Honeywell Outlines Data Centre Infrastructure Challenges

The AI revolution is transforming data centre power requirements, with rack densities climbing from 10-12 kilowatts (kW) one year ago to projections of 600kW per rack by 2027.1,2

The surge in large part stems from AI-specific workloads consisting of both training and inferencing of increasingly complex large language models (LLMs), alongside traditional cloud services that businesses now consider baseline workloads.

Blake Clark, Senior Solutions Architect at Honeywell, emphasises the role of data centres in delivering AI capabilities. 

“Every AI application depends heavily on data centre infrastructure, whether early in the lifecycle during training or in some cases even inferencing,” he says.

“When it comes to AI, we’re barely scratching the surface of what pivotal use cases and capabilities will materialise for both enterprises as well as consumers. Practically speaking, AI without data centres simply would not exist.”

Honeywell identifies power grid availability as primary deployment barrier

Energy supply is a primary constraint on data centre expansion, rivalling traditional obstacles including community opposition and skilled labour shortages. 

Alpesh Saraiya, Honeywell’s Senior Director, Data Centre Global Marketing and Growth, has observed that end users are challenged with sourcing a reliable – and ideally emissions-reducing – energy supply.

He cites the Republic of Ireland as a particular example, explaining how operators within the country are facing challenges with a government moratorium requiring data centres to source their own power.

Blake adds that power infrastructure development timelines create additional complications.

“Operators must build near main transmission lines, but developing power and water infrastructure takes longer (the lead times are often three to five years or more) and costs more than connectivity infrastructure,” he notes. “Despite these constraints, funding remains available for expansion with major data centre players having mandates to build rapidly.”

Additionally, data centre uptime requirements have reached levels where even seconds of interruption carry severe penalties. Revenue generation during normal operations can reach millions of dollars per minute for large facilities, making reliability paramount. 

“An outage for a data centre that could be as small as a couple of seconds, maybe a minute, can potentially take hours if not days to get operations back to normal,” Blake explains.

An emphasis on standardisation for hyperscale programmes

Hyperscale operators often now design and build up to 100 data centres sites simultaneously across global regions, creating an imperative for standardised solutions and using vendors with proven track records. 

This contrasts dramatically to smaller operators who typically build one or two facilities at a time, according to Honeywell.

Standardisation has become so important across the industry that companies are investing in retrofitting older facilities to match newer deployments. This has inevitably shifted priorities within the data centre, with velocity and quality of build being the critical factors over component-level differentiation.

Ultimately, vendor selection criteria prioritises company stability and deployment history over cutting-edge technology. For one thing, data centre operators often prefer two years of auditable shipment history to avoid becoming test cases for unproven solutions.

“While there can be an exceptional circumstance, in normal practice data centres don’t want to be the guinea pig to try new things – given their mission-critical operations where downtime is not an option,” Alpesh explains.

Increasing regulatory pressure adds additional urgency to standardisation efforts. For instance, carbon footprint reporting requirements implemented in 2024 now carry sanctions from January 2025 for some EU-based operators missing reporting or reduction targets. 

This has prompted operators to retrofit existing facilities to achieve standardisation benefits across their global footprint. 

With data centres now taking action to advance such operations, Honeywell has positioned itself to provide on-time, on-budget project management, delivery and service through its integrated deep solution portfolio to help data centres achieve up to 99.999% uptime. 

1 Data Center Knowledge, “Data Center Rack Density Has Doubled. And It's Still Not Enough,” Bill Kleyman, April 15, 2024. [Accessed June 9, 2025]

2 Data Center Dynamics, “Nvidia's Rubin Ultra NVL576 rack expected to be 600kW, coming second half of 2027,” Sebastian Moss, March 18, 2025. [Accessed June 9, 2025]

Disclosure: This article is an advertorial, and monetary payment was received from Honeywell. The content was crafted by the Data Centre Magazine editorial team.

Explore the latest edition of Data Centre Magazine  and be part of the conversation at our global conference series, Tech & AI LIVE and Data Centre LIVE. 

Discover all our upcoming events and secure your tickets today

Data Centre Magazine is a BizClik brand