foundation for AI to play a powerful role in the region, committing billions of dollars to cutting-edge infrastructure and technology partnerships that will power AI usage.
For example, Abu Dhabi’s state-linked G42 has announced a number of deals,1 Saudi Arabia’s HUMAIN is driving a massive build-out of AI data centers,2 and the Qatari government is investing in AI cloud capacity.3
Such supply-side activity reflects the region’s determination to position itself as an international hub for AI infrastructure and services—an AI superpower. But what about local demand? Are GCC organizations equally determined to adopt AI, and to what extent are they making progress?
At first glance, they appear to be gaining ground. In our 2023 survey on AI use in GCC countries, 62 percent of respondents said their organizations had adopted AI to some extent.4 Two years later, our latest survey puts that figure at 84 percent (see sidebar, “About the research”).5 But this masks uneven progress.
Some organizations in the region are deploying AI on an impressive scale. Saudi Aramco, for example, used decades of operational data to build a gen AI model with 250 billion parameters, helping it analyze drilling plans, geological data, historical drilling time, and costs.6 And in 2024, Qatar signed a five-year partnership with US firm Scale AI to drive the adoption of AI within government and enhance services.7 Most organizations, however, have not moved beyond pilots. Only 31 percent of respondents said their organizations had reached a level of AI maturity such that AI was being scaled or had been fully deployed across the organization.
Value creation also remains an issue, with many organizations having little to show for their efforts thus far. Only a handful of respondents (11 percent) categorized their organizations as value realizers—that is, organizations that have adopted AI in at least one business function, are scaling or have scaled deployment, and can attribute at least 5 percent of earnings to AI. In short, high usage of AI is out of sync with maturity or value.
Make no mistake—leaders in the GCC region are keen backers of AI. About three-quarters of respondents said their top executives were committed to scaling the technology, and even more said their organization’s AI budgets would likely increase in the coming year. Yet leadership intent might not suffice. Translating intent into wide deployment with bottom-line impact hinges on three components: a business-led AI strategy that chases value, delivery capabilities (the organization’s technology and talent, for example), and a change management program that encourages adoption at scale.8 The value realizers in the survey outperform their peers on precisely these fronts.
The pace of change is fast. Only a year ago, the focus was on gen AI,9 but technology has since leaped ahead; many organizations are now piloting AI agents,10 heralded as marking a major evolution in enterprise AI.11 As technology continues to evolve, building the right capabilities and adoption strategies could prove key to keeping up with leaders in AI adoption when it comes to creating value.
The state of AI
Survey results suggest that most GCC organizations are using AI. A few are doing so at scale and capturing value, but most organizations have yet to move beyond pilots.
Usage
In our 2023 survey, 62 percent of survey respondents said their organizations had adopted AI in at least one business function.12 In our latest survey, the figure has risen to 84 percent—four percentage points lower than that reported by organizations in McKinsey’s global survey (Exhibit 1).13
As in previous years, GCC organizations are most likely to regularly deploy AI in the service operations and marketing and sales functions, where the value of certain use cases has become apparent.14 However, the biggest jump in deployment in the past two years is in product and service development, in which organizations are rewiring end-to-end development workflows and innovation cycles by enhancing existing products with new features and creating new AI-based products (Exhibit 2).15
By sector, 32 of the GCC organizations that report using AI in at least one business function operate in the industry, energy, and infrastructure sector. Financial services and consumer and professional services each account for 22 organizations, followed by 16 organizations in the technology, media, and telecommunications sector and another 16 in the social, healthcare, and education sector. While level of usage differs across sectors, the uptake indicates that AI use is becoming embedded across the GCC economy.
Also noteworthy is the number of organizations using agentic AI—an application of AI technology that many believe will be more transformative in the workplace than gen AI, ushering in a new wave of productivity and innovation.16 Agentic software acts on behalf of a user or a system to perform tasks. Agents can orchestrate complex workflows, coordinate activities among multiple agents, apply logic to thorny problems, and evaluate answers to user queries.17 Essentially, they are transitioning from knowledge-based tools to ones that are more action-based, and they are becoming more accurate in the process.
Sixty percent of survey respondents said their organizations are using AI agents to some extent.18 “There is huge interest in specialized agentic AI models,” said one interviewee. “With earlier models like GPT-2 or GPT-3, you could get outputs but not trust them blindly, whereas some of today’s models are far more accurate and can support real business applications without manual processes. Moreover, these agents have a direct link to the bottom line, cutting SG&A and reducing development cycles, for instance. That makes them very attractive.”
