Sponsored Content.

Data centers have become modern megaliths in a new world of infrastructure powering the next wave of innovation, and artificial intelligence (AI) is redefining how they’re designed and powered. As AI workloads intensify, they’re creating new demands for efficiency, resilient power delivery, and how we define energy strategies that will determine the future of digital infrastructure.

“Data centers have become the places where AI model training and deployment occur, and that role is central to recent growth in electricity demand,” cites The International Energy Agency (IEA).¹ Because so much of the load is IT equipment, the IEA also points out that “Servers account for around 60% of electricity demand in modern data centers, underscoring why power has become such a hot topic when it comes to global growth in technology.”

Those power needs are colliding with practical limits on grid expansion. Building new transmission or waiting for interconnection approvals can take years, so data-center owners are combining strategies to secure capacity quickly. Many remain grid‑connected while signing long‑term renewable energy contracts and adding battery back-up to smooth short‑term variability. As McKinsey & Company explains, “Operators are pairing grid connections with behind‑the‑meter solutions, hybrid systems and advanced controls to handle fast-growing demand and interconnection constraints.”³ Where speed-to-power is more urgent, some operators install behind‑the‑meter generation—often modular gas turbines or generator sets—because on‑site assets can deliver capacity faster than waiting for grid upgrades. Siemens Energy captures this point plainly: “One solution to this challenge facing the data center industry is on-site generation of electricity and cooling.”⁶

Batteries are now a standard part of that toolbox. They provide immediate response during disturbances and shave short peaks so mechanical generators aren’t taxed beyond what they are designed to do. As the article by Facilities Dive observed, “Flexible battery or generator solutions can help data centers power up faster, reduce grid impacts and keep their owners’ sustainability goals within reach.”⁸ In short, hybrid mixes of grid power, long‑term renewable purchases, batteries and on‑site generation are practical and increasingly common.

Power generation equipment is not immune from these changes. Power plants tasked with serving AI loads may be asked to cycle more often and operate across a wider load range than many traditional baseload or peaking plants—behavior that increases thermal and mechanical stress on turbines, generators, and the gear drives linking them. McKinsey & Company highlights the problem in their online article, “Power supply is becoming an issue in markets that have traditionally attracted clusters of data centers, which is driving interest in dedicated generation and hybrid systems.”⁴ That means gear designs need to cope with more starts/stops, ramping, and generally more variable torque profiles than in historical applications.

With these increased demands, availability and predictability have become even more critical when it comes to power generation. Data centers expect near‑continuous operation, and backup systems are fundamental to achieving that availability. The Lawrence Berkeley Lab report states plainly that “UPS batteries and backup generators are there to keep the data center powered during outages and that these systems are essential to ensure the extremely high levels of reliability that data centers must meet.”² For gear suppliers, that translates into customers demanding not only mechanical robustness but also rapid serviceability.

So, what should gear companies do? There are several practical moves that align product and service offerings with data-center power needs. First, emphasize durability in designs intended for high‑cycle, variable‑torque duty: stronger tooth profiles, improved bearings, advanced sealing and lubrication strategies, and conservative designs that can handle a wider thermal range to reduce risk of early failure. Second, make the product serviceable by offering rapid response offerings, like Philadelphia Gear’s Onsite Technical Services (OTS)™, and high-quality OEM parts to reduce downtime for repairs during critical outages. This includes the ability to serve mobile power solutions deployed from trailers in addition to traditional, fixed brick-and-mortar facilities. Third, add simple but reliable condition monitoring for both mechanical and lubrication systems including vibration, temperature, and acoustic instrumentation so customers can incorporate predictive‑maintenance programs. McKinsey and others point to procurement trends where buyers weigh total cost of ownership, scalability, and service readiness—factors gear companies can influence through design and aftermarket offerings.^3,4

There are also product-level opportunities beyond the gear drive itself for companies that can offer more than just off-the-shelf products and instead provide engineered system solutions. So, whether the end-user is uprating current systems, needs efficiency improvements, or does not have the expertise to build a broader power‑system solution, all of these represent a competitive opportunity for OEMs. GE Vernova and Siemens Energy both emphasize the value of integrated approaches that combine generation, controls, and lifecycle services for data-center power applications.^5,6

In short, AI data centers are increasing electricity demand and, importantly, changing how that power is delivered and how generation equipment should function in this new world. Gear companies that combine proven mechanical reliability with rapid service capability, and partnership-oriented equipment solutions will be best positioned to support power plants and the various on-site power solutions serving AI workloads.

“We’ve already started working with AI data centers looking for help in meeting their energy demands,” said Carl Rapp, President of Philadelphia Gear. “With over 130 years of experience supporting the energy industry, we’ve been side-by-side with our customers as their energy needs have grown and changed. And during that time, we’ve remained true to our roots as subject matter experts for critical power generation equipment. Our approach has always been to be a trusted advisor and build custom engineered products that solve specific challenges. So, whether it’s a new gear design or servicing the equipment over its lifecycle with aftermarket repair, parts, and service, we have built our business on running to and solving our customers’ most complex problems.”

