The Data Center Frontier Executive Roundtable features insights from industry executives with lengthy experience in the data center industry.

Here’s a look at the Q3 2026 insights from David Shepard, Global Head of Data Center Business, Chemelex-Raychem.

David Shepard is Global Head of Data Center Business at Chemelex-Raychem, a company with 44 offices in 28 countries. In this role, he is responsible for building and leading the company's data center business worldwide. Prior to his current position, David served as Vice President of Sales for Chilldyne, where he built global partnerships, drove the sales pipeline to nearly $200 million, and helped orchestrate the company's acquisition by Daikin Corporation. During more than 30 years in the technology industry, David has worked across all aspects of IT, from field engineering to the C-suite, with a focus on data center sales and operations, cloud services and IaaS, enterprise software, managed services, outsourcing, multi-vendor support, and professional services sales and operations.

Reducing cost and complexity through the strategic use of technology has been a central theme throughout David's career. As an entrepreneur, he has worked in startup and turnaround environments, building and operating data centers and multi-cloud environments, optimizing workloads, developing secure cloud IaaS and PaaS solutions, conducting IT performance audits, and designing business continuity programs. He has also delivered disaster recovery solutions to help organizations meet critical business continuity requirements. David has extensive experience building sales organizations across global markets, including the Americas, Asia-Pacific, and EMEA, and has successfully brought technology solutions to market worldwide. Throughout his career, he has carried a sales quota, led services and sales organizations, rebuilt businesses, and established successful global partnerships. A U.S. Air Force veteran, David is also actively involved in supporting veteran-focused causes and initiatives.

Data Center Frontier: What is AI infrastructure exposing that the data center industry could get away with overlooking at lower densities — and which traditional design or redundancy assumptions now need to be reconsidered?  

David Shepard, Chemelex-Raychem: The deployment and rapid adoption of AI infrastructure is beginning to expose a few areas of risk that were previously not top of mind with operators.  

Holding aside the “new normal” of extended supply chain issues, deploying IT infrastructure that is, say, 10 times more dense and 10 times more expensive puts a lot of pressure on operators to mitigate this new level of risk and protect these now far more expensive assets.  

As a for instance, we do leak detection in the white space.  Leak detection SOP 5 years ago was ropes or point sensors on the floor, or under the raised floor.  With AI infrastructures, those individual racks run between $2-7M per rack. That means traditional leak detection is simply not good enough, and “in rack, in CDU, in server” has become necessary.

Data Center Frontier: As data centers become larger, denser and more power-constrained, where is the definition of “critical infrastructure” expanding — and are the interfaces between power, storage, cooling, controls and other systems becoming the real reliability risk?  

David Shepard, Chemelex-Raychem: “Critical infrastructure” – once simply power and connectivity – now must be thought of more holistically, and must now include adjacent technology like leak detection, freeze protection and Data Center Infrastructure Monitoring (DCIM).  

Data Center Frontier: With time-to-power compressing development schedules, where can operators safely accelerate deployment — and where do commissioning, testing and operational visibility need to become more rigorous rather than less?  

David Shepard, Chemelex-Raychem: Deployments on a smaller scale, where possible, enable the operators to gain real world operational competence. When done that way, the scale up is more easily managed, though it is always a challenge. And managing it is far more rigorous.  

Practically speaking, think of it like this: Operator A is a large colocation provider. The million square feet under management today are handled by expert staff who know the gear and the process. The operator wants to deploy AI infrastructure in a secure, well-run environment and introduce risk (density/cost, water/plumbing, etc…) into a massively risk-averse environment.  

Starting up a smaller version of the final deployment, building competency and comfort around the new infrastructure, will pay dividends.

Keep pace with the fast-moving world of data centers and cloud computing by connecting with Data Center Frontier on LinkedIn, following us on X/Twitter and Facebook, as well as on BlueSky, and signing up for our weekly newsletters using the form below.