Hyperscale data centers are expanding at an incredible pace, driven by the demands of AI and cloud computing. These facilities are growing larger, consuming more power, and requiring advanced technologies to keep up with rising workloads. Here's what you need to know:
The hyperscale boom is reshaping construction, technology, and workforce demands. Companies that secure skilled talent early and adopt new construction methods are best positioned to succeed in this rapidly evolving landscape.
Hyperscale Data Center Growth Statistics 2025-2030
AI workloads are transforming how data centers approach power and cooling. While traditional setups typically supported racks consuming 10–20 kW, modern AI clusters now operate at 60–120 kW per rack, with some exceeding 200 kW. This leap is tied to the rising Thermal Design Power (TDP) of AI accelerators. For instance, NVIDIA's GPU advancements highlight this trend: the V100 consumed 300 watts, the A100 required 400 watts, the H100 jumped to 700 watts, and the B200 now demands 1,000 watts. By 2029, cutting-edge GPUs are expected to surpass 4,000 watts in TDP.
The shift from 10–20 kW racks to systems often exceeding 100 kW per rack introduces significant challenges. AI training, which synchronizes thousands of GPUs, creates intense "peaky" power swings. During computation phases, peak power usage can surge up to 50% above steady-state TDP for brief periods. Electrical systems must be robust enough to handle these sudden spikes without failure.
To address these demands, data centers are upgrading their infrastructure. Higher voltages and specialized power cords (IEC 60320 C21/C22) are now standard to support increased rack loads and prevent overheating. Additionally, operators must assess floor strength, as AI racks can weigh over 6,600 pounds.
These rack-level adjustments are just the beginning, as the impact extends to the entire data center campus.
Beyond individual racks, entire campuses are evolving to support the power needs of AI workloads. Some facilities now require over 1 GW (1,000 MW) of power, with more than ten data center campuses in North America already operating at or above this threshold. By 2030, U.S. data center power demand is expected to hit 134 GW, nearly tripling the levels projected for 2024.
This rapid growth has shifted the focus from power generation to transmission. On average, grid interconnection timelines now stretch to four years or more, with some regions experiencing delays of up to seven years for new grid connections. To adapt, developers are adopting a "power + land + interconnect" approach, which integrates grid positioning and deliverable power into project planning. For example, in January 2026, CyrusOne and Eolian launched a 200 MW campus in Fort Worth by repurposing high-voltage infrastructure initially built for a 100 MW Battery Energy Storage System.
"The constraint is no longer buildings. It is power pathway certainty."
- Roxanne Marquis, Founder, 8888CRE
For contractors entering the hyperscale data center market, understanding these power and energy infrastructure challenges is critical. Managing gigawatt-scale campuses requires expertise in high-voltage distribution, grid integration, and on-site power generation solutions.
The demand for rapid AI infrastructure deployment has made modular construction a game-changer. By allowing off-site fabrication and on-site work to happen simultaneously, modular construction slashes traditional project timelines from 24–36 months to just 12–18 months.
This parallel approach tackles a major challenge in data center construction: the shortage of skilled labor. By reducing on-site labor needs by 60–70%, off-site assembly eases the pressure on specialized workforce demands.
With modular construction, prefabricated components arrive at the site up to 80% complete. This approach turns months of fieldwork into just days. For instance, electrical installations that once took 12 weeks can now be completed in as little as 10 days using prefabricated power modules.
A key factor in this efficiency is Factory Acceptance Testing (FAT). Systems are commissioned and certified before shipping, catching 95% of issues ahead of time and cutting on-site remediation by 90%. In 2024, Schneider Electric showcased this by delivering a 4MW facility supporting 320 NVIDIA H100 GPUs in just 11 months. The prefabricated modules arrived ready for immediate installation, streamlining the entire process.
"Prefabricated cooling modules are built, tested and certified in a controlled factory environment while the data center foundation is still being poured, allowing true parallel construction."
- Ashley Dirou, Global Senior Sales Developer, Grundfos
Coreweave also embraced this method in 2024, deploying 3,000 NVIDIA H100 GPUs across three modular facilities in just 10 months. This speed helped secure $180 million in contracts that would have been out of reach with a traditional 2-year build. The readiness of off-site components plays a crucial role in avoiding staffing mistakes that delay construction.
