Emirates Global Aluminium unlocks public and private cloud, industrial AI with Azure Local

Paving a path for sustainable, scalable computing infrastructure

Innovating aluminium to make modern life possible is all in a day’s work at Emirates Global Aluminium (EGA). Based in the United Arab Emirates (UAE), EGA has rapidly grown from a small regional smelter into a vertically integrated aluminium giant with mining, refining, smelting, and recycling facilities. Serving more than 400 customers spanning from aerospace and automotive to construction and consumer goods companies, EGA is the world’s largest premium aluminium producer and responsible for one in every 25 tons of aluminium made globally.

EGA has long used and evaluated technology including AI—increasingly so as part of its Industry 4.0 transformation strategy—to drive innovation and increase its competitiveness, agility, and flexibility. Although it uses hybrid cloud, which includes a mix of on-premises private cloud infrastructure that offers close proximity to company operations, the company’s leadership team believes public cloud is instrumental in democratizing innovation not just for EGA but for everyone. “Not so long ago, startups would have to spend $40,000 to $60,000 buying servers to get up and running, and then they’d need resources that knew how to run that infrastructure and scale it,” explains Carlo Khalil Nizam, Chief Digital Officer at Emirates Global Aluminium. “Cloud takes all the hardware and makes it available around the world for lease together with new software capabilities. That took the capital obstacles away from startups all the way up to big businesses like EGA and meant a lot more people could experiment and innovate.”

EGA decided to deploy a hybrid environment managed by Microsoft Azure Arc that connected private cloud services through on-premises datacenters, which are used to host operational data, quality control data, environmental and energy data, and supply chain and market data, with the public cloud. This was done to optimize latency, support advanced AI and automation solutions, offer sustaining commercial savings by applying intelligence at the edge, and streamline processing for massive amounts of real-time readings from sensors, machinery, and production lines. EGA specifically needed to manage increasingly complex and data-intensive operations with scalable computing infrastructure, which traditional on-premises systems struggled to deliver.

Once it came time to select a cloud vendor, EGA knew it had to keep core services available locally in the UAE to comply with data sovereignty law. It had already built a modern digital workplace with Microsoft 365 and Microsoft Copilot and upgraded its infrastructure to tap into the benefits of the cloud. As it turned out, Azure was well placed in the Middle East—in fact, the closest Azure datacenter was just 1,500 feet away from EGA’s facilities.

An adaptive cloud approach to making modern aluminium production possible

To support its on-site operations and broader cloud-based solutions, including its digital manufacturing platform, EGA moved one-third of its server base to the Azure public cloud platform and another third to run hybrid and at the edge with Azure Local, bringing together the best of the public and private cloud from one provider. “That had big implications both technically and commercially, favorably in both ways,” says Nizam. “Using the Azure public cloud platform gave us elastic power instantly, and at the press of a button, we had software-defined networking on demand. We could access as much compute and storage as we wanted very quickly, which allowed us to innovate much faster and be more agile.”

EGA now uses Azure Local to support safety-critical applications that use AI—to detect rod latch failures in real time in the rodding room, for example—and require very low latency and fast response times, which the company gets most reliably at the edge. “Running at the edge with Azure Local brings us not just infrastructure as a service capabilities, but also a lot of software and platform capabilities that open up new possibilities, such as Azure AI and Azure IoT, which we didn’t have in our on-premises environment,” says Nizam. “If we wanted to have them on-premises, it would have taken a much longer time to figure that out.”

Employees simply and centrally monitor both sides of the environment with Azure Arc, which EGA also uses to accelerate the deployment of use cases and shift models from training into operation. “With Azure Arc, we can deploy use cases fully automated from Azure to our premises, while also respecting our critical infrastructure security architecture,” says Alexander Simon Kraehenbuehl, Director of Digital Platforms and Infrastructure at Emirates Global Aluminium. “Teams on our digital manufacturing platform deploy multiple releases per day if needed, compared to the maximum weekly pace in our traditional environment. This allows them to be more agile and stay closely aligned with the fast pace of our business. Our new hybrid cloud, powered by Azure Arc, allows us to implement safety-critical use cases and thus unlock business value that before was out of reach for our digital ambition.”

