Gary Cudmore - Episode 172 - The Route to Networking

From Computer Rooms to Digital Cities

Gary Cudmore on scaling data centres, power constraints, and why infrastructure is still all about people

In this episode of the Route to Networking podcast, host Jack Rafferty speaks with Gary Cudmore, Senior Vice President at Rowan Digital Infrastructure.

With over four decades in the industry, Gary has seen the full evolution of digital infrastructure. From early mainframe computer rooms in the 1980s to today’s hyperscale, AI-driven data centres.

His experience spans engineering, construction, and large-scale infrastructure delivery, giving him a unique perspective on how the industry has changed, and where it is heading next.

What follows is a conversation about scale, power, people, and the realities of building the infrastructure behind the digital world.

When Data Centres Were Just “Computer Rooms”

Gary entered the industry in 1983, at a time when data centres looked nothing like they do today.

Back then, they were simple environments. Mainframes, raised floors, perimeter cooling, and relatively small footprints, often inside high-rise office buildings.

But while the technology was simpler, the challenges were not.

Getting equipment into buildings without suitable access meant cutting machinery in half, craning systems through windows, and reinforcing floors just to handle the load.

“It wasn’t the data centre that was complicated,” Gary explains. “It was the environment we had to build it in.”

Even approvals were a challenge. Inspectors had never seen this type of infrastructure before. Every project required education, negotiation, and persistence just to move forward.

The Moment That Changed Everything

Early in his career, Gary experienced a situation that shaped how he approaches design to this day.

During a site visit, a data centre lost power. The backup generator failed to start, and the site manager had no documentation to diagnose the issue.

No diagrams. No clear system visibility. No way to act.

“I made a commitment that I never wanted to put another operator in that position,” Gary says.

From that point on, his approach changed.

Design was no longer just about building infrastructure. It was about building systems that operators could understand, manage, and recover in real-world scenarios.

Documentation, clarity, and usability became just as important as engineering itself.

The Power Problem No One Can Ignore

Fast forward to today, and the challenges look very different.

The biggest constraint is no longer space or design. It is power.

Data centre demand is moving faster than utility infrastructure can keep up. In some regions, grid connections are not expected until the 2030s.

That delay is forcing developers to rethink how projects are delivered.

One solution is temporary on-site generation, often using natural gas, to bridge the gap until permanent power is available.

But that creates a new tension.

Speed versus sustainability.

Balancing environmental goals with the need to deploy infrastructure quickly is becoming one of the defining challenges of the industry.

Building at a Scale That Didn’t Exist Before

The scale of modern data centres is difficult to comprehend.

As Gary puts it, a single gigawatt campus is equivalent to the daily power consumption of an entire city.

“We’re not just building data centres anymore. We’re building digital cities.”

This shift brings new challenges.

Long-lead equipment like transformers can take years to secure. Developers must commit capital early, often before customers are confirmed, just to stay competitive.

At the same time, AI workloads are introducing unpredictable power spikes, placing even greater strain on infrastructure design.

In many cases, operators do not yet fully understand their own load profiles.

That uncertainty makes collaboration between engineering and IT teams more important than ever.

Why Nuclear Is Back in the Conversation

As power demand grows, so does interest in alternative energy sources.

One area gaining attention is small modular reactors (SMRs), a new generation of nuclear technology designed to be safer and more flexible.

Gary believes part of the challenge is perception.

When people hear “nuclear,” they think of worst-case scenarios. But modern systems are fundamentally different, with built-in safety mechanisms and proven use cases in other industries.

More importantly, they solve a key problem.

Reliable, always-on power.

For data centres, that is critical.

And for developing regions, it could be transformative, enabling both connectivity and economic growth without reliance on traditional grid infrastructure.

Why Community Matters More Than Ever

Despite the scale and complexity, Gary repeatedly comes back to one core idea.

People.

From planners and engineers to local communities, every stage of a data centre project depends on human relationships.

At Rowan, community engagement is not an afterthought. It is built into the process.

That means public meetings, local investment, infrastructure improvements, and ongoing dialogue with residents.

“We want unanimous approval,” Gary explains. “That only happens when you listen and engage properly.”

In an industry often seen as disruptive, that approach is becoming increasingly important.

The Talent Gap and the Opportunity Ahead

As the industry scales, so does the demand for skilled workers.

But the gap is growing.

From electricians and mechanical engineers to safety professionals, there is a clear shortage of talent capable of supporting projects at this scale.

Safety, in particular, stands out.

With thousands of workers on a single site, often operating heavy equipment in close proximity, the risks are significant.

For those entering the industry, Gary sees this as a major opportunity.

Not just to build a career, but to play a critical role in shaping how these environments operate.

Looking Toward the Future

So what comes next?

According to Gary, the future of data centres will be shaped by two things.

The chip and the network.

Advances in semiconductor technology will dictate how infrastructure is designed, while AI workloads will continue to push the limits of power and cooling.

Further ahead, the combination of AI and quantum computing could unlock entirely new possibilities.

But that creates another challenge.

Innovation may outpace our ability to physically build the infrastructure needed to support it.

A Final Piece of Advice

For those starting out, Gary keeps his advice simple.

Stay curious.

Do not rely solely on quick answers. Think first. Learn deeply. Test your understanding.

“The more I learn, the less I realise I know.”

In an industry evolving this quickly, that mindset may be the most valuable skill of all.

Listen to the Full Episode

In this episode of Route to Networking, Gary Cudmore shares insights from over 40 years in the industry, covering everything from early data centre builds to the future of AI-driven infrastructure.

From power constraints and nuclear energy to community engagement and career advice, it is a conversation that highlights both the scale of the challenge and the opportunity ahead.

Listen to the full episode to hear Gary’s story.