Data Centers

Data centers could account for 44% of U.S. electric load growth through 2028 and consume up to 9% of the country’s power supply by 2030, causing concerns over their impact on U.S. power availability and costs. Up to 40% of data center electricity use goes to cooling, according to the National Renewable Energy Laboratory, thus greater cooling efficiency is of great interest as one strategy for reducing energy consumption. Cooling is also integral to data center design, influencing how these facilities are developed, built and renovated. 

The second half of 2024 saw several notable announcements related to data center cooling systems, which protect high-performance processors and servers, enabling the advanced computations that artificial intelligence requires. In December, Microsoft and Schneider Electric separately released designs for high-efficiency liquid cooling systems to support increasingly powerful AI chips. Microsoft’s water-based design operates on a closed loop, eliminating waste from evaporation, while Schneider Electric’s data center reference design uses a non-water refrigerant. Earlier in 2024, Vertiv and Compass Datacenters showcased their “first-of-a-kind” liquid-air hybrid system, which they expected to deploy early this year.

Here’s what trends and developments data center cooling experts say they’re watching for 2025 and beyond.

Two-phase liquid cooling will break into the mainstream

Most data center professionals say they’re dissatisfied with their current cooling solutions, according to AFCOM’s 2024 State of the Data Center Industry report. Thirty-five percent of respondents said they regularly make adjustments due to inadequate cooling capacity, and 20% said they were actively seeking new, scalable systems.

Many data center cooling experts predict data center developers and operators will increasingly turn to two-phase, direct-to-chip cooling technology to improve cooling performance. These systems toggle the working fluid — typically a non-water refrigerant — between liquid and vapor states in a process that “plays a pivotal role in heat removal,” according to direct-to-chip liquid cooling system designer Accelsius.

2025 will be a “year of implementation” for two-phase systems as data center professionals get more comfortable with the technology, Accelsius CEO Josh Claman said in an interview. More sophisticated data centers with higher computing needs are more likely to seek out two-phase cooling, Claman said.

Traditional air cooling reaches its physical limit at server rack densities of about 70 kilowatts, the benchmark for state-of-the-art AI training facilities today, said Sarah Renaud, vice president of consulting services at ENCOR Advisors, a commercial real estate firm that works with data center clients.

Because future racks will be even denser, “two-phase is the future,” Renaud said. “It can handle higher power densities and heat fluxes, meaning it’s better-suited for handling AI workloads.”

Hybrid cooling will expand, but supply chain risks loom

Two-phase immersion cooling provides a lower 10-year total cost of ownership for data center operators than DTC or single-phase immersion cooling, according to a March 2024 study by Chemours, the Syska Hennessy Group and cooling system designer LiquidStack. But its high upfront costs, the long operational life of legacy cooling systems and variable cooling needs within individual data centers mean two-phase will continue to coexist alongside other technologies for some time, experts say.

“Almost no new [data center] builds will be exclusively air-cooled nor exclusively liquid [because] not all applications require intense liquid cooling — think of archived data that is rarely accessed versus generative AI,” Renaud said. “You can cool those [less demanding] racks more cost-effectively with air.”

Microsoft’s closed-loop, water-based cooling system “appears to align with an incremental strategy” that supports its near-term needs while “allowing [its] infrastructure to readily pivot to accommodate advanced cooling technologies like direct-to-chip two-phase when the time comes,” said Nick Schweissguth, director of product and commercial enablement at LiquidStack. 

But data center operators’ hybrid cooling plans could be complicated by supply chain issues that could be made worse by anticipated Trump administration tariffs, Schweissguth said. Direct-to-chip coolant distribution units, which keep processors bathed in fluid, are particularly at risk, he noted.

With CDU demand set to surge in 2025, “companies vying to capture the direct-to-chip market will ultimately prevail based on their ability to produce at scale and build bulletproof relationships with suppliers,” Schweissguth said. 

Building and system design will evolve to enable 24/7 uptime

Operators expect far more out of state-of-the-art AI data centers than they did from previous generations of these facilities, said Steven Carlini, vice president of innovation and data center at Schneider Electric.

Whereas earlier facilities might have variable workloads averaging 30% or 40% of total processing capacity, AI facilities typically run at 100% capacity for weeks or months when training models, necessitating more rugged and redundant design, Carlini said. 

“It takes the variability out of the equation, but you have to be very sure you design the cooling system to support that,” he said.

Carlini described a near future in which higher rack-power densities require heavier cooling infrastructure, which creates additional physical demands in data center design. The designs his team has worked on recently, for example, involve “huge” pipes with “big steel cages over the supercluster” or two-story floor plans, with the first level flush on a concrete slab to handle the added weight.

“All that water has to go somewhere,” he said.

“Slow but steady” retrofit activity ahead

Retrofitting an operating data center to accommodate more powerful processors is a big technical and logistical challenge that leads some to conclude that it’s easier to build new, Accelsius’ Claman said. 

But new buildings are significantly more resource-intensive, complicating corporate sustainability goals, he noted. And existing data centers often have more robust power supplies. “That’s why they are where they are, and it’s not easy for them to move,” he said.

The majority of an operating data center’s asset value lies within its power supply and infrastructure, such as electrical, plumbing and other technical systems, according to JLL’s 2025 Global Data Center Outlook. These assets are particularly valuable given the challenges of securing power for new developments. Thus, retrofits such as transitioning existing data centers to liquid cooling will “be a viable solution and an opportunity to increase asset value,” JLL’s outlook says. 

Meta is transitioning its existing data centers to liquid cooling “because they say they ‘have to,’” while colocation giant Equinix said in December 2023 that it would expand liquid cooling to 100 of its data center facilities, Renaud noted.