Modular construction reduces hyperscale project timelines by 30–50%. For example, Microsoft Azure AI used modular data centers in 14 global locations in 2024, completing projects in an average of 13 months from contract to operation. Standardized designs eliminated the usual 3–4 month architectural iteration phase, further speeding up the process.
In Malaysia, K2 Strategic completed two 40MW modular buildings in just 9 months in 2025. These fully containerized units arrived as self-contained systems with integrated power distribution and cooling, requiring only site interconnection. Additionally, commissioning times have been reduced significantly - modular facilities now take just 6–8 weeks for Level 5 commissioning, compared to the 12–16 weeks needed for traditional builds.
For AI workloads, the benefits are even more pronounced. As of late 2025, prefabricated liquid-cooled modules have cut deployment timelines to just 8–10 months. The financial impact is clear: a facility generating $2 million monthly from AI workloads can offset the 20–30% modular capital expenditure premium in just 8 months by starting operations sooner. This accelerated pace highlights the growing need for skilled professionals who can manage these fast-moving, technology-driven projects.
The move toward AI-driven infrastructure is reshaping the skillsets required for hyperscale data center projects. Traditional air-cooled systems are being replaced by liquid cooling and large-scale battery installations, but the workforce hasn’t kept pace. Today, about 80% of cooling loads in new data centers rely on liquid cooling, while only 20% still use conventional air-cooling systems. This shift means that MEP engineers, mechanical contractors, and operations teams must develop expertise in technologies that were rare just a few years ago.
The U.S. is also grappling with a significant labor challenge - an anticipated annual shortage of around 81,000 electricians through 2030. To address this, companies are turning to professionals from adjacent industries like military nuclear operations, oil and gas, and helicopter maintenance. Matt Landek, Data Centers Division President at JLL, highlights the difficulty, saying, "Finding the staff is extremely difficult, the logistics are overwhelming, and there's constant pressure to get the equipment tested and turn the servers on".
Liquid cooling, such as direct-to-chip and immersion cooling, can cut power consumption by 20–40% compared to air-only systems. However, these systems require a completely different approach. For instance, MEP professionals must shift from designing air-duct systems to creating complex liquid piping networks, often using stainless steel to ensure durability and compatibility. HVAC engineers also need to factor server-rack fans into their ventilation designs, treating them as integral components of the mechanical system.
Operations and maintenance are becoming more intricate, too. Teams must manage water quality and apply chemical treatments to prevent corrosion and inefficiency in large coolant loops. Operators also need training on advanced diagnostic systems for predictive maintenance of high-capacity pumps and heat exchangers.
"At this scale, liquid cooling is no longer just about removing heat, it's about system architecture now."
- Maciek Szadkowski, CTO, DCX
Commissioning has grown exponentially more complex. Travis Schumacher, Senior Project Manager at DPR Construction, explains, "The traditional commissioning method is not valid for AI liquid cooling design... we could have potentially 200 to 400 load banks on the floor just to test the various systems". That’s up to 10 times the load banks used in traditional builds, requiring specialized training and additional testing tools.
To address the skills gap, Microsoft launched Wisconsin’s first "Datacenter Academy" in partnership with Gateway Technical College in May 2024. This program aims to train over 1,000 students in five years, preparing them for high-demand technical roles in hyperscale data centers.
While cooling systems evolve, energy storage solutions are also advancing to meet the demands of AI-driven infrastructure.
Battery Energy Storage Systems (BESS), ranging from 5 MW to over 50 MW, are now a standard feature in hyperscale projects. Once limited to emergency backup, these systems now play a critical role in core infrastructure, supporting grid stability and peak load management. For example, short-duration lithium-ion BESS can reduce an 800 MW load to 600 MW by absorbing peak spikes. However, designing and running these systems requires expertise in microgrid controller programming and real-time telemetry.
Large-scale battery setups bring unique challenges, such as managing structural loads, implementing specialized fire safety systems, and handling complex electrical sequencing. Engineers must also design hybrid systems that combine on-site power generation - like gas turbines or solar - with BESS to ensure consistent performance under variable conditions.
In February 2026, Energy Vault partnered with Peak Energy to purchase 1.5 GWh of energy storage. This collaboration focuses on creating energy storage solutions for "AI Neoclouds" and AI-first data centers, using sodium-ion battery technology to meet the high-density demands of these facilities. Julien Dumoulin-Smith, Power Sector Analyst at Jefferies, remarks, "Battery energy storage is an increasingly critical part of the data center infrastructure".
