The rise of compute-intensive AI workloads has transformed the data center landscape for good.

Whereas capacity has historically been able to scale with demand, this is no longer the case. Now, ambitious projects like Stargate are necessary to meet the ambitious targets of hyperscalers and model makers. To address the shortfall in energy supply, firms are exploring a range of solutions, from on-site generation and small modular nuclear reactors to increased use of battery energy storage solutions, all working in conjunction with the grid.

But alongside the ambitious build-outs and investments, an operational revolution is necessary to meet the requirements of the AI age. At its heart is the convergence of information technology (IT) and operational technology (OT) systems. Seamless coordination between these elements, whether that’s telemetry and analytics or energy and maintenance, enables unparalleled operational efficiency.

The potential for IT/OT convergence cannot be overstated. Once connected, AIOps solutions can unlock new levels of data center intelligence, efficiency, resilience, and – crucially – scale.

The new landscape

Historically, IT and OT systems existed in siloes. Different teams were responsible for them, acting independently of one another. Separating the two made sense, for a time. IT provided the digital backbone to the data center, while OT provided its power, cooling, and environmental controls. Beyond capacity planning, the two could be managed as distinct clusters of systems.

Today, data centers are so much more than warehouses filled with servers. They are shaped by sophisticated, interrelated systems that straddle the digital and the physical.

AI is a driving force behind that change. Not only do AI workloads place greater demands on compute and network infrastructure, but they also require dynamic responses from cooling and energy systems. This can only be coordinated by breaking down the siloes between IT and OT, unifying data and controls across the two domains.

This kind of IT/OT convergence has the potential to improve resource utilization, improve response times, and drive a range of predictive and even autonomous processes – from workload distribution to maintenance.

When the rubber hits the road

Achieving IT/OT convergence is easier said than done. Data center operations could have any number of legacy systems, outdated policies, and siloed organizational structures, all of which put up barriers between systems.

Organizationally, there are cultural and strategic differences between IT and OT teams. It’s easy to fall into stereotypes here: the IT professional who likes to “move fast and break things” and the more deliberate OT worker who prioritizes resilience. This is, of course, an oversimplification. But there’s a grain of truth to it. These teams will have different ways of working and often work in different spaces. Bringing their ways of working closer together is part of the challenge of convergence, shaping cross-functional teams with shared goals and pooled expertise.

There are also technical barriers to change. In a very real sense, IT and OT systems are speaking different languages. Many OT systems rely on proprietary industrial protocols and physical hardware, while IT networks use protocols such as TCP/IP. Successful convergence requires these various protocols and systems to be translated and bridged by a variety of means. Integration is a leading solution. This can be made easier by working with providers that offer solutions across IT and OT – that way, they can help you on your convergence journey.

The four phases of IT/OT convergence

Phase I: Data connectivity

The foundation for any transformation is unified data. Connecting OT systems to a data platform and standardizing data formats delivers a real-time flow of data across operations. It’s best to start with non-critical systems before moving on to mission-critical infrastructure.

Phase II: Unified dashboards

Building on this data connectivity, operators can adopt visualization tools that bring that data to life for both IT and OT teams. That level of visibility spells an end to information siloes and the potential to unlock new synergies and operational efficiencies.

Phase III: Automated responses

The next, and potentially most impactful, phase is deploying automated solutions. A steady stream of data from across IT and OT can drive AIOps functionality capable of adjusting cooling and power management in real time.

Phase IV: Predictive operations

Predictive operations take automated systems a step further. Machine learning models draw on historical and real-time data to predict outcomes and respond accordingly. For example, predictive maintenance draws from a range of data points to determine when a machine needs upkeep – before anything goes wrong. More efficiency, less unplanned downtime.

The coming revolution

For many data center operators, this change is already well underway. As the scale of demand for AI-ready compute continues to rise, IT/OT convergence will progressively shift from a point of differentiation to a commercial imperative. Connected data streams and cross-functional teams will be the norm, as will predictive analytics and intelligent automation.

Looking forward – beyond the four phases set out in this article – the next revolution will be in fully autonomous data centers.

This will leverage the functionality unlocked by IT/OT convergence to its fullest extent, dramatically reducing the need for human intervention and delivering never-before-seen efficiency and scale.