AI has been the catalyst for a huge shift in the scale of digital infrastructure. The era of advanced compute is characterized by greater power and capacity, alongside bigger, more complex, interconnected components that now make up extensive AI factories – far surpassing the data halls of just the last decade.

These hubs of connectivity have been firmly cemented as the foundations of the digital economy, and this high-stakes reality – combined with the growing complexity of today’s facilities and operations – is driving the need to look beyond the data center industry itself for next-gen solutions and insights.

Increasingly, operators are recognizing that the answers to some of their most pressing challenges may already exist in more mature, adjacent sectors.

With a background in astrophysics and aerospace engineering, Arti Garg, data scientist and chief technologist at Aveva – a key strategic partner of Schneider Electric – explores the value of applying industrial lessons to AI infrastructure, and why digital twins are becoming mission critical for operating data centers at scale, in a recent DCD>Talks episode.

AI meets industrial

“Today’s powerful AI gigafactories are changing how AI intersects with the industrial sector,” says Garg. “That’s because models are becoming more capable and advanced, which means we can bring in more diverse data types.”

Garg is confident that the industrial sector is rich with opportunities for AI applications. From monitoring to optimizing operations, AI-powered industrial software presents a natural partnership for delivering the next generation of digital transformation and smarter, more efficient manufacturing.

In many ways, the industrial world – companies manufacturing goods, machinery, and resources that are used for producing other goods – has already laid the groundwork for the kind of data-driven optimization that AI now enables at scale. But with this wealth of opportunity comes an equal number of challenges.

The infrastructure required to support today’s AI factories and enable this valuable intersection is increasingly complex, spanning not just IT systems but also physical infrastructure layers that must operate in harmony.

“Everything from electrics, power, and cooling to the systems that support the computing infrastructure itself – the requirements of these AI factories are starting to resemble those of other industrial facilities,” explains Garg. “That means data center customers are now looking for the types of solutions we’ve long provided to industrial customers.”

Lessons learned from the industrial sector are unlocking new possibilities, as well as critical capabilities around ensuring uptime, delivering backup power, managing fluctuating loads, and prioritizing safety. These are areas where industrial operators have decades of experience operating under strict constraints and high-risk conditions.

“We’re actively bringing all of these capabilities into the data center world,” adds Garg. This transfer of knowledge is becoming essential as data centers evolve into increasingly complex, mission-critical environments.

Been there, done that

Managing extreme volumes of power and heat has long existed beyond the data center walls, offering a valuable history of trial and error to draw from. Industries such as oil and gas, chemicals, and heavy manufacturing have spent decades refining approaches to thermal management, efficiency, and operational resilience.

“Managing thermal loads and optimizing cooling equipment in a way that minimizes water usage safely is a complex challenge we’ve been working on with industrial customers for a very long time,” says Garg.

The shift from air- to liquid-cooled and hybrid environments is a key example. To manage today’s heat outputs, liquid cooling is quickly becoming a necessity. While this innovation improves efficiency and supports higher-density workloads, introducing coolants and managing flow adds a new layer of operational complexity that must be carefully controlled.

Optimizing these systems requires simulation and iteration – approaches that have been refined in industries like chemicals for decades. Rather than starting from scratch, data center operators can leverage these established methodologies to accelerate deployment and reduce risk. Garg offers another example of this valuable crossover:

“In the industrial space, our software spans the entire asset lifecycle – whether that’s an individual asset or a whole facility. Historically, it’s been used to design complex systems, like those in an oil refinery. We’ve supported energy performance certificates (EPCs) in building these systems, enabled operations from an HMI/SCADA standpoint, and optimized performance using time-series and other operational data.”

This lifecycle approach – design, build, operate, optimize – is increasingly relevant for data centers as they scale in both size and complexity. It enables a more integrated, data-driven way of managing infrastructure from inception through to ongoing management.