Claman predicted a “slow but steady” pace of retrofits and “a more balanced conversation” around their benefits. Schneider Electric is betting on this trend as well, recently partnering with Nvidia on the release of three retrofit reference designs for data center operators looking to boost performance without redesigning their facilities from scratch. 

The rapid increase in computing power means data centers on the bleeding edge today may rapidly fall behind, further complicating the already formidable challenge of designing facilities with both air and liquid cooling infrastructure, Carlini said. 

“Ten years ago, you’d try to [design data centers with] more capacity than you need and grow into it, but now you don’t know what [power] density you need to build at,” he said.

Facilities in Northern climates might get an edge

Air provides 20% to 30% of the cooling load, even in newer data centers, according to Carlini. That’s driving efficiency-minded developers to site more facilities in “the attic,” the informal industry term for cooler Northern regions, Renaud and Claman say.

“The market talks a lot about a ‘free cooling zone’” in the Northern United States, Northern Europe and Canada, Claman said. 

In cooler weather, energy use for air cooling systems could drop by as much as 95%, according to Renaud. “We are seeing a trend of hybrid colocation strategies in which data that does not require frequent access can be stored in more remote and colder locations,” leaving higher-access-frequency facilities to operate in warmer, more established data center hubs like northern Virginia, she said.

Cold-climate sites also are less likely to need water-hungry evaporative cooling systems, which are common in warmer, drier climates and have raised concerns around data centers’ environmental impacts, Claman said. He predicted a move toward closed-loop cooling systems that can take advantage of seasonal free cooling.

“There is a lot of scrutiny around emptying aquifers to cool data centers,” he said.

The global data center market is forecasted to expand at a 15% compound annual growth rate through 2027 and has the potential to grow at a 20% rate in that time, according to JLL’s 2025 Global Data Center Outlook. 

The firm’s outlook notes four megatrends that are set to define the market this year: artificial intelligence, power needs, liquid cooling and capital demand. 

Rapid progress in semiconductor technology has led to increasingly more dense and high-powered chips, with rack densities expected to grow from 41 kilowatts now to 130 kW in the next generation to about 250 kW in the future, per the report. Nvidia’s latest artificial intelligence chips consume up to 300% more power than their predecessors, for example, with industry forecasts suggesting that global data center energy demand will double in the next five years, JLL says.

The power demand for AI training facilities is also substantial, with new projects “regularly requesting 1 gigawatt of power or more,” or the same amount of electricity used to power 800,000 U.S. homes over the course of a year. Due to the immense power requirements, and lack of energy availability in many regions of the U.S., data centers tend to cluster together, JLL says, and they can represent a large portion of electricity consumption in certain regions.

Data center power demand is threatening existing power grids and pushing data center stakeholders to look to small modular nuclear reactors to address power availability challenges, JLL says. AI’s surging demand is also driving a battle for energy efficiency, sparking a new wave of innovation that includes more energy-efficient chip architectures.

But AI represents only one component of data center demand, the report states. Traditional, low-intensity workloads such as data storage and cloud-based applications still make up the majority. 

“Even optimistic adoption scenarios suggest that AI workloads will represent less than 50% of data center capacity in 2030,” JLL says. “Therefore, older data centers are not in danger of becoming obsolete. Instead, they will remain a valuable component of the global data center ecosystem.”

The immense demand growth in the sector means that it is “imperative that all existing capacity is retained while additional facilities are constructed to meet new demand,” JLL says. 

The majority of a data center’s asset value lies within its power supply and infrastructure, such as electrical, plumbing and other technical systems. These are especially valuable given the challenges of securing power for new developments, the report notes.

“For these reasons, data centers will not become obsolete within 10 years of being constructed,” JLL says. “Instead, with proper maintenance and investment, data centers will appreciate in value over time.” 

Liquid cooling systems have become the default in new construction to meet advancing computing and power density requirements, JLL says. Retrofits, or transitioning existing facilities to liquid cooling infrastructure to accommodate high-density workloads, “will be a viable solution and an opportunity to increase asset value,” according to the outlook. 

Immersion cooling is expected to become a common thermal management strategy as GPUs push above 150kW per rack. Broad implementation of immersion cooling is still a few years away, however, with the cooling technology introducing new challenges in structural design, the report states. 

Data center development is expected to require roughly $170 billion in construction lending or permanent financing in 2025, JLL says. It expects data centers to attract more investors, although this remains a niche sector, it notes.  

Sponsored content

By ifm efector

Demand for large language models that power generative artificial intelligence are outpacing current cloud capacity, constrained by the construction of new data centers. The accelerating growth trajectory of new data centers has captivated tech company executives as they search for new ways to accelerate construction and growth without sacrificing quality or security. Simultaneously, the energy consumption of data centers is growing unsustainably at current levels. This new era of technological progress needs industrial equipment that is as agile and flexible as the market it serves. The technology to support the growth trajectory of data centers, help reduce energy consumption, and futureproof industrial equipment already exists.

Building the technological foundation to accelerate data center deployment

Reducing deployment down to weeks drastically changes the competitive edge of cloud computing suppliers. Many machine builders still rely on analog controls from a previous generation to run cooling systems, but legacy controls are not equipped to handle the amount of data and computation needed to monitor data centers in real time. Because engineers are comfortable with the ubiquitous technology, they continue to build around Programmable Logics Controllers (PLCs) and create patchwork solutions to overcome their limitations. But, the fact is they were never designed to perform complex computational analysis. Engineers need better technology designed to meet modern expectations of machine data and functionality. The future of data center construction needs industrial equipment that simplifies:

• Implementation of equipment
• Interoperability of hardware
• Machine-level data collection
• Setup & programing

As data centers power cutting-edge technological advancements, the technology that powers their equipment needs to innovate too. Analog controls were invented in an era when less information was readily available or expected. Digital communication protocols today offer smarter options for industrial equipment with a design that provides greater connectivity and diagnostic information within a secure environment.