Scale and value
Despite the momentum that these top-line figures evidence, closer analysis suggests patchier progress: While nearly all organizations are using AI, more than two-thirds haven’t moved beyond pilots (Exhibit 3). Said one GCC executive, “Many people still equate AI with models such as ChatGPT. Knowledge of other AI tools and their potential is shallow. So I wouldn’t really say there is adoption at scale. It’s limited.”
Moreover, even though nearly all organizations are investing in AI to some extent, few have yet to extract value from those investments. Only 11 percent of organizations qualify as value realizers, according to survey results.
Capturing scale and value: The capabilities that count
Our survey results suggest that most organizations have aligned their business strategy with their AI strategy and have strong leadership buy-in. Far fewer have what it takes to translate strategic intent into impact.
With 89 percent of respondents planning to increase their AI budgets in the coming year, understanding how to achieve bottom-line impact is an imperative.
A strategic road map is an essential starting point. Often, an organization’s AI strategy is determined by its IT department. But for that strategy to deliver value, McKinsey research shows it should be owned by the most senior executives to ensure alignment with the organization’s business strategy.19 In this way, AI initiatives become a strategic program rather than a collection of scattered efforts.
Survey results suggest that most GCC organizations have strong strategic alignment. But this matters little if the organization is unable to translate strategic intent into impact. Follow-through matters most in the very areas where the survey’s value realizers are often markedly stronger than others: the talent and operating model, technology and data, and the change management initiatives taken to encourage adoption at scale.20 Of the 13 respondents from value-realizing organizations who answered this survey question, 11 said they were strong in each of these three areas. Not even half of the remaining respondents claimed the same (Exhibit 4).
Talent and the operating model
Nearly all GCC organizations in our survey have hired AI talent in the past year—most often data engineers, data scientists, and software engineers. Yet without the right operating model, organizations risk underusing even the best talent. Organizations capturing the most value from AI are often those that combine centralized AI expertise with the executional know-how of the business, working in agile squads.21
Centralized AI talent. New roles are emerging as AI deployment increases—such as forward deployed engineers,22 context engineers,23 and AI product owners—but relatively few people qualify to fill them. To maximize scarce resources, many organizations are turning to a model whereby AI talent is centralized, perhaps in a center of excellence, but deployed flexibly across different business functions and domains. McKinsey research in financial services suggests that 70 percent of organizations with centralized models had progressed to putting pilots into production compared with about 30 percent of those with a decentralized approach.24 Proximity to the business remains critical with cross-functional pods—small teams that bring together engineers and data scientists with business stakeholders. Such a model ensures that ownership sits with the business, domain priorities guide AI development, and value creation stays anchored in business outcomes.25
Additionally, as talent and operating models evolve, organizations must treat AI agents as part of the workforce, managing their performance and capabilities with the same discipline used for people. Those that master this early will translate agentic potential into lasting business value.
Agile ways of working. The value of agile ways of working to speed development is widely recognized, and many organizations are familiar with agile rituals such as sprint demos, quarterly planning, and daily stand-ups. Execution sometimes lacks discipline, however, becoming a box-checking exercise rather than a working practice that drives results.26 Organizations might therefore do well to review their agile working practices.
Technology and data
Deploying AI at scale can be costly given the technology and data requirements. “Many firms lack the capital, which is one factor slowing down adoption, since implementing AI recommendations requires significant investment in automated equipment and infrastructure,” said an executive of a GCC conglomerate. Yet meeting those requirements is fundamental to value creation, as our survey results confirm. According to respondents, most value realizers have a well-established tech foundation and strong data fundamentals.27 Only 37 percent of others boast the same.
Key features of a technology and data strategy that support the scaling of AI include the following:
- Scalable architecture. The architecture will need to be scalable as the deployment of AI evolves, which means building modular components that can be upgraded independently and fungible assets such as libraries of prewritten code that can be used repeatedly for more common tasks.28
- An ecosystem of partners. In fast-moving technology cycles, innovation and costs could suffer without vendor flexibility. Savvy organizations therefore balance best-in-class vendor solutions with open-source tools so they can pivot as technology advances. As a result, independence and bargaining power are protected while new AI technologies can be integrated without a wholesale system overhaul.29
- Data integrity. Reliable data is the lifeblood of AI, and the lack of it is often a barrier to scaling AI. Common issues include poor data quality, missing values, bias, and outliers, all of which can degrade output and expose an organization to risk.30 Fifty-three percent of survey respondents said output inaccuracy was one of the most significant barriers to AI adoption (Exhibit 5). Data integrity can be improved with good governance, which itself becomes easier if data is centralized and assetized to be reusable. But this can take time—which explains why, in the interim, some organizations prioritize AI domains that don’t rely heavily on old data. Within the talent acquisition domain, for example, automating CV screening, interview scheduling, and onboarding can improve efficiency without the need for vast proprietary data lakes.
