Solmar Insights
As artificial intelligence accelerates the transformation of digital infrastructure, data center operators are under pressure to support denser, more power-hungry compute environments with resilient facility designs and operations. Scott Longhurst, Data Center Services Director at Datalec Precision Installations (DPI), shared in a recent industry podcast the operational shifts underway as demand grows for AI-ready capacity and advanced cooling solutions.
Key figures
Higher rack densities driven by AI workloads
Increased power and liquid-to-chip cooling requirements
35+ years of mission-critical operations experience cited
AI spurs demand for dense compute
The rapid proliferation of artificial intelligence is directly impacting the technical requirements of data center design and operations. Longhurst reported that AI adoption is driving swift increases in rack density and overall facility power requirements. In practical terms, this means more equipment, and more heat, must be managed within the same footprint, challenging legacy mechanical and electrical systems.
Rack power requirements are moving beyond previously standard thresholds, putting pressure on operators to source sufficient electrical capacity and rethink their approach to thermal management. AI workloads, such as training and inferencing for machine learning models, frequently drive surges in compute density well above conventional enterprise applications. This evolving compute profile influences facility layouts, power delivery infrastructure, and underlying chassis and racks.
In this shifting environment, DPI’s work in fit-outs, modular deployments, and full-lifecycle data center support positions it at the intersection of physical infrastructure and advanced AI integration. These new deployment realities add complexity to both new builds and retrofits, with tenants demanding AI-ready features as a prerequisite for long-term occupancy.
Rise of liquid cooling and operational boundaries
To manage the thermal loads imposed by high-density AI deployments, Longhurst emphasized the importance of liquid-to-chip cooling. Unlike legacy air-cooling methods, liquid systems offer the efficiency required to directly dissipate concentrated heat from advanced processors, such as those powering generative AI and analytics workloads.
This move toward liquid cooling is not merely a technical challenge but also transforms operational practices. There emerges an explicit division of responsibilities between landlord-managed infrastructure and customer-managed systems. With more of the cooling managed at the rack and even chip level, close coordination is needed to mitigate fault domains, define service boundaries, and maintain clear accountability across operators, tenants, and equipment providers.
Real-time monitoring and advanced documentation become essential under these conditions. DPI offers both deployment and ongoing support for specialized cooling, illustrating how vendors now deliver not just build services but also maintenance and integration expertise for these more complex thermal solutions.
Beyond redundancy: comprehensive resilience
Longhurst argued that resilience in the AI era is no longer achieved purely through redundant electrical, cooling, or networking systems. Instead, it requires a combination of proven operational processes, comprehensive documentation, reliable supply chains, and highly trained operations teams. This multi-layered approach is increasingly vital as both the complexity and business impact of outages rise in high-performance compute environments.
Effective resilience demands predictive maintenance plans, trend data analysis, and robust response protocols that adapt to infrastructure and client changes. DPI’s role is to implement these strategies across the project life cycle, helping facility operators anticipate potential failure modes and intervene before disruptions impact clients.
With adoption outpacing the training of skilled technical staff, the industry faces a notable workforce challenge. The need for both AI-specific operational knowledge and traditional mission-critical skills gives operations teams a broader scope of responsibilities, raising the stakes for investment in training, documentation, and external support partnerships.
Proactive maintenance and lifecycle management
Proactive maintenance featured prominently in the discussion as a differentiator for resilient data center operations. Longhurst called for a shift away from strict calendar-based servicing toward predictive, condition-driven management. By using trend data, lifecycle planning, and root-cause analysis, operators gain earlier visibility into components that risk imminent failure.
This method enables more cost-effective asset management and risk reduction. Facility managers can prioritize investments in aging or stressed equipment, aligning replacement and upgrades with actual operational signals rather than arbitrary deadlines. The outcome is a more reliable, available environment for mission-critical workloads, increasing the value of both colocation and dedicated capacity in competitive digital markets.
Longhurst’s approach also supports customer commitments by providing clear data for service-level expectations, facilitating contractual clarity between landlords and tenants while supporting smarter investment in asset renewal and upgrades.
Efficiency and the path forward
Alongside resilience, energy efficiency remains central to the future of AI-driven data centers. The industry is adopting strategies such as higher allowable operating temperatures, more efficient uninterruptible power supply (UPS) systems, and variable-speed cooling equipment. These measures help temper rising energy use even as power densities climb.
DPI aims to support customers throughout the full facility lifecycle, from initial planning and modular construction to commissioning and ongoing operation of advanced cooling systems. The company’s involvement illustrates the emerging need for comprehensive partners able to support complex sites as the digital infrastructure market increasingly targets AI and high performance computing tenants.
As compute demands escalate, the intersection of operational resilience, efficiency, and innovative cooling approaches will continue to shape investment decisions and technical standards within the US data center industry.
What this means for buyers
Data center capacity and specialized cooling equipment in the US are directly affected by rising AI rack densities and new liquid-to-chip cooling requirements. As workload density grows and hybrid landlord-tenant operational models emerge, procurement choices must prioritize partners with lifecycle support and proven resilience strategies. Buyers should emphasize equipment with advanced monitoring and flexible maintenance, as proactive support is increasingly required to meet uptime commitments this quarter.
Reporting via the original publisher