Improving access to machine data

At the core of modern technological progress is the power of more data. With smarter sensors and standardized connectivity, monitoring systems are embedded in automated cooling system equipment. Modern data centers can implement remote monitoring and receive immediate diagnostic information at any hour by collecting more source data with simplified hardware. Central to this system architecture is IO-Link: a bi-directional digital communication technology between sensors and control systems. Established as an open consortium with the commitment of 485 member companies, this revolutionary technology changes what is possible with connected industrial equipment. To understand how this can accelerate data center deployment and equipment optimization, consider five key benefits of IO-Link.

Faster machine commissioning. IO-Link eliminates manual programming of equipment when commissioning a new data center. Field-mountable IO-Link hubs and plug-and-work smart sensors offer greater flexibility and simplify reconfiguration, enabling quick changes and reducing manual errors. Simply bring in equipment, connect components with quick-disconnect cabling, and start collecting data. This modular approach reduces the timeline to commission a new facility from months to days, regardless of facility size. Standardization of IO-Link among industrial suppliers improves interoperability, reduces friction during setup and programming, and improves supply chain resiliency by reducing reliance on proprietary technology.

Increased data accuracy and diagnostic information. With a fully digital signal, IO-Link increases the accuracy of sensor data compared to the conversion losses of traditional analog scaling and sends multiple process values over simplified wiring. It enables advanced diagnostics to detect problems earlier and perform comprehensive analysis that prevents equipment downtime, outperforming the limitations of analog monitoring systems. Smart sensors perform the calculations at the edge and send results immediately to any Data Center Infrastructure Management (DCIM) system.

Increased access to machine health and process data. With a continuous flow of real-time data, data centers can monitor and track the most critical machine parameters that signify condition and performance. The additional process data enables companies to build digital twins of each machine where their real-time data trains statistical models to predict machine failures. IO-Link increases the raw source data available to train these advanced models and optimize machine performance once a facility is fully operational.

Decentralized energy optimization with machine-level monitoring. Cooling systems must respond to real-time demands by moderating coolant temperature and flow accurately to improve system efficiency. Monitoring these critical points at the machine level ensures optimal system performance, mitigates threats of thermal throttling, and reduces energy consumption. Traditional analog signals are susceptible to interference, introducing errors that lead to improper cooling and inefficient energy usage. The digital signal of IO-Link minimizes the potential for interference and enables precise, repeatable measurements.

Future-ready equipment that operates at peak performance. Today, many data centers run below optimal performance, because of concerns about damaging servers due to inefficient cooling systems. With real-time digital data, facilities can optimize cooling systems to power AI models on their servers with higher performance. Transitioning to systems with digital technologies better prepares data centers to calculate cost per search or kilowatt-hours (kWh) metrics in the future by providing more granularity of their operational data. Instead of monitoring energy consumption by aggregate or zone level, coolant monitoring and energy monitoring by the individual cooling unit are possible. Field-mountable IO-Link provides an embedded method to analyze consumption data at the server level. Though data centers may not utilize this data at first, the access is part of the machine design and ready for the future at no additional cost.

Preparing for the future

Data centers can accelerate deployment and building automation solutions by adopting readily available technology to connect cooling equipment and increase access to machine-level data. IO-Link technology is not just another BUS system, but foundational technology for the next era of connecting, monitoring, and predicting conditions of industrial machinery. Of any industry, data centers understand the potential and have the capabilities to harness data to create the future. It is time for their technology stack and equipment to match that ambition.

What started as a relatively niche sector a few years ago has transformed into a powerhouse of construction activity, with no signs of a slowdown this year.

Data centers continue to drive significant growth in nonresidential construction planning, said Sarah Martin, associated director of forecasting at Dodge Construction Network. These projects contributed to a 19% increase in planning activity since December 2023, compared to just 5% growth without their inclusion, according to Dodge.

That momentum shows no signs of fading.

J.P. Morgan estimates spending on data centers could add between 10 to 20 basis points to U.S. economic growth in 2025 and 2026, according to Reuters. Big tech companies are a major reason for the surge, doubling down on data center investments in order to meet growing artificial intelligence demand.

For example, President Donald Trump recently announced Stargate, a joint venture between OpenAI, Softbank and Oracle. The JV plans to invest $100 billion in artificial intelligence infrastructure, with the potential to scale up to $500 billion by 2028. Each data center facility will span about 500,000 square feet, according to Stargate.

Here are some other major data center projects that top tech giants plan to build in 2025:

Amazon

Amazon is going big on data centers, with plans to invest $150 billion in the coming 15 years on its infrastructure to handle the expected demand for artificial intelligence and other cloud computing needs, Bloomberg reported. 

The tech giant first started opening data centers in central Ohio in 2016, and recently announced a $10 billion plan to further expand its cloud computing infrastructure in the state, per a December news release from Ohio Gov. Mike DeWine. The new investment follows a $7.8 billion plan announced in 2023, including work in New Albany, Ohio.

Those planned projects include a new $2 billion data center campus in Sunbury, Ohio. Amazon’s facility will occupy about 200 acres in an industrial park in the city, according to WBNS10. Construction of the first 450,000-square-foot building is expected to begin in January 2028 and be complete by December 2034, Data Center Dynamics reported.