To address immediate talent shortages, a North American OEM worked with NES Fircroft in early 2026 to deploy six experienced O&M engineers across three U.S. sites. This strategy of bringing in specialists from around the world is becoming essential, especially as hyperscale projects expand into remote, resource-rich areas with limited local workforces.
The construction industry is grappling with a shortage of 439,000 skilled workers as of late 2025, largely driven by the boom in hyperscale data center projects. Tech giants like Amazon, Google, and Microsoft are behind more than 400 data centers currently under development. To meet the growing demand for AI workloads, data center capacity is expected to grow by 130% by 2030. However, this surge has left construction firms with project backlogs stretching nearly a year, all due to the challenges in MEP hiring. These workforce gaps are seriously impacting project timelines and the delivery of hyperscale facilities.
The demand is especially high for highly skilled trade positions. Te-Ping Chen described this trend aptly, saying, "Data centers are a 'gold rush' for construction workers". These jobs offer wages up to 30% higher than typical construction roles, which has spurred increased interest in skilled trades. Still, the labor shortage highlights the pressing need for more effective recruitment strategies, particularly for mission-critical infrastructure projects.
Certain roles are especially critical to hyperscale construction projects. Electricians and pipe layers are indispensable for managing the complex electrical systems and infrastructure required to support high-density AI workloads. Project managers with experience in large-scale mission-critical projects are also in high demand, as are MEP specialists skilled in advanced HVAC systems and liquid cooling. Additionally, commissioning experts are needed to test load banks and validate intricate system integrations.
Other essential positions include electrical engineers and high-voltage specialists, who design and maintain power distribution systems capable of supporting racks drawing over 100 kW. Filling these roles promptly is crucial to keeping projects on track and avoiding costly delays.
Given these workforce challenges, targeted recruitment solutions have become essential. iRecruit.co is a platform specializing in construction recruitment for mission-critical projects. Their focus includes roles like project managers, MEP specialists, commissioning experts, and other field-level positions. The platform operates on a pay-on-hire model, meaning you only pay a success fee when a candidate is successfully placed. Every candidate is pre-screened for experience in sectors like data centers, infrastructure, energy, and advanced manufacturing.
iRecruit.co simplifies the hiring process by delivering pre-qualified candidates with the skills required for hyperscale projects. If a hire doesn’t work out, the service offers a 90-day search credit to find a replacement, minimizing the risk of a poor fit. This approach is tailored for builders and developers working on hyperscale data centers, where securing the right talent quickly can be the difference between meeting deadlines and facing expensive delays.
Here’s a breakdown of the pricing options:
| Plan Name | Monthly Fee | Success Fee | Best For |
|---|---|---|---|
| 1 Open Role | $0/month | 25% of first year's salary or 3% monthly for 12 months | Single critical hire with no upfront cost |
| 2 Open Roles | $8,000/month ($4,000 per role) | 20% of first year's salary or 2% monthly for 12 months | Growing teams with multiple positions |
| 3+ Open Roles | $10,500+/month ($3,500 per role) | 20% of first year's salary or 2% monthly for 12 months | Large-scale hiring for major projects |
The pricing model is designed to scale with your hiring needs. Companies filling multiple roles benefit from reduced success fees, and monthly charges apply only to active roles. Success fees are only incurred when a hire is successfully placed, offering flexibility and cost-efficiency for firms managing hyperscale construction projects.
Hyperscale operators are pushing hard to upgrade their networks from 100G to 400G and 800G, driven by the demands of AI workloads. In 2024, nearly 80% of switch ports shipped to hyperscalers were 200G or 400G. According to Jimmy Yu, an analyst at Dell’Oro Group, “The whole AI thing is driving the 800G upgrade cycle”.
But it’s not just about boosting speed. Network architecture is undergoing a transformation, moving from traditional Top-of-Rack setups to spine-leaf and fabric mesh networks - a design better suited to handle the immense "east-west" traffic generated by AI clusters. New advancements like 51.2T-class switching silicon, such as Broadcom's Tomahawk 5, allow for larger and more efficient network pods with fewer devices and flatter fabric tiers. Currently, 400G connections dominate leaf-to-server links, while 800G is being rolled out primarily for fabric uplinks and spine switches, helping to cut down on bottlenecks.
