WESCO Q&A: Craig Doyle on Data Centre Supply Chain Planning

At Data Centre LIVE, WESCO Datacentre Solutions' Craig Doyle said that the biggest challenge facing AI infrastructure is not demand itself.

Instead, he debated whether developers are planning early enough to keep pace.

Speaking to BizClik Studio, Craig argued that the AI boom has changed workable operator strategies, with supply chain decisions needing to be made long before construction begins.

Planning before construction

While long lead times are a familiar challenge for data centre operators, Craig believes many organisations still think about procurement too late.

"If you're beginning with the end goal of construction in mind, and the time it takes to build, then your supply chain conversation really should start at the design and concept phase," he said.

Instead of waiting until equipment is specified, developers should consider procurement alongside the earliest design decisions.

That is especially important as some electrical infrastructure is subject to lengthy manufacturing times.

Craig explained: "There are some critical components in the data centre space that could have more than 52 weeks' lead time.

"Some of them could be two to three years for the grid components that actually connect the data centre to the electrical grid."

Given that many data centres themselves take between nine and 18 months to construct, he says early planning is the only practical way to reduce project risk while accommodating rapid market growth.

A more resilient supply chain

Although geopolitical tensions continue to affect global trade, Craig believes the industry's experience during the COVID-19 pandemic has made supply chains significantly stronger.

According to Craig, today's conflicts tend to affect regional operations and transport costs more than global availability.

He said that companies across the sector have learned valuable lessons since the pandemic, with supply chain resilience becoming a strategic priority.

He also added that the changes implemented after COVID-19 mean current disruptions have "less of an effect globally" than they otherwise would have had.

Sovereign AI drives local infrastructure

Beyond hyperscale expansion, Craig highlighted sovereign AI as one of the biggest forces reshaping data centre investment.

As governments place greater emphasis on keeping data within national borders, demand is increasing for domestic AI infrastructure capable of supporting both large language models and inference workloads.

Sign up for next year's event!

If you missed out on Data Centre LIVE 2026, or would love to attend again, you can register your interest in LIVE: The World Tour 2027.

Craig said: "I think if you polled 100 countries, almost 80, 90% would tell you they'd rather have control over that resource and that asset. So now that trend around enabling sovereign AI is rapidly increasing."

He pointed to growing government-backed investment programmes across Europe, where countries are developing what are often described as 'AI factories' or 'gigafactories'.

However, he said that whether those projects reach gigawatt scale or only hundreds of megawatts depends on the size of each nation.

Why edge still matters

Despite much of the spotlight being on hyperscale campuses, Craig does not believe edge infrastructure is becoming less important.

He said: "These mega sites, these giga sites and edge data centres – whether they're used for sovereign AI or for any service provision – there really is a place for both styles, and there's obviously a lot in between."

He explained that while AI models may be trained in large facilities, many organisations still need inference workloads delivered much closer to where they're operating.

"We're seeing more and more now is that traditional enterprises – traditional health, healthcare, governmental, educational, research utilisation – tend to happen closer to the premises of the organisation.

"Yes, certain models can be trained far afield [...] but the utilisation of the model is often done then in-market."

From WESCO's perspective, supporting edge deployments requires the same fundamentals as hyperscale construction.

"The reality is... most of these environments, whether they be edge or large-scale, giga-scale environments, a lot of the general tasks have occurred, and will continue to occur, in a very similar fashion."