Google

The Mountain View, California-based multinational corporation plans to continue its aggressive expansion into data center construction, with a number of high-profile projects slated for 2025, according to Devon Smiley, communications manager of technical infrastructure at Google.

For example, Google broke ground in July 2023 on its first data center in Arizona. The site will span 750,000 square feet, with the first phase due in July, according to Data Center Dynamics.

Plans feature a $600 million facility to support some of the firm’s signature services, such as Gmail and Google Cloud. The Mesa project will tie into the establishment of a new Phoenix cloud region, expanding Google’s digital infrastructure footprint in the area. No general contractor has been confirmed yet, said Smiley.

In the Midwest, Google broke ground in April 2024 on a $2 billion data center in Fort Wayne, Indiana. The facility will power its global AI and cloud infrastructure, according to the company. The Indiana Economic Development Corp. will give Google a 35-year data center sales tax exemption for the first $800 million invested, with options to extend up to 50 years. Atlanta-based Holder Construction is the general contractor on the project, according to local outlet 21Alive.

Meanwhile, in Nebraska, the tech giant’s newest date center project in Lincoln is under construction on roughly 580 acres, according to the Nebraska Examiner. That project complements a separate $750 million facility under construction in Omaha, according to a government release. Once complete, Google will have four data center campuses in the state, according to the company. Google did not disclose the projects’ general contractors.

Meta

Meta, the parent company of Facebook and Instagram, has big plans for data center construction. On Jan. 24, CEO Mark Zuckerburg called 2025 “a defining year for AI” in a Facebook post, and claimed the company would construct a data center large enough to “cover a significant part of Manhattan.” Zuckerburg said Meta will bring on about 1 gigawatt of computing power in 2025, and plans to invest heavily in growing the firm’s AI teams.  

That ambtious goal aligns with previously shared plans. In 2024, the firm announced new data center campuses in:

• Jeffersonville, Indiana, awarded to Turner Construction.
• Rosemont, Minnesota, awarded to Mortenson, according to Finance and Commerce.
• Montgomery, Alabama, awarded to Hensel Phelps. 
• Cheyenne Wyoming, awarded to Portland, Oregon-based Fortis Construction. 
• Aiken, South Carolina (no contractor announced). 

Each of those campuses represents an $800 million investment in the region, Meta claims.

In December, the Menlo Park, California-based tech giant announced a larger investment near Monroe, Louisiana. The custom-designed 4 million-square-foot campus will be Meta’s largest to date, the company said. When finished, the data center will represent over $10 billion in investment and the company anticipates over 5,000 construction workers will be on site at peak construction. 

The social media giant has tapped New York City-based Turner, Redwood City, California-based DPR and Minneapolis-based Mortenson to build the multibillion-dollar project. No completion date has been announced.

Microsoft

One of the largest and most recognizable tech companies in the world has been busy on the data center front. Redmond, Washington-based Microsoft has announced plans to pump more than $80 billion into cloud data centers, with half of that money going to facilities in the U.S.

“AI promises to drive innovation and boost productivity in every sector of the economy,” said Brad Smith, vice chair and president of Microsoft, in a Jan. 3 company blog post. “The United States is poised to stand at the forefront of this new technology wave, especially if it doubles down on its strengths and effectively partners internationally.”

The firm, which is also one of the largest investors in ChatGPT parent company OpenAI — with nearly $14 billion pumped into the company so far, according to CNBC — has jump-started its plans. It’s building a $3.3 billion data center on the site of the former Foxconn site in Mount Pleasant, Wisconsin, which it tapped Chicago-based Walsh Construction to lead. 

It’s also planning to invest $1 billion on three data centers in Licking County, Ohio, according to Data Center Dynamics. Burnsville, Minnesota-based Ames Construction will build one of the three data centers, a $420 million project in New Albany, the tech firm said in a construction update blog. Two other data centers worth $1 billion in West Des Moines, Iowa, and La Porte, Indiana, are also in the works.

Oracle

While Oracle CEO Larry Ellison made news with the White House announcement unveiling the Stargate initiative on Jan. 21, he had ramped up his commitment to building massive, power-hungry data centers long before that.

The co-founder and chief technology officer of Austin, Texas-based tech giant Oracle told attendees of Oracle Cloud World 2024 in September that the firm now has 160 public and private data centers, according to Data Center Frontier. 

Additionally, he told analysts the company was working on a data center requiring more than a gigawatt of electricity that would be powered by three small nuclear reactors, according to CNBC. While he didn’t reveal the exact location of the facility, he said it was in a region where permits had already been approved to build the reactors as a power source.

Those announcements followed the tech firm’s pledge from last year to spend $10 billion on data center expansion in 2025. At the time, Ellison told investors that an AI-focused data center the firm is building in Salt Lake City is so large, eight Boeing 747s would fit nose-to-tail inside it. 

“We’re bringing on enormous amounts of capacity over the next 24 months,” Ellison said.

Wesco International announced plans to acquire data center facilities management service provider Ascent for $185 million, noting that the deal will enable it to provide engineering expertise and professional services spanning the entire lifecycle of data centers.  

Ascent delivers solutions for mission critical data center operations and infrastructure, from “next-generation AI data center design and builds all the way through the operations and maintenance lifecycle,” Wesco said in an Oct. 31 news release. Services include liquid cooling design and implementation, repairs of critical systems, emergency services and management of third-party maintenance workflows, per the release.

Ascent also provides an integrated client dashboard through its Navigator platform designed to support critical data center infrastructure and facilities, Wesco said. The data center company has $115 million in trailing 12-month sales and a three-year sales compound annual growth rate of 30%, according to the release. 

