Facility executives and managers need to re-evaluate the way they distribute power internally, from the power grid to the racks and eventually to the computer chips — so-called grid to chip. Historically, data centers relied on alternating current (AC) architectures, but that is evolving very quickly, says Bin Lu, executive vice president of power products at Schneider Electric.
“Traditionally, it was more alternating current based,” he says. “Now, more and more you see hybrid AC, DC (direct current) architecture, especially in the last year as Nvidia announced 800-volt DC architecture.”
Now the entire industry is preparing for 800-volt DC systems, which offer many benefits. First, it improves energy efficiency by minimizing energy losses during power distribution. It also enhances reliability, lowers maintenance costs and allows for higher power capacity and better voltage stability, all of which makes it suitable for data centers and other mission critical facilities.
Another emerging strategy is localized DC power conversion.
“It’s basically a cabinet that you put next to the computer rack,” Lu says of Schneider’s 800-volt DC sidecar concept. “It converts the incoming AC power into 800-volt DC power.”
Localized DC power conversion also enables distributed energy storage.
“A sidecar will have modular battery storage as a part of the cabinet,” he says. “That helps you to manage short-term backup.”
Busways from Starline that range from 250-6,000 amps can be installed above the racks and plug-in units added to the system to deliver more power where needed.
“As long as the busway is sized appropriately, they just need to add more plug-in units,” he says. “The backbone infrastructure doesn’t need to be replaced if the busway is sized appropriately from the beginning.”
But facility managers must be able to look down the road and consider their future energy needs, Lu says: “Doing this will help design a system that can scale with them.”
AI is part of the problem but also part of the solution, says David Chernis, director of flexible compute platforms for CPower Energy.
“Whether it is for utilities making the grid smarter or a virtual power plant provider automating hourly market bids and operating a data center’s flexible demand response assets in near-real time, AI takes on the complex work behind the scenes,” he says.
Flexible energy, demand response and virtual power plants also offer fast, scalable pathways to speed connections to power and unlock innovation.
As complexity grows, AI and unified data platforms are becoming essential tools for managers.
“AI is real,” Lu says. “We talk about AI as it relates to energy demand by data centers, but we forget that GenAI is also a great enabler. Platforms that leverage AI capabilities open up a whole new world for facility managers. These tools help them manage their facilities with better transparency from energy usage patterns to space utilization, asset management, condition-based monitoring, diagnostic capabilities and more. These tools bring the physical side of the facility and the energy and power side of the facility together into one pane of glass.”
When these platforms also are AI enabled, managers gain visibility and predictive insight. They can monitor energy use and see the amount of energy each asset uses. It provides managers with equipment condition monitoring in the data center.
Chernis says data center infrastructure management (DCIM) platforms are increasingly used to coordinate these systems. These platforms offer software tools designed to assist data center operators in overseeing, controlling and enhancing the physical and IT aspects of a data center. They provide tools for tracking assets, energy consumption and environmental conditions to improve efficiency and reduce operational costs.
“Whether it is coordinating predictive maintenance, optimizing energy usage or maximizing revenue generation and on-bill savings in demand response and energy flexibility programs, AI-driven monitoring and automation can improve operations for data centers specifically and buildings in general,” Chernis says.
When it comes to preparing for data center growth, Lu says there are four priorities for facility executives and managers: reliability and availability; energy efficiency; sustainability; and digitalization and AI adoption
He says managers should start with reliability and availability by adding redundant power. Beyond this, he says managers should check that the facility has the most appropriate system in place, whether it is a traditional uninterruptible power supply (UPS) or energy storage.
“You need to make sure that if you lose power from the grid, your lights stay on, your data center stays on,” he says.
Next, he recommends that managers improve energy efficiency with facility and electrical upgrades that can be done now. Once those improvements have been made, Lu says managers should consider on-site power generation — such as a microgrid or solar panels, which also contributes to the building’s sustainability.
Finally, he says managers should embrace digitization, which requires managers to learn about and engage with new technologies.
“Challenge yourself to understand the new technologies becoming available — the architecture evolution, hybrid AC, DC and then digitalization,” Lu says.
Sadiq Syed, senior vice president of the digital energy software business with Schneider Electric. agrees, adding, “Technology has a key role to play. It helps you get to the heart of what is driving energy consumption and makes it actionable for facility managers so they can move the needle and use less energy.”
Addressing power needs through advanced technologies can be phased, says Mark Swift, engineering and product management lead at Starline.
“You don’t need to replace all of your equipment at one time,” he says. “Maybe upgrade a smaller subset of the existing data center but leave the main data center infrastructure alone. This gives facility managers a nice upgradeable path forward.”
The takeaway for facility executives and managers is clear: managing a modern data center requires a broader, more strategic skill set.
“The world is moving very quickly,” Lu says. “And congratulations, facility managers, especially those overseeing data centers. You're at the tip of the iceberg.”
From high-capacity power distribution and BESS to DCIM, unified data platforms and flexible energy markets, facility managers are operating at the intersection of IT, energy and sustainability.
“What we’re seeing now is like no other time in recent history for the data center space,” Swift says. “The iterations on design have accelerated so greatly. It’s almost unfathomable.”
For facility executives and managers, the message is unmistakable: The AI data center is here, and it is rewriting the playbook on managing mission-critical facilities.
Ronnie Wendt is a freelance writer based in Minocqua, Wisconsin.
Key Advances Rewrite Data Center Management
The AI Data Center Demands a New Skill Set
