GBM establishes UAE AI lab with Cisco, NVIDIA

Gulf Business Machines has established what it describes as the UAE’s first AI Lab built on the Cisco Secure AI Factory with NVIDIA, giving organisations across the Gulf a controlled environment for moving artificial intelligence projects from trials into production.

The Dubai-based technology company said on Wednesday that the facility combines pre-validated computing, networking, security and AI software to support on-premises and hybrid deployments. GBM said the approach could shorten deployment schedules from months to weeks while helping customers meet requirements for data residency, regulatory compliance, cybersecurity and low-latency processing.

The lab is aimed at organisations seeking to test production-grade AI systems without first assembling separate infrastructure components and security controls. It is also intended to address sovereign AI requirements, where organisations need greater control over where data and workloads are stored, processed and governed.

At the core of the facility is Cisco Secure AI Factory with NVIDIA, a modular reference architecture combining high-performance AI infrastructure with security and observability across the technology stack. Cisco says the architecture is designed to protect models, agents, applications, workloads and infrastructure rather than adding security only after AI systems have been deployed.

The GBM facility includes NVIDIA AI Enterprise software and provides continuous monitoring of AI workloads and agent activity. Security policies can be enforced through multiple layers, including hardware-level controls, while the architecture covers data paths as well as models and workloads.

Pieter Bil, GBM’s chief executive, said enterprises moving beyond AI experimentation face a choice between deploying quickly and maintaining strict governance. “Building production-grade AI requires a delicate balance between rapid innovation and strict data governance,” he said, adding that GBM had combined global technology frameworks with locally managed platforms designed for Gulf requirements.

Bil took over as GBM chief executive in September, with responsibility for the company’s regional growth and a mandate that includes accelerating AI adoption among enterprise and public-sector customers. He previously served as Kyndryl’s senior vice-president and managing director for the Middle East and Africa.

Cisco introduced the Secure AI Factory with NVIDIA in 2025 and has since expanded the architecture from central data centres towards edge deployments. Its current design supports Cisco UCS servers using NVIDIA accelerated computing, high-performance Ethernet networking and integrated security technologies, alongside monitoring and management capabilities.

Cisco expanded the platform in March this year to support AI inference closer to where data is generated, and in August added rack-scale computing options intended for larger AI workloads. The architecture now ranges from individual GPU servers to high-density systems, allowing customers to use a common operational and security framework as deployments grow.

For organisations evaluating the lab, the proposition is a test environment that mirrors the architecture intended for production, reducing the need to redesign security and infrastructure after a successful pilot. That can be especially relevant to regulated enterprises weighing cloud flexibility against requirements to retain control.

Cassie Roach, Cisco’s global vice-president for Cloud and AI Infrastructure Partner Sales, said the GBM collaboration was intended to help Gulf organisations turn AI plans into commercial deployments while retaining sovereign control and operational resilience.

The emphasis on sovereignty reflects a practical constraint for sectors handling sensitive or regulated information. On-premises and hybrid systems can allow organisations to keep specified workloads and data within required geographic or organisational boundaries, although compliance ultimately depends on how individual systems are configured, governed and operated.

Dirk Barfuss, NVIDIA’s director for EMEA Channel, said enterprise AI infrastructure was shifting towards inference-optimised and agentic systems. He said NVIDIA accelerated computing and AI Enterprise software provided the performance and scale required for those workloads, while GBM’s lab would demonstrate how such infrastructure could be deployed securely across the GCC.