“This acquisition will allow us to further extend our end-to-end service offerings — including advanced liquid cooling solutions — in collaboration with our supplier and contractor partners to meet the demands of the rapidly growing data center market," Bill Geary, executive vice president and general Manager of Wesco Communications and Security Solutions, said in a statement. 

Overall data center vacancy rates for primary markets fell to a record low of 2.8% in the first half of 2024, from 3.3% in the previous year, with nearly 80% of more than 3.87 gigawatts under construction in primary markets already preleased, according to a data center trends report by CBRE. 

The agreement Wesco has signed to buy Ascent highlights a trend of record profits and boosted earnings for many service, systems and technology providers that have continued to lean into the burgeoning data center market. 

CBRE, for example, has scaled its efforts to capitalize on the market, acquiring data center construction and operations company Direct Line Global in June. During CBRE’s third-quarter earnings call in October, its CFO Emma Giamartino noted that the acquisition would enhance CBRE’s capabilities in a rapidly expanding data center market, valued at approximately $30 billion, according to its estimates. 

Meanwhile, JLL hired Kristen Vosmaer, former vice president of global operations at development firm GLP, to lead its Work Dynamics data center business. This move followed JLL’s agreement in May to acquire SKAE Power Solutions, a New-York based data center design and engineering firm that will enable the real estate services firm to provide new technical services and add depth to its existing offerings, the firm said.

In June, Johnson Controls formed a data center-focused organization, Global Data Center Solutions, focused on providing integrated solutions to data center customers. Johnson Controls’ data center vertical drove growth of more than 20% in its applied HVAC and controls business in its fourth fiscal quarter of 2024, facilitating an organic 10% year-over-year rise in sales, CFO Marc Vandiepenbeeck said on an earnings call Wednesday. 

Data center exposure is currently about 10% of Johnson Controls’ revenue and “continues to grow faster than the rest of the portfolio with very solid double-digit growth sequentially at least for the foreseeable future,” Vandiepenbeeck said. 

Honeywell and Schneider Electric also attributed strong earnings growth to their work in the vertical. Honeywell saw a 25% year-over-year growth in its building solutions segment, due in part to strong demand for data centers, CFO Greg Lewis said on an Oct. 24 earnings call. 

To capitalize on the data center momentum, Schneider Electric announced an agreement to buy a 75% stake in data center cooling firm Motivair. The deal, slated to close in 2025, will “enhance our thermal management abilities to ensure we’re ready to meet all data centers,” Schneider Electric CFO Hilary Maxson said during the company’s third-quarter earnings call last week, pointing to an accelerated growth in the data center market. 

HVAC companies are increasing their focus on data centers too. Trane Technologies reported a backlog of $7.2 billion, up $300 million from the end of 2023. The company attributed this growth in its commercial HVAC segment, which accounted for roughly 90% of its backlog, to particular strength in its data center vertical. 

Another HVAC firm that saw significant growth figures during is Modine. The company ended its second 2025 fiscal quarter with a 102% year-over-year jump in data center sales, amid strong demand from hyperscale and colocation customers across North America.

Wesco said its acquisition of Ascent is expected to close in the fourth quarter, following customer regulatory approval.

Data center operators, hyperscalers, electricity providers, local regulators, municipalities and real estate firms must collaborate on how best to balance environmental, business and societal goals, according to a Deloitte report released Tuesday. 

That collaboration should involve determining potential strategic colocation needs — where a data center company rents computing and server resources to one or more companies — as well as needs for cooling, recycling, and heat and wastewater managment, according to the report. 

Small and midsized organizations that may find it challenging to harness generative AI data center capacity should work with specialized data center operators and cloud service providers to meet their needs, Deloitte says. In turn, data centers can actively track usage and availability for potential opportunities and demand pockets to help deliver near-term colocation services, per the report. 

As data center energy consumption rapidly accelerates to meet power-intensive generative AI demand, many companies are looking to use a combination of energy-efficient data center technologies and more carbon-free energy sources to reduce environmental impact, the report says. Although data centers are expected to make up only about 2% of global electricity consumption, or 536 terawatt hours, in 2025, annual energy use could roughly double to 1,065 TWh by 2030, Deloitte’s analysis suggests.

In the U.S., data centers are anticipated to represent 6%, or about 260 TWh, of total electricity consumption in 2026, Deloitte says, citing estimates from the International Energy Agency. The nation’s data center market has doubled in size in four years, with vacancy rates hitting a record low of 3% in the first half of 2024, according to a report by JLL released in August. 

The surge in demand for data center capacity comes “against the backdrop of a multi-decade power industry transformation,” as electric companies build, upgrade, and decarbonize electric grid infrastructure, and digitalize systems and assets,” Deloitte says. Utilities are also working to fortify assets against increasingly severe weather and protect networks from rising cybersecurity threats, according to the report. 

As demand for high-power computing capability intensifies, a significant price gap between new data centers and legacy facilities has also emerged, with existing data centers lacking the infrastructure to support the demanding workloads required, according to a CBRE report released in August. 

Liquid cooling, despite its potential to support more data-intensive server racks, potentially eliminating the need for chillers, is still at a nascent stage and yet to be widely adopted or integrated into AI data centers globally, Deloitte notes, citing a March 2024 article on the website of engineering and consulting firm Ramboll. 

The urgency and geographic concentration of data center demand, along with the need for round-the-clock, carbon-free energy, exacerbate challenges for technology companies and electricity providers, compounded by new demand from electrification, manufacturing and other sources, Deloitte says. 

