FEATURE
The ability to change infrastructure arrangements as business requirements evolve is becoming particularly valuable. Flexible consumption models allow enterprises to experiment with new applications and subsequently reassess where those workloads should operate once they reach production.
This is especially relevant to AI projects, where initial experimentation may require different infrastructure capabilities from large-scale deployment.
For Migally, resilience is another major driver of hybrid and multicloud adoption.
“ Hybrid and multi-cloud strategies are increasingly being driven by three priorities: resilience, sovereignty and AI readiness,” he said.
Organisations want to reduce dependence on individual platforms, meet regulatory obligations and maintain business continuity regardless of where applications are hosted.
However, operating across multiple environments introduces management challenges. Without consistent governance, security and visibility, hybrid architectures can create operational blind spots and increase complexity.
Abousaleh pointed to the continuing importance of legacy investments, particularly in sectors such as banking and insurance, where replacing highly customised systems can be expensive and disruptive.
We’ re seeing organisations across the Middle East move away from a cloud-first mindset towards a business-outcomefirst approach. The question is no longer where workloads should run, but whether organisations can trust, govern and protect the data that powers those workloads.
For example, Everpure customer Options Technology has deployed the company’ s technology to support AI workloads through flexible consumption and on-demand scalability, helping the business overcome infrastructure constraints associated with global expansion.
Migally believes the AI infrastructure conversation must extend beyond computing power and storage capacity.
“ AI is fundamentally changing the conversation because infrastructure is no longer just about compute and storage. It is about data quality, governance and trust,” he said.
The challenge is particularly significant for organisations that lack a clear understanding of what information they hold, where it resides and whether it is suitable for AI applications.
Performance considerations are also influencing infrastructure choices. Manufacturing, logistics and smart city applications often require rapid local processing, making edge computing and on-premises infrastructure important components of enterprise architectures.
Regional volatility has further reinforced the importance of business continuity, encouraging organisations to strengthen backup arrangements, availability strategies and infrastructure resilience.
AI rewrites the infrastructure equation
Artificial intelligence is fundamentally changing enterprise infrastructure requirements, introducing new pressures on compute, storage, networking and data management.
Unlike many traditional enterprise applications, AI workloads can generate unpredictable demand, consume significant processing resources and require access to information distributed across multiple systems.
For CIOs, infrastructure planning must increasingly account for the relationship between AI performance, data quality and operational cost.
Akar said AI is challenging established assumptions about capacity planning and data storage.
Enterprise information has traditionally been organised around individual applications, but AI systems frequently need to draw on information from across the organisation. This means existing data silos, inconsistent definitions and governance gaps can directly affect the effectiveness of AI initiatives.
Data primacy, he argued, provides a framework for making information accessible and governed across different environments.
“ As organisations across the Middle East accelerate their AI ambitions, the conversation is shifting from infrastructure alone to trust. AI can only be as trustworthy as the data behind it. The organisations that will lead in the AI era are those that combine innovation with strong data governance, resilience and visibility. In other words, they are building not just AI capabilities, but Data and AI Trust,” said Migally.
Datadog’ s research highlights another dimension of the challenge: the growing complexity of AI workloads.
Its 2026 State of AI Engineering report found that, globally, tokens per request more than doubled for median customers and quadrupled for the heaviest users year on year.
Longer prompts and increasingly complex agentic workflows are placing additional pressure on infrastructure capacity, application response times and operating costs.
The research also found that one in 20 analysed large language model calls reported errors, with 60 % of those errors attributed to capacity limitations.
Abousaleh said these findings demonstrate why enterprises must look beyond simply
26
INTELLIGENT CIO MIDDLE EAST www. intelligentcio. com