Adds Nizam, “The teams that use Azure Arc love it because it’s a single pane of glass that allows you to use assets and run models without it mattering if they’re here or there.” The seamless connections between Azure products also help EGA avoid reengineering solutions every time it decides to move a model or application between the public and private cloud.

Accelerating industrial AI use cases to improve production processes

After bringing the power of the cloud into its operation areas, EGA experienced 10 to 13 times faster AI response time, lower latency, and 86% cost savings associated with AI image and video use cases. As a result—and as anticipated—EGA can implement new industrial AI use cases, including using narrow AI and visual AI to inspect carbon anodes that are essential to its production processes. “These very big carbon blocks have different surface patterns, and the human inspectors would only have time to look at 1.5 to 2.5% of all the anodes produced,” explains Nizam. “That meant they could reject some, but it also meant that some of variable quality inevitably made it through to the production process and could have affected the quality of the aluminium.”

Using visual AI and a neural network, EGA developed and trained a model on 100,000 images to define and differentiate between what makes a good anode and what makes a bad one. “Now, we don’t inspect 1.5 to 2.5%—we inspect 100% of all anodes to one common objective standard,” continues Nizam. “That has boosted the quality of our anodes, reduced rejections, and overall helped improve the quality of our aluminium.”

This use case speaks to a larger attitude across EGA of embracing technology and innovation as new opportunities arise. “You can use AI as foe or friend, and we want to see how we can use it as friend to augment our productivity, develop new types of alloys, and make modern life possible through innovation,” notes Nizam.

Tackling sustainability from every angle

As part of its broader pursuit to set global best practices for the aluminium industry, EGA views sustainability through two lenses. There’s the traditional environmental perspective, for which Nizam explains, “Consolidating one-third of our servers into Azure reduced our reoccurring total cost of ownership by 40%, compared to our on-premises setup, and was equivalent to 1.3 million date trees in annual metric tons of CO2 avoidance.” But EGA is also dedicated to keeping computing operations sustainable commercially. “All our compute inside EGA runs on 100% renewable power from our Azure Local portfolio, which means that the AI use cases that run at the edge are green AI,” says Nizam. “It’s a much more sustainable way of computing and also a much more flexible, safe, and secure way of computing.”

A similar attitude applies across the company’s data strategy. At the outset of EGA’s digital transformation, its metal value chain had created a digital footprint that was relatively untapped. EGA put in place a foundational data lakehouse based on Azure Databricks and delicately configured the lakehouse for its unique architecture, getting support from Microsoft Global Black Belts to speed up project execution. “Having that world-class data platform with Azure Databricks was so powerful for us and critical to our ability to ingest, aggregate, organize, and assimilate the data as a first step,” says Nizam. “That data platform is now the backbone of many other platforms that exist in Azure, including analytics, AI, and native IoT, which are all available in hybrid format.” 

“The icing on the cake,” as Nizam describes it, is that EGA also took the opportunity to reduce the number of on-premises datacenters from 19 to 10, designating two as its primaries. “Those datacenters use 50% less power than the original datacenters, and the power they do use is 100% renewable,” adds Nizam.

Feeling and seeing the difference

These use cases only scratch the surface of what EGA is now capable of—not to mention what it’s exploring for the future. “Going forward, digital transformation will touch all walks of life in our company, including the employee experience, the customer experience, and how we run our physical and corporate operations,” says Nizam. “We’re using green AI and virtual computing to predict commodity pricing and help us make decisions there, to help us look at how to use autonomous vehicles on-site to optimize our planning and production processes, and to build our own language models.”

These new opportunities give EGA added incentive to upskill employees who traditionally kept on-premises infrastructure up and running. They’re now learning how to use and maximize the new services and functionality of the enhanced digital manufacturing platform. “It’s been poetry in motion to watch these advanced edge capabilities come to life, delivered perfectly and on time,” notes Nizam. “You feel the difference, and you see the difference.”

Discover more about Emirates Global Aluminium on Facebook, Instagram, LinkedIn, X/Twitter, and YouTube.