Building the AI Factory: Power, Cooling, and Execution at Scale Meets the Deployment Reality Gap - Q2 Executive Roundtable
At Data Center Frontier, we rely on industry leaders not only to help us understand the most urgent challenges reshaping digital infrastructure, but also to illuminate the broader technological, operational, and market forces driving the industry's evolution. And in the Second Quarter of 2026, those challenges increasingly revolve around a fundamental shift in emphasis: the industry is moving beyond discussing AI infrastructure in theory and into the far more demanding work of deploying, operating, and scaling it in production.
The era when hyperscale announcements and GPU roadmaps dominated the conversation is giving way to one defined by execution; where power availability, thermal management, construction schedules, supply chains, and operational discipline determine whether ambitious plans become functioning AI factories. That transition is exposing new realities. Rack densities continue to climb, liquid cooling is becoming mainstream, electrical architectures are evolving, and project timelines are compressing even as capital commitments reach unprecedented levels.
Success increasingly depends not on optimizing individual systems in isolation but on orchestrating tightly integrated environments where compute, power, cooling, networking, and facility operations function as a unified whole. At the same time, moving from pilot deployments to industrial-scale AI infrastructure introduces an entirely different class of challenges around reliability, maintainability, commissioning, and repeatable execution.
For our Q2 Executive Roundtable, we brought together senior leaders whose expertise spans AI infrastructure design, mission-critical deployment, advanced thermal management, and engineering innovation to examine where the industry stands today, and what it will take to bridge the gap between AI ambition and AI deployment at scale. Drawing on perspectives from hyperscale execution, liquid cooling, and next-generation power and facility engineering, their insights explore the practical realities of building the AI factory at industrial scale.
Our distinguished Executive Roundtable panelists for Q2 of 2026 includes:
- Steve Altizer, President and CEO, Compu Dynamics
- Joe Capes, VP, Trane Technologies and Head of LiquidStack
- Robert Danforth, Director–Engineering, Simulation, and Advanced Development, Rehlko
Today: The Deployment Reality Gap
Today’s kickoff article begins on the premise of how the AI infrastructure race is no longer constrained by vision or capital. Across the industry, hyperscalers, cloud providers, enterprises, and emerging AI platforms have demonstrated an extraordinary willingness to invest in next-generation capacity. The harder question is whether the digital infrastructure ecosystem can translate that ambition into operational reality on the timelines now being demanded.
Today's bottlenecks extend well beyond GPUs. Power availability, thermal management, supply chains, construction execution, and systems integration have become deeply interdependent, creating a widening gap between what can be conceived on paper and what can be commissioned in production. As rack densities surge and liquid cooling moves into the mainstream, successful AI deployment increasingly depends not on any single technology breakthrough, but on the industry's ability to synchronize dozens of complex variables across the entire delivery chain.
Against that backdrop, we asked our Executive Roundtable participants to identify where they see the greatest disconnect between AI infrastructure ambition and deployment reality, and what those gaps reveal about the next phase of building the AI factory.
What's Ahead in the Roundtable
Over the rest of the week, our panel will also share their perspectives on several other pressing themes shaping the next phase of AI infrastructure development, including:
- The Rise of Integrated Infrastructure: Why the convergence of power, cooling, networking, and facility operations is becoming essential to support increasingly dense and complex AI deployments.
- Scaling Beyond the Prototype Phase: The operational, engineering, and execution challenges that emerge as AI environments transition from proof-of-concept installations to industrialized production at scale.
Now let's move onto Data Center Frontier's first Executive Roundtable question for the Second Quarter of 2025:
Data Center Frontier: As rack densities, cooling demands, and power requirements accelerate simultaneously, where do you see the greatest disconnect today between AI infrastructure ambition and deployment reality?
Steve Altizer, Compu Dynamics: Today, the traditional data center development industry is trying to support a new class of AI workloads inside environments that were largely designed and optimized for a different operating profile.
Demand for large scale and multi-tenant data centers still exists and will continue to grow. Those facilities play a critical role in the market. But many were built around more cloud-level rack densities, longer refresh cycles, and a relatively static handshake between the supporting infrastructure and the enterprise IT environment inside the data hall.
AI workloads function better in a different hosting environment - one that offers more flexibility than today’s data centers. We are seeing GPU deployments move well into the hundreds of kilowatts per rack. This introduces highly disruptive thermal profiles, heavier equipment, more dense power distribution, massive interconnect requirements, and hardware roadmaps that can shift every six to twelve months.