Carl continued, “As a part of Timken Power Systems (TPS), Philadelphia Gear® is integrated within a network of manufacturing and service centers that provide electro-mechanical expertise for complex engineered systems that include gear drives, electric motors, generators, bearings, and control systems. For data center operators, expertise in a single discipline is no longer enough. That’s what TPS is all about, evolving alongside our customers’ needs to deliver broader, integrated capabilities that simplify operations and help their businesses run more efficiently.”

To learn more about Philadelphia Gear or TPS, visit our websites or scan the QR code to take a virtual tour.

Authors: Carl Rapp, president of Philadelphia Gear; and Rob Fisher, marketing & product manager for Philadelphia Gear.

---

References

1. IEA — Energy and AI: Energy demand from AI: https://www.iea.org/reports/energy-and-ai/energy-demand-from-ai
2. Lawrence Berkeley National Laboratory, 2024 United States Data Center Energy Usage Report: https://eta-publications.lbl.gov/sites/default/files/2024-12/lbnl-2024-united-states-data-center-energy-usage-report_1.pdf
3. McKinsey & Company — AI power: Expanding data center capacity to meet growing demand: https://www.mckinsey.com/industries/technology-media-and-telecommunications/our-insights/ai-power-expanding-data-center-capacity-to-meet-growing-demand
4. McKinsey& Company — How data centers and the energy sector can sate AI’s hunger for power: https://www.mckinsey.com/industries/private-capital/our-insights/how-data-centers-and-the-energy-sector-can-sate-ais-hunger-for-power
5. Siemens Energy — On‑site Power Generation for Data Centers (white paper): https://assets.new.siemens.com/siemens/assets/api/uuid:5d02c989-8681-4320-b4e6-5445fb1b9a60/sie-us-si-rss-data-centers-power-generation-whitepaper-en.pdf
6. GE Vernova — Gas Power Technology for Data Centers: https://www.gevernova.com/gas-power/industries/data-centers
7. Facilities Dive — Data centers seek flexible power solutions for resilience, sustainability: https://www.facilitiesdive.com/news/data-centers-seek-flexible-power-solutions-for-resilience-sustainability/753811/

---

By William Kaewert, Strategic Advisor and Board Member, Stored Energy Systems

Part 1 showed why outages and price spikes will pull on-site assets like generators and battery storage into weekly service. This continuation focuses on what equipment must do to succeed. Availability comes first with economics a close second. 

Fast forward to 2030: we’ll have demand outrunning firm capacity, and the grid shows it: 

Aging thermal plants are cycled hard, so forced outages creep up, while the gear that could relieve some of the pressure – large transformers, combined cycle gas turbine powerplants and new transmission – takes years to arrive. Grid batteries help for hours, but not for days of wind and solar drought. The result: more curtailments and more requests to dispatch on-site storage and power generation systems. Prices mirror the strain: negative at night, punishingly high in the afternoon and evening. On-site power will soon be on weekly or even daily duty rather than on standby. 

Now, if you want your system to become 2030-ready, it must be able to do two jobs: to keep critical loads online during blackouts, and generate cash by decoupling peak power usage away from periods of peak power pricing.   

Availability comes first with economics a close second.

This will result in a dramatic change in the requirements for on-site power. You will need storage that tolerates frequent cycling, bidirectional conversion that charges batteries quickly and can export back to the grid, and controls that let the plant act as a grid-synchronized resource without adding safety risk. 

What on-site power will look like in 2030 

While it’s too early to predict specific technologies or suppliers, we can reasonably identify some general requirements: 

The general concept of power flow in 2030 is that electricity will in many places flow into and out of customer facilities much like oceanic tides roll in and out. Although the reversal of flows will be much less predictable than tidal movement, changing flows of power are the consequence of increasing renewables penetration. Daily cycles of the sun, power demand, and the seasons mean that power generation will be increasingly difficult to match with demand. The only solution will be on site power generation and energy storage. These assets will be necessary to stabilize the inevitable mismatches between demand and renewable generation.

Storage 

Daily or weekly cycling materially changes the requirements for energy storage. Technologies to support an expected 200 cycles per year and 15-to-20-year calendar life places very different demands on batteries than legacy “standby” applications, such as UPS or switchgear backup duty.  

Batteries for revenue service must be sized much larger than today’s standby batteries. On top of the “insurance” job of powering critical systems during blackouts the batteries of 2030 will stack another job. This second job is the time-shifting of facility loads such that peak demand occurs during periods of low electricity prices, rather than during when energy is consumed. Many sites would need to upsize today’s stationary battery systems by ten times or more to satisfy the time-shifting requirement. 

The much larger size of future high-capacity batteries, combined with fire safety concerns, will drive these stacked application batteries out of basements and outside the building. 

The outdoor location and the electrical and fire safety, seismic and other AHJ requirements mean that tomorrow’s critical power battery systems will look a lot like best practice for outdoor BESS systems. 