- AI-native data. Beyond integrity, AI-native transformations rely on data with contextualized storage—capturing conversations, case history, and workflow state—and traceable memory that records how decisions are made and verified. Built on governed, dynamic data building blocks and designed to interoperate seamlessly across entities, this AI-native data enables agents to act autonomously and responsibly, powering AI ecosystems built on trust and continuous learning.
Change management
What differentiates organizations that successfully scale pilots in a transformation from those whose efforts stall is the strength of their change management programs.31 AI transformations are no different. Time and again, interviewees told us the major barrier to adoption was resistance to change. “People resist change, believing their current processes are the best. Unless they are educated that AI will make their jobs easier, adoption will be slow. AI should be framed as a way to reduce hours, improve efficiency, and enhance work–life balance, not a threat,” said the cofounder of a GCC consultancy.
Change management strategies are key to tackling such resistance. In the survey, all but one respondent from the value realizers said their organizations had clearly defined adoption and scaling strategies backed by change management initiatives. That compares with just 41 percent of others.
Part of such a program will likely include building AI literacy widely across the organization, securing strong executive sponsorship, and publicly celebrating AI achievements. All can make a difference. But several interviewees mentioned the importance of driving both adoption and value. Interviewees highlighted three ways to maximize value at the same time as scale:
- Accountability. Too often, pilots are launched without clear links to business outcomes and without monitoring the level of adoption. Performance checks counter this by tying deployment to measurable KPIs such as cost savings, revenue uplift, or cycle-time reduction. “The premise of everything we do is that every use case must have a clear definition of success,” said the chief AI officer of a luxury goods company. “If it’s successful, it moves to the business team for ownership. If not, we stop and move on.” Another interviewee told us that any use case proposed at his organization required the AI committee approval before being built, and the foreseen financial benefits had to be included in profit and loss forecasts. Monitoring value in this way focuses minds on capturing it and can encourage further adoption when the value becomes apparent to all.
- Using AI to build AI. Sequencing the introduction of AI domains and use cases to capture quick wins is another way of demonstrating AI’s value and encouraging adoption. “Start small, ship fast, show value, and the business will scale it for you. Don’t overcomplicate; use AI to build AI,” is the advice of one interviewee. While AI can accelerate processes, using AI to build that AI can quicken delivery of an AI solution because using AI can generate the prompts and scaffold code needed for new use cases and even design workflows. As one executive explained, “For new use cases, more than 90 percent of the heavy lifting—from idea engineering to writing software, creating agents, and workflows—is done by AI.” Using AI to build AI also requires far smaller development teams, which means scarce talent can get more done, delivering greater value.
- Workflow redesign. Organizations will likely leave value on the table unless they redesign workflows. This is because AI changes how tasks are performed, by whom, and what might be possible, explaining why AI’s value often comes not from bolting it onto existing tasks and ways of working but rethinking these processes. Indeed, workflow rewiring has the strongest link to earnings improvement through AI.32 The link is likely to be stronger still in an era of AI agents. Interviewees suggest this kind of redesign is still at an early stage in GCC organizations, however. Most organizations are not yet ready to fully disrupt existing ways of working.
GCC organizations are keenly aware that they need to respond to AI’s potential, but the response isn’t easy to get right. Some appear unsure how best to move forward: “Boards and executives are excited about AI, but many still don’t know how to convert intent into action. What they need is a blueprint on where to invest and how to prioritize,” said one interviewee. Others find that AI initiatives fail to live up to expectations and struggle to integrate them into existing processes across the organization. Such challenges contribute to the widening gap in AI adoption and impact between leading organizations and the rest.33
Senior executives determined to adopt AI at scale will face a bewildering list of to-dos. Just a handful of high-level guidelines could help ensure their efforts are tightly orchestrated for impact:
- Set direction at the top. Boards and senior executives must own the AI strategy, link it to business priorities, and drive usage with clear communication and visible sponsorship.
- Build the right tech foundations. Invest in modular, scalable tech stacks; balance vendor and open-source tools; and treat data as a well-governed, reusable enterprise product.
- Reorganize how work gets done. Deploy scarce talent centrally in cross-functional squads and double down on agile ways of working.
- Link adoption to performance. Countering cultural resistance will require a strong change management program. But performance measurements also need to be in place to ensure that use cases create value. Workflows may also need to be redesigned to capture AI’s potential.
- Start smart, scale fast. Sequence quick wins to prove AI’s value and create bottom-up demand for AI at scale. Using AI to build AI speeds delivery and hence scale and value creation.
AI is transforming how organizations compete. Our survey suggests that many GCC organizations may need to pick up the pace to stay ahead, backing up experimentation and pilots with a plan that first builds the capabilities required to support AI at scale and then drives adoption with value creation as the constant goal.