To expedite the jump to carbon-free energy sources for data centers, major tech industry players like Amazon and Google continue to be aggressive in boosting their use of renewable energy-sourced electricity and establishing power purchase agreements. The technology sector “consistently dominates” U.S. corporate renewable procurement and accounted for more than 68% of the nearly 200 deals of associated capacity tracked between February 2023 and February 2024, Deloitte’s report notes, citing its analysis of information gathered from multiple reports from S&P Global Market Intelligence, published during March and August. 

The 24/7 carbon-free electricity that many tech companies seek can be hard to come by, however, “especially in the short term,” with many electricity providers announcing plans to build carbon-emitting natural gas-fired power plants, Deloitte says. “This could potentially make it more challenging to meet utility, state, and even national decarbonization targets,” the report notes.

Deloitte suggests that stakeholders form strategic partnerships to serve local and cluster-level AI data center needs. 

The technology industry should also work closely with electricity providers to determine how to fund and scale new energy technologies, “which is a vital step to bring more clean energy to the grid,” the report states. 

Deloitte also suggests optimizing generative AI uses and shifting processing to edge devices, which can help save network and server bandwidth, while supporting applications where response times are critical or where sensitive data is involved.

Amazon Web Services on Monday announced new data center components designed to support advanced, high-power chips and existing infrastructure. 

The improvements include simplified electrical and mechanical systems. The simplified systems can reduce the potential number of racks impacted by electrical issues by 89%, driving reliability and reducing overall energy consumption, according to a news release. The new modular components are designed to support liquid-to-chip cooling systems and high-density AI workloads while being “flexible enough to accommodate a wide range of other hardware types,” AWS said. 

“These data center capabilities represent an important step forward with increased energy efficiency and flexible support for emerging workloads,” Prasad Kalyanaraman, vice president of infrastructure services at AWS, said in a statement. “But what is even more exciting is that they are designed to be modular, so that we are able to retrofit our existing infrastructure for liquid cooling and energy efficiency to power generative AI applications and lower our carbon footprint.”

The growing demand for data centers, especially high-intensity facilities that can handle the power required for generative AI applications, comes as Amazon works to reduce its carbon emissions and meet internal and external decarbonization targets, including reaching net-zero carbon emissions by 2040. The company cut its absolute carbon emissions 3% year over year in 2023 and reduced its carbon intensity by 13% in that time. 

In its 2023 Sustainability report, released in July, Amazon attributed the overall decline in its absolute carbon emissions to an 11% year-over-year decrease in scope 2 emissions, or emissions from electricity, and a 5% decrease in scope 3 emissions, or indirect and supply chain emissions. Its increased use of renewable energy-sourced electricity and its purchase of additional environmental attributes such as renewable energy credits contributed to the reduction in its scope 2 emissions, the company’s report states.

.The two most important ways Amazon can lower electricity-related carbon emissions are by improving energy efficiency and transitioning to carbon-free energy, according to Amazon’s sustainability report. These strategies include on-site renewables, refrigerants with low global warming potential and alternative fuels for backup power sources and data center cooling, it says.

With electricity passing through multiple conversion and distribution systems before reaching IT equipment in a data center, each step naturally introduces inefficiency, energy loss and potential failure points, AWS said in the release. 

“We saw an opportunity to really pull those things apart and simplify … the whole system. We assume that any component in the data center is going to fail,” Kevin Miller, vice president of global data centers at AWS, said in an interview. “And our goal then is to make sure that an individual component failure can never bubble up to become something that’s customer-visible.” 

The new components include upgraded cooling systems that are expected to reduce mechanical energy consumption by up to 46% compared with its previous design during peak cooling conditions, without increasing water use on a per-megawatt basis, according to the release. The systems will integrate with both air and liquid cooling, “to provide maximum performance and efficiency at the lowest cost, whether running traditional workloads or AI models,” AWS said.

“We are already starting to deploy some of that liquid cooling capacity, but we also know …  that we need to be flexible in meeting the specific needs of different generations over time,” Miller said. “It's going to be a transition where we're going to continue to support both air and liquid in our buildings for a long time, and we need to have components that are very modular, so that we can right size [systems] and really make it efficient for just the amount of liquid capacity we need.”

“For our customers and for others, it can seem like a data center is a data center,” Miller added. “The reality is that they are very different.” 

A new power shelf will efficiently deliver data center power throughout racks, reducing conversion losses. AWS said the engineering innovations it has developed in its power delivery system will support a six-fold increase in rack density over the next two years and another three-fold increase in the future. Taken together, the energy-efficient system design and power delivery systems will enable AWS to deliver 12% more compute power per site for customer workloads, it said. 

Design changes include a new single-sided cooling system, a reduction in cooling equipment and the introduction of liquid cooling capabilities, AWS said. Additionally, backup generators can run on renewable diesel, reducing GHG emissions by up to 90% over the fuel’s lifecycle when compared with fossil diesel. A control system is also being rolled out across AWS’ electrical and mechanical devices that can standardize monitoring, alarming and operational sequences, AWS said.

Intersect Power is partnering with Google and TPG Rise Climate, a subsidiary of private equity firm TPG, to drive up to $20 billion in renewable power infrastructure investment to power new data centers and scaling clean energy generation by the end of the decade, the companies said last week.

The partnership’s “power-first” approach will involve colocating data centers with Google as an anchor tenant near “large amounts of high capacity factor, low-cost, clean electricity and added battery storage [to] achieve high percentages of renewable energy,” while reducing the need for new transmission infrastructure, the companies said in a Dec. 10 news release.

