E-Paper
Sign In
UAE firms face rising AI costs as adoption moves beyond pilot stage
As Gulf businesses accelerate AI deployment, expert warns that hidden consumption costs, governance gaps and infrastructure inefficiencies could undermine returns on investment
- PUBLISHED: Wed 5 Aug 2026, 2:51 PM
As governments and businesses across the UAE and the wider Gulf race to embed artificial intelligence into their operations, the challenge is shifting from adopting the technology to controlling its cost and proving its business value.
While the region has positioned itself as a global AI hub through sovereign investments, national strategies and large-scale digital transformation projects, many organisations are discovering that deploying AI at scale comes with hidden costs that are often overlooked during early experimentation. According to industry experts, managing those costs could become one of the biggest determinants of successful AI rollouts over the next few years.
Recommended For You
“AI adoption is no longer a technology initiative, it is becoming a business imperative,” Sherif Tawfik, Chief Business Officer at Core42, told Khaleej Times. But he warned that organisations must focus on whether AI can deliver “measurable value repeatedly, securely and economically” rather than simply implementing the technology. Many companies, he said, are embracing AI “without clear visibility into where it is being used, what it is costing, or whether it is delivering meaningful business outcomes.”
Stay up to date with the latest news. Follow KT on WhatsApp Channels.
A growing concern for businesses is that AI economics differ significantly from traditional software deployment. Unlike conventional applications, every AI interaction carries a cost, which can quickly escalate as organisations move from limited pilot projects to enterprise-wide adoption.
“The biggest hidden cost is not the model itself, it is the unmanaged consumption,” Tawfik said. “Unlike traditional software, every interaction carries a cost, and organisations often underestimate how quickly that grows as usage expands from a pilot to thousands of users or autonomous agents executing multi-step tasks.”
He said many enterprises struggle because financial controls, usage monitoring and governance frameworks are often introduced only after systems have already been deployed. “The real challenge is that consumption compounds before the controls to manage it are in place,” he said, adding that AI spending can accumulate without being clearly linked to business outcomes if organisations fail to establish proper oversight from the outset.
Infrastructure choices are also emerging as a critical factor in controlling costs. As AI applications become more diverse, experts say organisations must avoid routing every task through the largest and most expensive AI models.
“Running both on the same infrastructure, with the largest available model by default, is a reliable way to waste capacity and inflate the cost per token,” Tawfik said. Instead, businesses should match workloads with the most appropriate models and computing resources to improve utilisation and reduce unnecessary expenditure.
Against this backdrop, Core42 has launched Compass, a full-stack AI inference platform aimed at helping enterprises move from experimentation to large-scale deployment. The platform provides access to more than 60 AI models and multiple accelerator technologies, while offering tools to monitor usage, manage budgets and enforce governance policies.
According to the company, Compass provides centralised visibility into AI expenditure, real-time budgets and alerts, hierarchical user controls and security safeguards including in-country data residency and encryption. The platform is designed to help organisations maintain tighter control over AI spending and compliance as usage expands.
The increasing focus on governance reflects a broader trend in the AI market as technology leaders and finance chiefs become more closely aligned on investment decisions. Tawfik said CIOs and CFOs now need a shared framework that evaluates AI according to business outcomes rather than model costs alone. “Leaders should evaluate AI based on the cost, speed and quality of achieving a business outcome, rather than the price of an individual model interaction,” he said.
Industries expected to benefit most from more efficient AI deployment include financial services, healthcare, government, retail, travel, telecoms and energy, particularly where organisations handle high volumes of customer interactions and data-intensive processes. Efficient inference is becoming essential as AI moves from isolated pilots into daily operations.
Looking ahead, the outlook for the Gulf AI sector remains strong. IDC expects global AI infrastructure spending to surpass $1 trillion by 2029, supported by a compound annual growth rate of around 31 per cent from 2025. But industry leaders believe the next phase of growth will be defined less by model capabilities and more by how efficiently, securely and economically organisations can deploy AI at scale. As agentic AI systems drive a surge in inference demand, companies that can balance innovation with robust governance and cost management are likely to emerge as the biggest winners.
Associate Editor Somshankar Bandyopadhyay brings more than three decades of experience to the KT newsroom. Currently, he manages the news production of the business section, ensuring that the top economic and business news of the day reaches its readers.