e& UAE has partnered with Intel to set up an AI Centre of Excellence and to co-develop edge AI solutions for the UAE. The COE will focus on testing and tuning LLMs on Intel Gaudi 3, then moving working models into production across e&’s network and enterprise customers. Edge deployments will lean on Intel’s Core Ultra, Xeon and GPU stacks for real-time use cases. The goal is simple: make AI services faster to build, cheaper to run, and easier to scale across the country.
What the AI Centre of Excellence actually does
The COE is not a showroom. It’s a workflow engine for getting models live.
- Test and optimise LLMs on Intel Gaudi 3
- Standardise the path from experiment to deployment
- Improve performance and cost of AI services across e& UAE
- Run evaluations that map to real network and enterprise workloads
The centre gives e& UAE a controlled environment to benchmark and harden models before they touch live traffic. Using Gaudi 3 for inference lets teams iterate quickly while keeping an eye on total cost of ownership. The outcome should be fewer “nice demos” and more production AI that survives scale and UAE compliance.
Edge AI: bringing the smarts closer to users
Most UAE use cases need low latency and privacy. That’s edge AI.
- Hardware mix: Intel Core Ultra, Xeon and Intel GPUs
- Real-time inference near the data source
- Industry-specific deployments for enterprise customers
- Designed for secure, scalable operations at the network edge
By placing compute at the edge, e& can serve use cases like video analytics, retail optimisation, and smart city functions without shuttling everything back to a central cloud. Intel’s portfolio covers client, server and accelerator tiers, which helps match the workload to the right silicon and budget.
Why this matters for telco and the UAE market
This is about turning a telecom network into an AI platform.
- AI-native network operations and services
- Faster time-to-value for enterprise AI projects
- A foundation for sector rollouts across government, utilities, finance and more
- Support for UAE data, security and sovereignty needs
e& UAE’s network is already a national backbone. With a COE plus edge builds, the operator can host, monitor and continuously optimise AI services where they run—on cell sites, metro edges or customer premises—while keeping costs predictable. That’s key when pilots need to graduate into long-lived services.
What each company brings
Clear split of roles—and less vendor lock-in talk, more delivery.
- e& UAE: network footprint, ops, enterprise reach
- Intel: Gaudi 3 accelerators, Xeon/Core Ultra/GPUs, edge software stack
- Joint: model lifecycle, deployment playbooks, industry solution patterns
e& contributes live networks and customers; Intel supplies the silicon and software to run modern AI efficiently. The partnership framework is set up to help ISVs build vertical solutions on top, which should broaden the UAE ecosystem without locking buyers to one rigid platform.
What to watch next
Early signals that this is working will be practical.
- Pilots moving to paid production within quarters, not years
- LLM inference benchmarks from the COE on Gaudi 3
- Edge rollouts in sectors that need real-time decisions
- Clear cost/performance metrics vs. “GPU only” stacks
If the COE publishes repeatable test methods and cost-per-inference numbers, buyers can compare apples to apples. Expect first proofs to appear where latency, privacy and bandwidth costs are most sensitive—think transport, energy, retail, and public services.
What is the AI Centre of Excellence?
A dedicated facility run by e& UAE with Intel to test, optimise and standardise AI models—especially LLMs—before deployment across the operator’s network and enterprise clients.
Which hardware will be used?
The COE will use Intel Gaudi 3 accelerators for AI inference, while edge solutions will be powered by Intel Core Ultra, Intel Xeon and Intel GPUs.
Why run AI at the edge?
To cut latency, ease bandwidth, and improve data control for UAE customers. Many real-world tasks are faster and cheaper when inference happens close to where data is created.