Power conversion 

The single-purpose, uni-directional charger of yesteryear will give way to power conversion that can charge, support loads, and export back to the grid. It will be capable of reversing direction (i.e. charger to inverter, and vice-versa) quickly, and follow both site commands and external dispatch signals from a virtual power plant operator or other grid source.  

Fast recharge speed will become an economic requirement. The former 24-hour recharge requirement typical of standby battery sites could shrink to four hours to exploit unpredictable opportunities of negative electric power prices. This sixfold faster charging requirement combined with batteries ten times larger than today’s standby batteries means the capacity of tomorrow’s power conversion could be sixty times greater than the power of today’s stationary chargers. Today’s 10 kW charger becomes tomorrow’s 600 kW bi-directional power converter.

Because the concept of negative prices for electricity is a mind-bender, an analogy to explain the much larger size of power converters is in order. Imagine for a moment that food is super expensive most of the day. For a few hours most days, however, you’re paid to eat. You take breakfast and lunch in short succession – because this deal only lasts a few hours. Plus, the more you can eat during the short “pay to eat” window, the more money you earn.  

The motivation is to eat a lot in a short period. That’s exactly what the big power converter does: delivers a lot of power into a big battery in a short time.  

To handle the much higher power, operating voltage must increase to maintain reasonably sized conductors. A higher storage bus, at least 800 volts DC, enables reduced wire cross section and circuit breaker ratings. Legacy 250, 125, and 48 V buses must then be fed through DC-DC conversion. Isolation, protection, and low-voltage disconnect behavior must be clearly defined. 

Meanwhile physical design should be for availability and service. Modular power stages allow N+1 operation and replacement under lockout without dropping the plant. Technicians will need adequate clearances, lifting paths, and access they need to do the job safely. The shift to higher voltages and power demands new operator training in high energy and arc flash hazards.

A quick ​​sizing example to align expectations 

Suppose a site currently has 100 kWh of standby storage maintained at float with a 10 kW charger. To support both a one-hour ride-through reserve and daily four-hour time shifting, battery capacity is increased to 500 kWh, with 30% capacity (150 kWH) left in reserve in case of blackout. This leaves 350 kWH for time shifting. To recharge 350 kWh in four hours, the converter needs roughly 90 kW at the DC bus after efficiency. Allowing for conversion losses and margin, a 110 to 120 kW converter is a practical minimum. If the site wants to exploit two-hour negative-price windows, double that power. This is why many facilities end up with both larger batteries and much higher conversion power than legacy systems. 

This is more than idle speculation. Forces are already in motion that, by the end of this decade, will result in electricity shortages. Blackouts will become more frequent, longer, and more widespread than ever. The price of electricity will reach staggering heights, while at other times, businesses will be paid to consume. As grid disruptions and volatile prices affect bottom lines, many C&I enterprises will add or repurpose stationary batteries. 

If “standby” power is going to work every week, it will look quite different from the equipment we use today. Storage must cycle, conversion must move power both ways, and operations must treat on-site power not just as an insurance policy, but also as a cash-generating asset. 

The good news 

While predictions of growing grid instability and looming blackouts are alarming, the very industries demanding the most from our grids are actually in the best position to support the grid. Commercial and industrial facilities have at their disposal an incredibly powerful resource: standby power. According to an address at the EGSA 2025 Spring Conference, the combined capacity of C&I standby generators installed in the United States is between 130 and 200 gigawatts, with about 45 gigawatts of that capacity located at hyperscale data centers. 

Properly harnessed with the surrounding infrastructure outlined above, this installed base could replace retiring dispatchable generation faster than any other option, with minimal upgrades to distribution. 

For us in the on-site power industry, the old ways of thinking are facing unprecedented challenges. As an industry, we will need to seize the opportunities this presents – or make way for those who do. 

This concludes Part 2. If you need the case behind these recommendations, see Part 1, which explains why load growth, plant retirements, and aging assets are pushing critical power into daily service in the first place. 

For deeper explanations, examples, and references behind these requirements, see the paper Bill delivered at Battcon 2025.  

---

William Kaewert is Strategic Advisor and Board Member of Colorado-based Stored Energy Systems LLC (SENS), an industry leading supplier of non-stop DC power systems that are an essential part of the nation’s critical infrastructure. SENS products provide non-stop power that enable 24/7 operation of the power grid, energy production, data centers, health care facilities, the financial system and other services that sustain modern life. Mr. Kaewert received his BA in history from Dartmouth College and MBA from Boston University. 

He has served on the board of directors of several economic development organizations and the Electrical Generation Systems Association (EGSA). He is an active member of InfraGard, a public/private partnership of private industry and the FBI to protect United States critical infrastructures from deliberate attack.

Bill co-founded Resilient Utilities Now, a non-profit working to improve US resilience against long-duration electric system failures. Bill has in the past served in other roles related to power system resilience, including director and Chairman of the Board for the Foundation for Resilient Societies a NH-based non-profit.