Data center construction costs soar 13% in 2026
U.S. data center construction costs rose 13% year over year, a sharp jump that shows how expensive the AI buildout has become. Turner & Townsend said the increase was more than twice the global rate and about three times the average annual increase over the previous decade. The article calls the period “this year”; it does not label the 13% figure specifically as a 2026 result. The report examined more than 500 live or recently completed projects across 62 global markets, along with a survey of 400 industry experts. Costs climbed as demand for AI computing grew faster than the supply of powered land and the workers needed to build facilities. Faster delivery schedules also added risk and expense for contractors. Turner & Townsend expects cost increases to moderate to 6% to 10% in the next year. Many major projects have already started or are starting, Barry said, so their cost risks are already reflected in contracts. He also expects a surge of major project starts in 2026.
How much did U.S. data center construction costs rise in 2026, and how did that compare with global cost inflation and the past decade’s average?
U.S. data center construction costs rose 13% year over year, a sharp jump that shows how expensive the AI buildout has become. Turner & Townsend said the increase was more than twice the global rate and about three times the average annual increase over the previous decade. The article calls the period “this year”; it does not label the 13% figure specifically as a 2026 result.
The report examined more than 500 live or recently completed projects across 62 global markets, along with a survey of 400 industry experts. Costs climbed as demand for AI computing grew faster than the supply of powered land and the workers needed to build facilities. Faster delivery schedules also added risk and expense for contractors.
Turner & Townsend expects cost increases to moderate to 6% to 10% in the next year. Many major projects have already started or are starting, Barry said, so their cost risks are already reflected in contracts. He also expects a surge of major project starts in 2026.
What does “powered land” mean, and why is it important when choosing a data center site?
“Powered land” generally means a site where enough electricity capacity is available or can be delivered for a development. The article uses the term but does not define it. For a data center, the key issue is not just owning land; the site also needs a viable path to the power needed to run the facility.
That makes power availability a crucial site-selection factor. A developer may have a suitable parcel, but delays or limits in getting an electricity connection can hold up construction or operations. The article reports that 56% of surveyed experts globally considered power a major obstacle to data center delivery. It also says utility connections commonly take one to two years, and many take longer.
The article gives regional examples of power costs: $15.01 per watt in Silicon Valley and $10.51 per watt in Dallas for a typical 60-megawatt design. Those differences underline why developers assess both access and cost. The source does not detail how “powered land” is secured, but its figures show why power is central to choosing a site.
What supply shortages and surging demands drove data center construction costs higher?
Data center costs rose as demand for AI computing power grew faster than the resources needed to build new facilities. The article points to shortages of powered land and labor, alongside strain on the supply chain. These constraints matter because they limit how quickly developers can deliver the large-scale facilities that growing computing needs require.
The report found that just over one quarter of respondents considered contractor capacity sufficient to manage construction backlogs. Power was another bottleneck: 56% of surveyed experts said it was a major obstacle to delivery. Utility connections often take years, adding another limit on how quickly projects can proceed. Meanwhile, project schedules became more aggressive, raising contractors’ exposure to risk.
Together, those pressures pushed U.S. data center construction costs up 13% year over year, according to Turner & Townsend. The company expects increases to moderate to 6% to 10% in the next year, but the article anticipates a surge of major project starts in 2026. Continued demand could mean more projects competing for land, labor, power, and delivery capacity.
Why do tighter construction schedules lead contractors and subcontractors to charge more?
When a data center must be built faster than a contractor would traditionally plan, there is less room to absorb delays or unexpected problems. Contractors and subcontractors respond by including extra risk allowances in their prices. That helps explain why compressed schedules can raise project costs even beyond the cost of materials and labor themselves.
Paul Barry of Turner & Townsend said those allowances can cover acceleration, more overtime, and efforts to attract workers. Contractors may also charge higher fees for accepting the risk of delivering on a faster timetable. The article links this pricing behavior to the surge in demand and strain on the supply chain, which together make fast delivery harder to guarantee.
These schedule pressures contributed to the reported 13% year-over-year rise in U.S. data center construction costs. Barry expects increases to moderate to 6% to 10% in the next year, partly because many major projects have already started or are starting now. For those projects, he said, cost risk has been baked into contracts.
Why can connecting a data center to the electricity grid take years, and how long do utility connections typically take?
The article says 56% of surveyed data center experts globally considered power a major obstacle to delivery. Just over half said utility connections typically take between one and two years, while almost a third reported timelines longer than two years. Those waits can hold back projects that need dependable electricity to move forward.
The article does not explain the specific causes of long connection times. As general context, grid connections can require coordination with utilities, available capacity, and any needed network upgrades or approvals. These processes may take time, but the source does not say which factors account for the delays reported by its survey respondents.
The figures show why power access is a central planning issue for data center projects. The article also reports regional power-cost differences: $15.01 per watt in Silicon Valley versus $10.51 in Dallas for a typical 60-megawatt design. Long waits and varying costs can affect where and when developers build, even as demand for AI computing drives new projects.
How large is the expected increase in global data center investment, and what share is the U.S. expected to attract?
Global data center capital expenditure is projected to rise from about $800 billion this year to $1.8 trillion per year in 2050, according to a recent PwC report. That forecast signals a large expansion in spending on facilities as demand for computing power grows. The figures refer to global data center capital expenditure, not just U.S. projects.
The U.S. is expected to capture almost half of all global investment in AI infrastructure. The article says that category includes both data centers and chips, so the share is not a figure for data centers alone. Strong demand for AI computing is the background to this investment outlook and the construction surge discussed in the article.
That spending outlook comes as powered land, labor, power connections, and contractor capacity constrain project delivery. Turner & Townsend expects construction cost increases to moderate to 6% to 10% in the next year, but Paul Barry expects a major surge in project starts in 2026. If demand continues, the article says, further project starts may follow.
What is a data center, and why do AI systems need facilities with so much computing equipment and electricity?
A data center is a facility that houses computer servers and related equipment used to store, process, or transmit digital information. This basic definition is general background; the article focuses on construction costs rather than describing the facilities themselves. It does say that demand for computing power for artificial intelligence is driving a major data center buildout.
AI systems require computing resources to process data and carry out calculations. Those systems run on equipment housed in data centers, which in turn requires electricity. The article stresses power’s importance: 56% of surveyed experts said it was a major obstacle to delivery. It does not give technical details about AI hardware or facility cooling.
The rising need for AI computing has helped push demand for large-scale U.S. data centers beyond the supply of powered land and construction labor. That imbalance, along with supply-chain strain and faster schedules, contributed to costs rising 13% year over year. PwC projects global data center capital expenditure will reach $1.8 trillion per year in 2050.
This brief was written by AI from the original reporting and checked by other models. Names, figures and quotes come from the source; read it for full context.
Read more in the JupiteX app
Pulse is free. New stories every 4 hours, each one broken into the questions that explain it.
Or read more news on the web