Garg explains that lessons learned in reducing environmental impact can also be applied to data centers directly:

“It comes back to optimizing resources. Cooling is a hugely important part of this, and matching workloads to resource availability is key. For example, you might ramp up compute usage at times when electricity is cheaper.”

This kind of dynamic enhancement highlights how operational intelligence can drive both cost efficiency and sustainability outcomes.

Garg believes a key missing piece lies in establishing shared standards to measure and understand environmental impact. Without consistent benchmarks, it becomes difficult to compare performance or drive meaningful progress across the industry.

“It’s important that we come to a common definition,” she explains. “What we do at Aveva is help people measure, track, and contextualize data – so that as industry-wide standards emerge, we’re already positioned to support reporting against them.”

Eyes everywhere

In environments such as oil refineries, software is highly specialized across each stage of the lifecycle. This is where digital twins – and the digital thread connecting lifecycle data – become vital components of modern operations.

“To service something, you need to know how it was designed, how it was built, and its service history,” says Garg. “Having all three pieces of information becomes essential for managing rising complexity.”

Across industries, this level of visibility is helping optimize resources and reduce inefficiencies. As data centers continue to scale globally in number and size, the pressure to operate sustainably and responsibly is increasing at the same pace.

Improving efficiency across the entire ecosystem from design and planning through to operations requires a digital edge that is proven, scalable, and ready to deploy. Garg highlights why this holistic view is no longer optional:

“We’re constantly hearing eye-popping investment figures from AI companies building data center capacity. And the real question is: what does it take to deliver these massive projects? We know how to do that in other industries – but doing it at AI speed and scale requires the ability to execute quickly and repeatedly.”

Digital twins play a key role here. In partnership with NVIDIA, Aveva has introduced sim-ready asset documents that support modular, standardized, yet flexible and repeatable design and deployment. This approach enables organizations to replicate successful designs, reduce errors, and accelerate time to market.

This speed and optimization extend into ongoing operations. Data centers are becoming increasingly autonomous, shifting from reactive to proactive management – and digital twins make this transition more reliable and scalable.

“Firstly, it helps determine when and where to send a human operator across multi-acre facilities,” explains Garg. “Over time, as autonomous systems and robots take on servicing roles, the digital twin becomes the context in which they operate.”

With multiple disparate data streams flowing across the data center – from IT systems to mechanical and electrical infrastructure – information must be unified, centralized, and normalized to support effective decision-making.

The sky’s the limit

Historically, the industrial sector has been slower to adopt new technologies due to safety, liability, and security concerns, as well as practical challenges like remote locations and limited connectivity. These constraints have often made experimentation more difficult and increased the stakes of failure. However, Garg believes AI is accelerating adoption in new and unexpected ways:

“Having worked in industrial AI, I was surprised by how quickly generative AI was adopted. In the past, people didn’t trust even well-established machine learning models. Now, customers are actively asking about generative AI even though it’s less explainable. I believe it’s because it’s more intuitive.”

This shift toward more intuitive, user-friendly technologies is changing how innovation is embraced across industries. It lowers the barrier to entry and enables a wider range of users to interact with complex systems.

“In environments where people aren’t just sitting at a computer – where they’re wearing gloves or operating machinery – we’re moving beyond screen-based interaction,” continues Garg. “Voice interfaces, for example, could allow someone to request a repair plan for a piece of equipment in real time, using a phone to facilitate the interaction.”

These developments point to a broader transformation in how people interact with technology in industrial and operational settings. The move toward more natural interfaces will further accelerate adoption and unlock new use cases.

Ultimately, it’s a process of give and take. As AI continues to evolve, adoption will increase, and as adoption grows, AI itself will continue to improve. This feedback loop is already driving rapid change across sectors.

For the data center industry to scale by solving tomorrow’s challenges before they arise, the boundaries between sectors must continue to blur. This cross-sector exchange creates a powerful, circular dynamic where advancements in one domain fuel progress in another.

To learn more about how the industry can benefit from knowledge sharing, watch the full DCD>Talks episode with Arti Garg, here.