“The scale of AI presents an opportunity to completely rethink data center development — by colocating them where possible with the grid-connected carbon-free energy that keeps them up and running,” Amanda Peterson Corio, global head of data center energy at Google, said in a statement.

By 2030, Google aims to achieve net-zero emissions across its operations and value chain and reduce its combined Scope 1, 2 and 3 absolute emissions by 50% from 2019 levels, the company says.  The tech giant also plans to “invest in nature-based and technology-based carbon removal solutions to neutralize our remaining emissions,” according to its sustainability website.

Amid an artificial intelligence boom driving a rapid increase in data centers’ power consumption, however, Google’s total greenhouse gas emissions increased 13% year over year to 2023, according to its 2024 Environmental Report released in July.

For years, Google and other tech giants have offset GHG emissions from electricity consumption and other activities through direct and virtual power purchase agreements with wind and solar power facilities. More recently, Meta signed an agreement with geothermal energy firm Sage Geosystems to power its data center growth and boost its renewable energy use, while Amazon announced deals with three energy companies to develop and deploy modular nuclear reactors. 

But those technologies are earlier in the commercialization process than wind, solar and battery energy storage and thus unlikely to be widely deployed until the next decade, according to the U.S. Department of Energy’s next-generation geothermal and advanced nuclear commercial liftoff reports. In contrast, the first clean energy project to be developed under the Intersect-Google-TPG partnership is expected to be operational by 2026 and “fully complete” by 2027, the companies said. 

“No one else is [developing] on this timeline,” Intersect Power founder and CEO Sheldon Kimber said in an interview. Kimber declined to describe the scope or location of the first project but said an announcement would be coming “shortly.”

“We expect this partnership to achieve unprecedented scale at our first colocated project, and we have set ourselves on a course to deliver several more large scale colocated data centers and clean energy power plants across the U.S.,” Ed Beckley, managing partner at TPG Rise Climate, said in a statement.

Intersect wants to develop “powered land” with robust electric generation and distribution infrastructure for large data centers in regions with high-quality wind and solar resources, said Kimber, whose company on Tuesday also announced more than $800 million in a funding round led by Google and TPG Rise Climate. 

A one-gigawatt data center colocated with a GW each of wind, solar and battery storage could satisfy up to 80% of its annual power needs from those carbon-free resources and make up the difference with grid electricity, Kimber said.

Rather than placing additional stress on the electric grid by “[demanding] power at the worst possible time,” such as a hot summer evening or cold winter morning, Intersect’s data centers “can actually support the grid” by drawing power from colocated renewables or batteries during those peak periods and relying on the grid only during periods of lower demand, Kimber added.

Even in areas with high-quality renewable resources, data centers may need gas-fired backup generators in the near term to mitigate grid outages or long periods of low wind and sun, Kimber said. As lithium-ion battery prices fall and emerging longer-duration energy storage technologies, like Form Energy’s iron-air batteries, become cost-competitive, however, data center operators “can start to squeeze out reliance on the grid and fossil fuel backup,” he said.

Schneider Electric has introduced a new data center reference design, co-developed with NVIDIA, to support liquid-cooled, high-density AI clusters, amid “urgent energy and sustainability challenges driven by high demand for AI systems,” the company said in a news release Wednesday. 

The company also unveiled its Galaxy VXL uninterruptible power supply, which it describes as the industry’s most compact, high-density UPS. It is designed to support AI-ready data centers and large-scale electrical workloads, offering 52% in space savings compared with the industry average, Schneider Electric said. 

Both solutions aim to benefit data center owners and operators as they deploy energy-efficient, high-density infrastructure to support AI workloads “as sustainably as possible,” Schneider Electric said. 

AI’s projected electricity demand encompasses a wide array of interconnected issues, ranging from infrastructure challenges to supply chain disruptions and socioeconomic concerns, with both global and local data center developments raising significant concerns about grid stress and power capacity, according to a report by Schneider Electric Sustainability Research Institute released this month. 

Despite the rapid growth of data center electricity use, with AI energy demand expected to multiply by at least 10 between 2023 and 2026, “a significant body of evidence suggests that it is possible to implement AI technology at scale in ways that minimize environmental impacts and maximize societal benefits,” the report says.

“By 2027, data center electricity consumption is projected to account for 2.5% of global demand, with the remaining 97.5% spread across industries such as buildings, manufacturing, transportation, and energy," Sean Graham, research director of cloud to edge data center trends at International Data Corporation, said in a statement. "While data centers pursue their own net-zero goals amid unprecedented growth, the real sustainability promise lies in leveraging AI to decarbonize entire value chains across industries. As Schneider Electric and NVIDIA have demonstrated, long-term collaboration and innovation are essential to driving efficiency and sustainability,” Graham noted.

Both innovations form part of its end-to-end, AI-ready data center solutions, which focus on three key areas — developing an energy strategy for the AI era, deploying advanced infrastructure and sustainability consulting, Schneider Electric said. Its energy strategy for the AI era includes supporting companies in their efforts to secure renewable energy and optimizing on-site power generation; providing services like site selection and geographical analysis based on customers’ deployment plans; and enabling on-site power generation through its AlphaStruxure platform, according to its release. 

To deploy advanced infrastructure, the company has also developed a portfolio of high-density, energy-efficient infrastructure systems “for a host of AI requirements exceeding 100 kilowatts per rack,” Schneider Electric said in the release. This includes data center infrastructure components “from grid-to-chip and chip-to-chiller, AI-powered remote monitoring and energy management software, and digital services for lifecycle optimization,” per the release.

