By Michael Delgado
Data center construction has exploded, with U.S. spending reaching $41.1 billion last year — up from $1.8 billion in 2014. The work is there for electrical contractors ready to take it on, but whether a firm can capitalize increasingly depends on that firm’s ability to operate efficiently under intense time pressure with narrower margins for error.
Identifying High-Impact Bottlenecks
Data center builds are, on average, larger, more tightly scheduled, and more complex than traditional commercial and residential work. When a firm goes looking for the bottleneck that threatens margin, the instinct is to look first at the field, where labor costs are highest and delays are most visible.
The field offers real ways to gain efficiency, yet it subjects a contractor to forces beyond their control. Weather, government actions, and site conditions will set a job back no matter how well it’s run, and trade sequencing routinely stalls the crews waiting downstream through no fault of their own. What a contractor can control comes down to crew speed and productivity, but the hardware meant to lift those, like layout robots and exoskeletons, must be deployed, learned, and maintained one jobsite at a time. Gains spread slowly across an organization even as data center demand climbs.
The back office carries the same level of risk to the schedule, but more of that risk is under a contractor’s control. A single change to how the business runs affects every active project at once, making office-level bottlenecks higher- leverage opportunities for efficiency gains.
A data center build makes it clear which of those bottlenecks needs to be addressed most.
Higher Volumes, Longer Lead Times
A data center packs in more electrical gear than traditional commercial and industrial jobs while carrying lead times two to four times their pre-pandemic norms:
- Medium-voltage switchgear: 40 to 65 weeks
- Distribution transformers: 40 weeks or more
- Standby generators: up to 60 weeks
In response to the growing demand, hyperscalers like AWS, Google, and Microsoft are now buying in bulk, which concentrates production capacity in a handful of large buyers and leaves everyone else waiting longer. Most data center work is colocation, enterprise, and regional builds, and those jobs are less likely to be owner-furnished than hyperscale campuses — putting sourcing pressure for long-lead equipment directly on contractors. A quoting or ordering error that once cost a shop a day can now cost it a place in an extremely tight and expensive manufacturing queue.
Long-lead gear is only part of the problem. A data center is built to hit a fixed power-on date, and the schedule around it is sequenced so tightly that even routine materials can derail it. Conduit, wire, and fittings carry little lead-time risk, but a crew scheduled to install them on a set day has no slack for a wrong count or a missed delivery.
Procurement Process Breakdown
Naming material as a leading cause of delay puts the spotlight on the work of buying, tracking, and managing it. For most contractors, that work is still done by hand. Purchasing teams and project managers juggle dozens of suppliers across multiple document formats and communication channels. Every quote, confirmation, and update must be reviewed, consolidated, and keyed by hand into a spreadsheet or accounting system to keep information centralized and up to date. One mistyped number or a missed shipment delay update that never reaches the right person can throw off a job. Across dozens of suppliers on every project, those small errors compound into stockouts, overorders, and work stoppages.
For a data center project, that process meets higher volumes and tighter tolerances, so mistakes become more inevitable and intolerable.
Why AI Changes the Equation
Historically, construction has been harder to digitize than most industries. The work is project-specific, the players operate in separate systems, and the information moving between them is messy by nature. AI is the first technology suited to that reality. AI can work with unstructured information, and it can absorb the coordination and handoffs between teams and systems that defeated earlier software. The McKinsey Global Institute estimates AI and automation could add roughly $228 billion in annual value to the U.S. architecture, engineering, and construction industry by 2030, and that AI could automate 39% of the sector’s nonphysical work, with invoicing and data entry among the areas expected to change most, precisely the manual, error-prone core of procurement.
What Modern Material Management Requires
Applying AI to procurement in practice looks like addressing manual points of failure with new solutions that:
- Put the whole process in one system. The gains come when the field, the office, and purchasing all transact in the same place, so that a request, a quote, and a purchase order stop scattering across inboxes and spreadsheets and no one rekeys the same number twice. Plugging into the accounting and project- management tools a contractor already runs on keeps that record accurate without adding another silo.
- Meet each role where it already works. The field should be able to send a request from the jobsite in whatever form is fastest — a photo of a handwritten list, a voice note, a few typed lines — and have it come back as a clean, structured order, while purchasing and project managers work from views built for their part of the job.
- Connect directly to suppliers. Quotes and availability should return from a distributor’s own system in real time, priced on the terms a contractor has already negotiated, so a team sees what is actually deliverable without maintaining brittle portals and electronic data interchange links.
- Automate the error-prone steps. A system that ingests any supplier quote, in any format, and levels bids automatically removes the manual leveling and reentry where mistakes enter, which matters most on long- lead gear, where one wrong quantity can cost a slot in the queue.
- Give everyone one live view of the job. The whole team should see the same status and history on every order and line item, quote through closeout, with committed, released, and remaining quantities tracked against each buyout, so a slipped date surfaces while there is still time to act.
The Firms Positioned to Scale
The contractors best positioned for the data center boom are the ones treating procurement as operating infrastructure, built to move material accurately and at speed. Handled that way, procurement protects the schedule, defends the margin, and creates the capacity to take on the next project. As the build- out accelerates, procurement speed, operational visibility, and back-office efficiency are what enable firms to meet the moment and scale with confidence.
Michael Delgado is CEO and co-founder at Canals, a Miami-based company that offers AI models trained specifically for the complexities of manufacturing, distribution, and construction across key verticals. Meet Michael and Canals team members at SPARK 2026, Booth 334.