Big Tech Tenants Push Data Center Developers To Stay On Schedule As Supply Chain Woes Linger

Data center developers continue to struggle with supply chain problems, and their largest tenants are paying close attention to who has answers.

Persistent supply chain headaches remain a central challenge for data center developers, particularly providers focused on leasing entire facilities or campuses to big tech hyperscale tenants.

Wait times for key equipment — from generators and transformers to cooling and power systems — remain long and unpredictable. Global suppliers are struggling to keep pace as an artificial intelligence arms race drives the world’s largest data center tenants to seek ever-larger facilities on shorter development timelines.

Tech giants such as Microsoft — who have bet their futures on technologies like generative AI and are frantically trying to scale up the infrastructure to support it — have identified supply chain problems delaying new capacity as a significant risk factor that could limit their competitiveness.

Microsoft Senior Program Manager Kartik Atyam, who works on expanding the tech giant’s data center portfolio, said that Microsoft has identified this continued supply chain volatility as the most significant challenge to successfully executing the company’s digital infrastructure strategy.

“Timelines are still extending, and even if some elements are coming in faster there are others that are taking longer,” Atyam said at Bisnow’s National DICE Construction, Design and Development West summit last month. “The capacity and flow-through rate to be able to support the builds that we desire is just not there."

Data center developers and tenants at the event said these hyperscalers have begun paying much closer attention to which data center providers have the supply chain savvy to deliver capacity on time, and that those able to successfully navigate these stubbornly choppy supply chain waters will be the beneficiaries of the accelerating wave of AI-driven hyperscale demand.

“We looked at three different [requests from hyperscale tenants] we received over the last three or four weeks, and all had one thing in common: They wanted to know what is our supply chain plan, do we have our [long lead-time equipment] ordered, what is the timeline for it, and what is our risk mitigation plan in the event that we see additional delays there?” Brett Severson, vice president for real estate at data center provider Evoque Data Center Solutions, said at the DICE event, which was held at the San Jose Marriott. 

“It has become a major theme," he added. 

More than three years after the onset of the pandemic wreaked havoc on global supply chains, data center builders continue to face long lead times for equipment from major suppliers, and expected wait times that shift dramatically from week to week. That can mean waiting anywhere between 12 and 18 months for emergency generators, while backup power systems, switching gear, and the chillers and HVAC equipment needed to cool data centers are all looking at lead times of close to a year, industry insiders told Bisnow. 

Power infrastructure like transformers and substations, increasingly the responsibility of developers instead of utilities, often has lead times measured in years, not months. 

Why has this supply chain bottleneck continued to plague the data center industry?

While developers and tenants said there are a range of factors at play, most pointed to one as the most significant: Data centers are getting bigger, faster. 

It is not news that data centers have been getting bigger over time. A five-megawatt data center development would have been considered a large project just four years ago, whereas hyperscalers now rarely lease buildings under 30 megawatts, with large projects in the hundreds of megawatts. 

But data center experts said that the past six to eight months have seen a rapid increase in the scale of facilities that hyperscale tenants are looking for. This means more generators, more transformers and more HVAC systems — new demand that suppliers have yet to scale up to meet. At the same time, hyperscale tenants want shorter timelines for building those facilities out to full capacity, with speed to market becoming a growing priority. 

“Traditionally, you were looking at projects that had longer-term ramp schedules, where they'd build out to 50 megawatts and then lease up over a 10-year period or 36 megawatts over three years on the hyperscale side of things,” said Andy Cvengros, a managing director on JLL’s data centers team. “In the past six to eight months, right now what you're seeing is really massive requirements with people looking for gigawatts of power for AI.”

Indeed, generative AI has been the driving force behind hyperscalers’ desire for greater scale and at a faster pace. Locked in an AI arms race, firms like Microsoft, Amazon Web Services, Google and Meta are scrambling to build out or lease the data center infrastructure needed to host the massive computing power these technologies require. Delays in the delivery of new data center capacity means potential delays of AI products and services that are increasingly central to the business models of the world's largest tech companies.

“We’re building for generative AI, and it’s just faster, bigger, as soon as possible and as much as possible,” Microsoft’s Atyam said. “When we’re looking at the AI space, whether it's on the inference or on the training models, wherever we can find the capacity, we're building now. If there is an open space that we think customers will come in, we're saying please move aside. We're going to build.”

With such significant risk for tenants tied to potential supply chain-driven delays, companies like Microsoft are taking an increasingly proactive approach to avoiding supply chain disruptions and ensuring the consequences of any delays that do take place are felt primarily by the data center provider. 

Atyam said Microsoft is expanding teams devoted to evaluating supply chain and other development risks and is increasingly engaged with data center providers to better understand their ability to navigate supply chain issues before sitting at the negotiating table. The company also insists on contract terms for capacity under development that levies significant penalties on developers for any delay in delivery. 

“When it comes to putting the onus of the risk on the lease providers, that's absolutely what we're doing: We sign a contract with the expectation of capacity at a certain date,” Atyam said. “What's really valuable to us is just getting that capacity up as soon as possible … we're definitely taking a close eye and working much more closely with our lease providers to understand what the risks are, and if they're going to push out their delivery then we're looking at other lease providers to also make up that gap. “

It’s not just Microsoft. Evoque’s Severson said multiple hyperscale tenants are increasingly focused on reducing their exposure to risk from supply chain problems and have been ramping up their due diligence on providers’ capabilities when it comes to managing disruptions.

He said data center providers need to get comfortable with this increased scrutiny if they want to capture a piece of the AI-driven demand surge that has only just begun. 

 “The customer is taking an active role in their risk mitigation by putting the onus on us to figure those things out, and we should be able to,” Severson said. “We’ve got to have that dialed in and we've got to have a risk mitigation plan so we can be a long-term solution that's going to be delivering on time.”