Bridging that gap between the facility and the workload is exactly what we do. For decades, Compu Dynamics has been a market leader in designing and building high-efficiency, mission critical environments. AI is changing virtually every element of the traditional environment. Data centers now have to be planned around the application, the cooling demands, especially the shift toward liquid cooling, and the supporting infrastructure from the beginning.
We are having these conversations frequently with data center owners, hyperscalers, and neocloud companies. In essence, AI infrastructure is not simply a denser version of cloud infrastructure. It operates best on a completely different kind of power and cooling foundation, with a different deployment model, and at a pace that is unprecedented.
Joe Capes, Trane Technologies/LiquidStack: Power is still the number one issue in the scale-up of AI. Whether it’s grid capacity, challenges with interconnects, or long lead times for transformers and switchgear, there are many headwinds affecting power availability and distribution. Beyond that, the biggest disconnect I see is that getting infrastructure shipped and getting infrastructure deployed are two very different things, especially when it comes to liquid cooling.
There's a huge amount of focus on securing power, GPUs, and capacity. All of those matter, but none of it creates value until the system is operating the way it's supposed to. What we're seeing across the industry is a push to bring capacity online as quickly as possible, even if it means that the general contractor, operator and tenant have to cut corners.
The challenge is that power, cooling, controls, installation, commissioning, and IT all have to come together at the same time. If one piece isn't ready, the whole project can break down. Building the infrastructure is only part of the job. Getting it running reliably is what really matters.
Robert Danforth, Rehlko: I think the biggest disconnect right now is that the industry is still talking a lot about capacity, while not talking about performance needed for the future.
Everyone is focused on securing enough power to support AI growth, and that's obviously important. But AI workloads are introducing operating conditions that look very different from what most data center infrastructure was originally designed for.
Historically, power systems were built around relatively stable operating conditions with occasional transient events. AI changes that equation. Large-scale training environments create highly dynamic, synchronized load patterns where what used to be an occasional disturbance increasingly becomes part of normal operation.
At the same time, we're seeing grid constraints, long interconnection timelines, and increasing pressure to bring capacity online faster than ever. The result is that simply having access to megawatts isn't enough. The bigger question is whether the infrastructure can deliver stable, reliable performance under real AI operating conditions.
As AI deployments continue to scale, I think we'll see the conversation shift from "How much power do we have?" to "How well does the entire system perform when AI workloads are running at full scale?" That's also why we're seeing growing interest in simulation and modeling tools that help operators understand system behavior before infrastructure is deployed.
The organizations that get ahead will be the ones designing around real-world operating conditions rather than theoretical capacity.
NEXT: The Rise of Integrated Infrastructure
About the Author
Matt Vincent
Matt Vincent is Editor in Chief of Data Center Frontier, where he leads editorial strategy and coverage focused on the infrastructure powering cloud computing, artificial intelligence, and the digital economy. A veteran B2B technology journalist with more than two decades of experience, Vincent specializes in the intersection of data centers, power, cooling, and emerging AI-era infrastructure. Since assuming the EIC role in 2023, he has helped guide Data Center Frontier’s coverage of the industry’s transition into the gigawatt-scale AI era, with a focus on hyperscale development, behind-the-meter power strategies, liquid cooling architectures, and the evolving energy demands of high-density compute, while working closely with the Digital Infrastructure Group at Endeavor Business Media to expand the brand’s analytical and multimedia footprint. Vincent also hosts The Data Center Frontier Show podcast, where he interviews industry leaders across hyperscale, colocation, utilities, and the data center supply chain to examine the technologies and business models reshaping digital infrastructure. Since its inception he serves as Head of Content for the Data Center Frontier Trends Summit. Before becoming Editor in Chief, he served in multiple senior editorial roles across Endeavor Business Media’s digital infrastructure portfolio, with coverage spanning data centers and hyperscale infrastructure, structured cabling and networking, telecom and datacom, IP physical security, and wireless and Pro AV markets. He began his career in 2005 within PennWell’s Advanced Technology Division and later held senior editorial positions supporting brands such as Cabling Installation & Maintenance, Lightwave Online, Broadband Technology Report, and Smart Buildings Technology. Vincent is a frequent moderator, interviewer, and keynote speaker at industry events including the HPC Forum, where he delivers forward-looking analysis on how AI and high-performance computing are reshaping digital infrastructure. He graduated with honors from Indiana University Bloomington with a B.A. in English Literature and Creative Writing and lives in southern New Hampshire with his family, remaining an active musician in his spare time.
Voices of the Industry
