The UPS As Critical Infrastructure: Powering AI-Driven Futures

Author: Schneider Electric

This audio was created using Microsoft Azure Speech Services

Imagine a scenario where millions of dollars disappear not due to a bank heist, but because of a single keystroke that cripples a vast hyperscaler cloud. This isn’t science fiction; it happened in 2017 when a small typo caused a domino effect, disrupting their services. When our vast connected infrastructure loses its ability to compute, what then?

The culprit? A weakness in our digital infrastructure.

Let’s take a moment to appreciate the “ALWAYS ON” mantra that defines our digital age. The 2017 hyperscaler’s outage served as a wake-up call, showcasing the fragility of our digital backbone and the immense cost of disruptions. But what if a similar event hit healthcare systems or financial institutions? The potential consequences are catastrophic.

Moreover, in our “always on” world, the urgency for resilient infrastructure mounts as our world faces more disruption from climate change to utilities cybercrimes. Power outages, once sporadic, are becoming more frequent and geographically diverse.

Consider this: While the average frequency of outages in the U.S. is 1.62 annually, the variance is significant by location. States like Maine experience over 4x more outages than Washington D.C. Manufacturing and data center hubs such as Texas, Tennessee, Virginia, and Florida are all over-indexed. The average outage duration ranges are stunning: from 90 minutes in Delaware to an astonishing 25 hours in Louisiana.

These vulnerabilities became apparent during the 2021 Texas winter storm, “Uri.” Local power provider Austin Energy cut off the city’s biggest electricity users, including chip factories owned by Samsung, NXP Semiconductors, and Infineon Semiconductors. This planned outage led to $360 million loss for Samsung alone, highlighting the global ripple effect of power disruptions. Similar situations are becoming more frequent. Today, the Electric Reliability Council of Texas, ERCOT is asking San Antonio to cut electricity use, proving the point on these ongoing challenges.

As businesses reshore their operations (nearly 70% of US manufacturers are doing so according to Forbes manufacturing expert, Jim Vinoski) and our reliance on AI, automation, and quantum computing grows, the demand for reliable power intensifies. It’s clear that investing in resilient infrastructure is not a luxury but a necessity.

Moving forward: Investing in a stable digital future

The good news is in the advancements being made. Utilities and the U.S. Department of Energy (DOE) are exploring AI and quantum-powered grid optimization for better simulations and increased use of renewables. Businesses are taking a more proactive approach by securing power to be able to address future growth.

However, a more intelligent and adaptable system, with a focus on resilience, reliability, and flexibility, is still needed. The ever-growing demand for power due to advancements in AI, automation, and quantum computing necessitates immediate action.

When people think about critical infrastructure, the typical mental model often includes power plants, servers, and significant transportation networks—those large-scale elements that appear most prominent in downtime concerns. But that is not sufficient in today’s tech-driven world. A business’s energy infrastructure must also have reliability, resilience, and trustworthiness. These are not always about grand things but the connective components that act as bridges and supports.

UPS for the win

The UPS is now more important to protect data centers, large businesses and facilities. There are two main reasons why AI data centers absolutely need uninterruptible power supplies (UPSs):

To prevent data loss and downtime: AI systems are constantly processing information and learning. A power interruption, even for a brief moment, can cause data loss or corrupt ongoing calculations. This can set back training progress or lead to inaccurate results. Additionally, downtime in an AI data center can disrupt services that rely on AI, leading to lost revenue or frustrated users.

To protect sensitive equipment: AI data centers house powerful computer systems that are susceptible to damage from power fluctuations or surges. A UPS provides a clean and consistent power supply, safeguarding this expensive equipment and ensuring its smooth operation.

UPS are fit for purpose tools

Many are tailored to work well in specific environments. Manufacturers need consistent power capacity and quality for high load machines; they need exceptional durability and offer runtimes sufficient to prevent production disruptions. Hospitals have life-supporting equipment, so redundancy tops the list – with zero points of failure and extended run times. They also have regulatory compliance to consider. For data centers: high capacity, efficiency, integrations for networking/remote monitoring and modularity are important. All of this comes alongside the core considerations of product quality, maintenance, and cost.

Consider scalability, efficiency, and sustainability in a UPS as well

Efficiency isn’t just about operating costs – it influences the environmental footprint. This goes to sustainability and circularity.

As power demand continues to grow, a flexible solution that is scalable/modular is critical – especially in this “AI-driven, always on” world. This means choosing UPSs that are scalable by design.

If you build for the future load, you want to balance capacity and utilization – too much unused capacity is not environmentally friendly, with greater energy loss.

Using software capabilities can provide an actionable picture of capacity and load and while maintaining reliability and resilience. It also helps you understand when expansion is needed.

Advanced software capabilities in UPS systems enable precise monitoring and management for preventive maintenance while helping to optimize performance for sustainability goals.

AI, Quantum Computing, and other future technologies increase the pressure for “uptime”

Ensuring power continuity in the face of increasing energy usage and unforeseen challenges will become increasingly central. This continuity can only be achieved when we look at the whole picture, including foundational elements like UPS, Batteries, and Power Distribution Units. And while we need to provide for today’s needs, sustainability for our future must be at the forefront of these system selections.

The key takeaway is clear

We are in an era of constant evolution. Increased instability of the grid necessitates investing in the resilience of “supporting” critical infrastructure and doing it sustainably. After all, we are depending on an “ALWAYS ON” digital future, where even brief disruptions can lead to chaos. It’s often the unsung heroes of critical infrastructure that keep us connected and thriving.

CALLS TO ACTION

To help operators and facilities managers make better decisions on UPS, the Energy Management Research Center has created two tools. Each Efficiency Comparison Calculator Tradeoff Tool compares the efficiencies of two UPS systems and shows the impact on electricity cost and carbon footprint.

Single Phase UPS Efficiency Comparison Calculator

Three Phase UPS Efficiency Comparison Calculator

Tags: data center, Energy Management Research Center, power continuity, resiliency, UPS systems