In February 2025, DE-CIX, an internet exchange operator, partnered with Nokia to upgrade its New York backbone to 400G with 800G readiness, preparing for bandwidth demand that’s growing at 20% annually. These upgrades require expertise in dense fiber setups, high-radix switching, and next-gen transceiver form factors like QSFP-DD and OSFP. For a deeper dive into how these advancements affect project workflows, check out our guide on data center construction. These rapid changes in network speed and architecture are also reshaping the global data center landscape.
Thanks to these network innovations, hyperscale projects are expanding into new regions at a remarkable pace. The geography of hyperscale development is shifting. Andrew Batson, Global Head of Data Center Research at JLL, noted, “The geography of digital infrastructure is changing faster than most people realize”. Recent figures reveal that 64% of the 35 GW pipeline is now focused on frontier markets, marking a significant move away from traditional hubs like Northern Virginia and Silicon Valley.
Texas is leading this transformation, with 6.5 GW under construction, signaling its rise as a top data center market by 2030. In September 2025, Oracle unveiled the first two buildings of its Stargate I campus in Abilene, Texas, offering 1.2 GW of capacity and supporting over 450,000 NVIDIA GB200 GPUs. Meanwhile, Meta is making waves with its "Hyperion" campus in Louisiana - a $27 billion joint venture with Blue Owl Capital. This massive project, spanning 2,250 acres, will initially provide 2 GW of capacity and scale up to 5 GW.
The Midwest and Southeast are also seeing a surge in activity. For example, in February 2026, Meta began construction on its 30th data center in Beaver Dam, Wisconsin. This $1 billion project is expected to create 1,000 construction jobs and 100 permanent high-tech roles. These regional expansions are driving significant demand for skilled workers, including electricians, MEP specialists, network engineers, and commissioning experts, all of whom are essential for building and maintaining these massive gigawatt-scale facilities. This growing need for expertise highlights the critical role of professionals in supporting this rapid evolution of digital infrastructure.
AI workloads and modular construction are transforming the landscape of hyperscale data centers. However, the real challenge lies in workforce capacity. As highlighted earlier, these advancements require specialized expertise in areas like liquid cooling, BESS integration, and advanced commissioning. To avoid costly delays, securing this talent six to twelve months before breaking ground is no longer optional - it’s essential.
"In 2026, the defining limitation in data center construction is execution capacity"
- MSUITE
With over 80% of construction firms struggling to fill both hourly craft and salaried positions, early talent acquisition has become a non-negotiable priority. The stakes are high: commissioning delays for a typical 60 MW data center can result in monthly revenue losses of around $14.2 million. While nearly $7 trillion is projected to be invested in data center construction over the next five years, success depends on integrated systems and proactive workforce strategies.
"Having all the information we need integrated within CMiC's single source of truth database... has provided our staff with the visibility they need to make better and quicker decisions"
- BJ VanOrman, ERP Strategic Director, JE Dunn Construction
The push for advanced technology and strategic workforce planning is driving growth into emerging regions like Texas and the Midwest. But with this shift comes increased competition for skilled labor.
"The operators who secure skilled labor now, through early contractor commitments and creative workforce strategies, will have significant advantages over those waiting until projects break ground"
- Tony Qorri, DataBank
Organizations that prioritize procurement planning, integrate workflows, and lock in talent early will be best positioned to thrive in this competitive growth cycle. For more guidance on managing these complex developments, check out our comprehensive guide on data center construction.
For an AI-focused hyperscale project, it's crucial to bring in the right expertise. Key roles to prioritize include project managers, MEP coordinators and engineers, commissioning specialists, electrical infrastructure specialists, and project engineers. These professionals play a vital role in handling the intricate infrastructure and maintaining system reliability. Concentrating on these essential positions will help establish a solid framework for your project.
Recruitment efforts should kick off early during the planning phase - ideally, several months ahead of the commissioning date. This proactive approach helps sidestep delays that can arise from shortages of skilled labor or the extended lead times often required to secure critical roles and equipment. By starting early, you can ensure the right workforce is ready to keep the project on track and meet deadlines efficiently.
Experts working with liquid cooling and Battery Energy Storage Systems (BESS) need a strong grasp of facility-scale liquid cooling designs, warm-water chillerless systems, and BESS rapid dispatch techniques for managing peak load shaving. Additionally, in-depth knowledge of thermal management, high-density power delivery, and the integration of advanced cooling technologies is crucial for success in these areas.