The Galaxy VXL UPS is a highly efficient, compact, modular, scalable and redundant 500 to 1,250 kW, three-phase UPS, including enhanced cybersecurity, software and safety features, according to a separate news release issued Wednesday. Schneider Electric says the Galaxy VXL UPS has a power density of up to 1,042 kilowatts per square meter, includes power-protection and saving modes and is compatible with both lithium-ion and VRLA batteries, providing greater uptime and easier serviceability. 

The company says its recently signed agreement to acquire a majority stake in Motivair is expected to enhance its liquid cooling portfolio and strengthen its expertise in direct-to-chip liquid cooling and high-capacity thermal solutions. Additionally, its sustainability consulting business will help customers reach decarbonization goals via tailored sustainability strategies and emissions assessments through its EcoStuxure Resource Advisor, Schneider Electric said in the release.

This end-to-end AI-ready data center strategy supports a science-based approach to “bend the curve” of energy consumption, which can be accomplished by integrating data center infrastructure with AI-driven applications, Schneider Electric said. 

“The energy and environmental impact of AI is growing at unprecedented pace, and it’s paramount we bend the energy curve downward by finding new ways to decarbonize data centers and the digital infrastructure,” Pankaj Sharma, executive vice president of data centers and networks at Schneider Electric, said in a statement.

President Biden on Tuesday signed an executive order to accelerate the development of data centers to enable artificial intelligence, while also ensuring the new facilities are powered by emissions-free electricity.

The order directs the U.S. Department of Defense and U.S. Department of Energy to lease sites for gigawatt-scale AI data centers and power generation facilities, and “to facilitate this infrastructure’s interconnection to the electric grid, fulfill permitting obligations expeditiously, and advance transmission development around federal sites.”

Growth of AI could lead data centers to consume 9% of the United States’ electricity by 2030, according to the Electric Power Research Institute. Biden’s executive order, however, gives more subsidies to big tech companies while sending consumer power bills higher, one clean energy advocate warned.

“We will not let America be out-built when it comes to the technology that will define the future,” Biden said in a statement.

The White House wants to move quickly, expecting that within three years data centers requiring 5 GW of power may be required to train AI models, White House technology adviser Tarun Chhabra said in a call with reporters.

DOD and DOE “will select sites where the private sector can build AI data centers and clean power facilities based on those sites’ accessibility to high-capacity transmission infrastructure and minimized adverse effects,” the White House said. “After selecting sites, DOD and DOE will hold competitive solicitations for proposals to lease these sites for building, owning, and operating large-scale AI infrastructure — all at private expense.”

Developers “will be required to bring online sufficient clean energy generation resources to match the full electricity needs of their data centers,” the statement said.

To support the efforts, the Department of Interior will identify areas for clean energy that can support data centers on DOE and DOD sites while also “enhancing permitting processes” for geothermal projects, the White House said. DOE will take further steps to “promote distributed energy resources, advance siting of clean generation resources at existing points of interconnection, and support the safe and responsible deployment of nuclear energy.”

The White House said the EO prioritizes “full and expeditious permitting of AI infrastructure on federal sites” and calls for federal agencies to identify “further opportunities to support expeditious permitting at these sites, such as by applying or establishing ‘categorical exclusions’ ... for infrastructure that does not significantly affect the environment.”

DOE last year finalized new and revised categorical exclusions for energy storage projects, transmission lines and solar projects, exempting eligible projects from environmental assessments or environmental impact statements.

“These efforts are designed to accelerate the clean energy transition in a way that is responsible and respectful to local communities, and in a way that does not impose any new costs on American families,” Biden said.

U.S. electricity demand could rise 128 GW over the next five years, driven by data centers and manufacturing growth, according to a December report from Grid Strategies. Electricity prices will likely continue to rise, experts say, with increasing demand an important factor.

Ben Inskeep, program director at Citizens Action Coalition of Indiana, was wary of the White House announcement on X.

The administration is “giving energy-guzzling AI data centers preferential access to siting and exclusions from bedrock environmental laws,” he wrote. “Big Tech companies get more subsidies and special treatment. Meanwhile, your utility bill goes up and the climate burns.”

Ener.co’s spray-on coating has significantly improved energy efficiency across 36 of Verizon’s urban data centers in the Northeastern U.S., the company announced this month.

The proprietary carbon nanotechnology coating, which is applied to HVAC condenser coils, has helped Verizon save over $1.28 million in annual energy costs and slash its greenhouse gas emissions by about 1,700 metric tons a year, Ener.co said.

The project achieved a 30-month payback period through energy savings and is projected to save an additional $16 million for Verizon’s data centers over the next decade, according to Ener.co.

Ener.co promotes its Enercoat spray-on condenser coil coating as an alternative to equipment replacements that can restore cooling efficiency and protect condenser coils. 

Ener.co has been working with Verizon for four years, President Patrick Rathje said in a statement. In the third quarter of 2023, Ener.co treated 300 HVAC systems across 36 of Verizon’s data centers, covering more than 14,000 tons of air-cooled cooling equipment, it said.  The Enercoat applications reduced annual electricity consumption by 7.7 million kilowatt-hours, resulting in over $1.2 million in annual cost savings, Ener.co said. 

Building on the success of these results, Ener.co and Verizon are preparing a $30 million nationwide data center HVAC performance upgrade project, according to the release.

“More efficient cooling performance is critical to managing data center energy and climate impacts,” Rathje said.

Ener.co’s other customers include Google, Cushman & Wakefield, Marriott International and NYU Langone Medical Center, according to the company. 

To date, it claims to have helped customers save about 67 million kWh in energy, resulting in roughly $11 million in cost savings and almost 14 million metric tons of carbon emissions reductions.