Data Power Technology Group has been serving the data center industry since the 1980’s, the same era as Marty McFly and Doc Brown from the film classic “Back to the Future”.. Over the past four decades, we’ve watched data center design evolve through steady, predictable progress.
Today, however, AI is rewriting the playbook, creating infrastructure challenges that would have seemed almost as improbable as powering the DeLorean four decades ago. New demands for power, cooling, scalability, and deployment speed are transforming the way facilities are designed, built, and operated, impacting everyone from owners and consulting engineers to contractors, construction managers, system integrators, and end users..
During our organization’s most recent annual market review, one trend stood out above all others: between June 2025 and June 2026, we experienced a sevenfold increase in AI-related applications across the markets we serve.. Those opportunities span hyperscale, colocation, enterprise, and Edge deployments.
Each year, our data center sales engineers and application specialists come together to review customer projects, emerging technologies, and evolving market needs across mechanical, electrical, and IT infrastructure. This year’s discussion confirmed what many across the industry are experiencing firsthand: AI is accelerating change at a pace unlike anything we’ve seen before..
Power constraints, resource availability, scaling and futureproofing: Today’s AI challenge. “1.21 Jigawatts!”; “What the hell is a Jigawatt?”; … “Tom – How am I going to generate that kind of power?”; … “It can’t be done, it can’t”; … “Doc, you’re my only hope.”.
At times, we might feel like we’re part of that iconic exchange between Doc and Marty as they struggle to generate enough power for the Flux Capacitor. Many of the same themes – power availability, timing, scalability, and futureproofing – are echoed in today’s conversations with our clients, although thankfully with real-world solutions..
Some of the most common questions and design challenges our clients are bringing to us today include:. Electrical power utility availability and timing: will I have enough and have it when I need it during construction and phasing?. What about this 800V DC power in the data center concept?.
Server rack densities and corresponding heat loads – what racks can I buy today that will still make sense in two or five years?. What I need today for cooling loads changes as we build out. How can I design a chilled water/liquid cooling loop or plant to accommodate that?.
How can we phase this install to ramp up compute power in the short term and scale up in the future? How do we perform commissioning under temporary power conditions?. How do I handle these spiky electrical loads generated by AI workloads?.
How do we get enough skilled labor to our jobsite to build our project?. Water: How do I use less water?. How can I design my data center to be a good neighbor and mitigate community pushback?.
The power of learning from the future: Capturing lightning. “Marty I’m sorry, but the only power source capable of generating 1.21 Jigawatts of electricity is a bolt of lightning.” “What did you say?” “A bolt of lightning, unfortunately you never know when or where it’s ever going to strike.” … “We do now.”.
For some in the data center ecosystem, they may be thinking about or have started to implement their first AI deployment. For others, multiple iterations of AI system deployments have yielded significant lessons learned. Whether it be a one- or two-rack initial deployment or 100+ MW scale, the lessons are powerful.
Tapping into the knowledge base and lessons learned within the industry is a valuable resource. Some of the best practices and emerging technologies shaping AI infrastructure today include:. Always start with the chipset (today’s chipset and future; look two generations ahead) and design the infrastructure out from there..
Find infrastructure partners that are integrated with and speak the vernacular of the chip providers. A proactive, integrated partner is invaluable in delivering at speed and scale.. Get the best team assembled as early in the process as possible.
Designs, practices, and equipment lead times are changing rapidly; however, system component alternatives and lead time issues can be mitigated by looking at all possible solutions early in the process.. Battery energy storage systems (BESS) and fuel cell technology for on-site, behind-the-meter, off-the-grid power solutions..
Dynamic load smoothing UPS solutions.. Hot aisle containment integrated systems incorporating racks, manifolds, busway, structural support and all downstream infrastructure.. CDUs with built-in harmonic mitigation VFD & AI-ready technology..
Chillers with automatic restart and onboard UPS technology.. Cooling towers, dry coolers and adiabatic coolers with modularized blocks for large scale, maximized cooling capacity in one single cell.. Plume abatement and reduced sound emission evaporative cooling technologies..
Water-reducing, hybrid cooling technologies; including indirect evaporative cooling.. Pre-developed, customizable mechanical, electrical, and IT component blocks for rapid deployment – streamlining design, fabrication, shipping, installation, start-up, and commissioning while reducing overall project lead times..
Closing perspective: Channel our inner Doc Brown and Marty McFly. If you get a few minutes, it is worth watching the 2:30ish clip of Doc and Marty’s “Eureka” moment. Faced with what seemed like an impossible technological challenge, they worked through the problem step by step until they found a solution.
Our industry is doing much the same today. Every new AI deployment expands our collective knowledge, helping us solve increasingly complex challenges with greater confidence and speed.. Like Doc and Marty, we’re tackling challenges that once seemed impossible.
By building the right teams, applying the best available knowledge, embracing new technologies, and learning from every deployment, we’re delivering AI infrastructure that is more scalable, more resilient, and better prepared for what’s next.. Who knew Doc Brown and Marty McFly would be helping us build AI data centers?.
At Data Power Technology Group, we partner with data center stakeholders to deliver scalable, high-performance infrastructure solutions that keep pace with accelerating demand. If you would like to continue the conversation, I welcome you to reach out to me directly at [email protected].
