The Grid Is the Schedule

Speed is a design problem before it's a construction problem.

Every client meeting I've had this year opens with the same question. Not "what will it look like" or even "what will it cost." It's "when can we go live?"

The industry outlooks for 2026 say the same thing in business terms: schedule has overtaken cost as the priority. Owners still watch capital spend, but whoever delivers capacity first collects the revenue. In 2018, a 400 MW facility could be built in about a year. Today, with contractors stretched thin, longer equipment lead times and power scarce in many markets, that pace is a very big ask. For a newcomer without an established supply chain, it may not be possible at all.

Most of the speed conversation happens in the field: more crews, more shifts, earlier equipment orders. All of that matters. But in my experience, the days a building gains or loses are mostly decided much earlier, at the drawing board, through choices about the parti, the grid and the kit of parts.

Start with the parti, not the program

A data center's program is almost brutally simple: a large clear-span volume, a band of support spaces and a big equipment yard. Because it's so simple, it's tempting to treat the architecture as an afterthought, a box wrapped around the equipment.

I think that has it backwards. When the program is this lean, the organizing idea carries more weight, not less.

One of my projects, a multi-building hyperscale campus, is under construction now. The data halls sit side by side and stack like Lego pieces. The power diagram reads in a straight line: utility feed, transformers, UPS, generators and switchgear, all lined up one after another like cars on a train. That clarity really pays off once you add redundant equipment to the mix. A tangled power diagram turns every backup path into a coordination puzzle. A linear one lets you see the whole chain at a glance, and so can the contractor installing it.

The structural grid is the schedule

When people ask me where a building's speed comes from, I tell them to look at the column grid.

A disciplined, repeating structural bay means repeating precast or steel members, repeating connections and repeating envelope panels. Fabricators run a production line instead of a hundred one-offs. Erection crews find their rhythm by the third bay. Every special condition, whether it's a transfer girder, an odd bay squeezed in for a property line or a mezzanine that doesn't align, breaks that rhythm.

The large operators are now building in capacity blocks of roughly 30 to 50 MW, or fitting out halls one at a time behind a moving screen inside a larger shell. To an architect, that's just phasing done honestly. Design the shell as permanent architecture, treat the interior as infill and keep the joint between them clean enough that the next phase can start without disturbing the last one.

Repetition at two scales

Modularity works at the scale of the site as well as the building. On that same campus, the macro move was simply to repeat the same building side by side, with a substation mirrored at each end of the site. One design, built several times over. Every lesson learned in the first building carries straight into the next.

At the micro scale, most of my projects have been precast: wall panels, columns, double-tee floors. There's a real beauty in reusing the same kit of parts. Fewer piece types, fewer shop drawings, fewer surprises.

The catch is geography. Precast is only as fast as the trip from the plant. Distance to the casting yard drives freight cost, and road limits drive panel size. An oversized panel won't ride flat, so it gets tilted on an A-frame trailer to stay within legal load limits. Those limits should shape the panel module from the first sketch, not get discovered during shop drawings.

Borrow from housing

The data center world could learn a lot from multifamily and student housing, where volumetric modular construction is further along than most people realize.

On a residential project in our office now, the units arrive as complete, self-contained boxes: windows installed, floors and finishes done, walls and ceilings painted, cabinets and doors hung, and packaged HVAC already in place. They're engineered stiff enough to be lifted straight off the trailer and set by crane without temporary bracing, and strong enough to carry several more units stacked above them. Each unit ships fully enclosed and watertight. On site, the crew only has to daisy-chain the electrical, plumbing and mechanical connections from one unit to the next. After that, the only thing left is to put up the facade in front of them.

Data centers are heading the same way. Self-contained modular server rooms, built in the factory with their MEP systems already installed, are gaining real traction, and so are prefabricated UPS modules. The idea even extends to power generation. On a previous project, we proposed aeroderivative turbines as packaged, modular power on site. The housing sector has already worked through many of the same questions, including lifting points, transport tolerances, mate-line connections and inspections at the factory versus on site. We should be paying attention to that.

This is definitely one way to go. What I'd really like to see is a fully modular data center taken all the way through, with every major system built off site. How long would it take? What would it cost? Where is the capacity ceiling? Which sites would it suit and which would it not? And how much faster would it actually get to market? I suspect the answers would surprise a lot of us.

Material choice is a schedule decision

A practical example comes from Brazil. Steel-framed facilities near the coast have suffered from salt-air corrosion, and at least one is already undergoing major repairs. Precast concrete is often the better choice there. It can be produced faster than steel, and its thermal mass reduces cooling load.

The broader lesson is that tectonics and schedule are tied together. The fastest structure on paper isn't the fastest if its fabricator is booked for a year, if it needs a long coating regime before close-in, or if the plant is too far away to ship economically. Standardized design still needs a local architect's judgment.

Draw it twice, build it once

The common advice now is to build the facility virtually in BIM before anyone breaks ground. I agree, with one condition: a model only saves time if it's coordinated. That means architecture, structure and MEP working in the same model and resolving clashes there instead of through RFIs.

It's usually the small things: a duct bank that runs straight through a precast beam, a cable tray fighting a sprinkler main, a generator pad that doesn't line up with its louvered opening. In the model, each one takes a few minutes to fix. In the field, after the concrete has cured, it means coring, redesign and waiting. Catching a handful of those in the model before shop drawings go out can save weeks.

Speed has a price

None of this comes free. Some owners have accepted tender prices as much as 25 percent higher just to hold an aggressive schedule. AI-ready facilities cost far more per megawatt than cloud facilities, so scale isn't delivering the savings people expected. Lead times for power equipment in North America have stretched to two or three years.

Design can't fix a transformer backlog. What it can do is make sure the building is never the reason the schedule slips. When the gear arrives, the shell should be ready, the pads poured and the interfaces already resolved.

I'll add one honest caveat. "Speed at all costs" really only works during design. That's when changes are cheap, because they're still lines on paper. Once a design is approved and permitted, every change costs more, and it doesn't necessarily buy any speed. More often it slows things down. New materials have to be ordered, permits revised, and sometimes work that's already in place has to be torn out. The fastest projects I've seen are the ones that made their hard decisions early and then left the drawings alone.

The architect's role, compressed

There's one more shift I see every week: the supply chain now chooses its clients. Contractors and consultants are going to owners with long pipelines, fair risk allocation and designs that are ready to build.

For architects, our value on these projects isn't the beauty of a single building. It's the clarity of a repeatable one: a disciplined grid, a resolved kit of parts, clean phasing joints and material choices that respect the real supply chain. That's how a designer contributes to speed.

One example from our own work: we helped a client establish a global design language for the building skin. On its face, it was a branding exercise, a consistent look for their facilities everywhere. In practice it did far more. Once the skin had a defined vocabulary, the facade panel sizes, joint details and materials became standard as well. Each new project started from a resolved envelope rather than a blank page. Fabricators knew what was coming, details didn't need to be reinvented, and the design team could spend its time on the things that really were site-specific.

Final Thought

That's the quiet secret of fast buildings. The speed doesn't come from rushing. It comes from decisions made once, made well and then repeated.

About the Author

Frederick is an award-winning architect and planner with over 35 years of architectural design experience and nearly 20 years in the data center sector, bringing an international portfolio that spans large-scale facilities and diverse project types. His background across global markets gives him a broad, practical lens that shapes the insights he shares.

Throughout this series, he closely follows the evolution of the data center market, drawing on industry events, newsletters and ongoing dialogue with fellow experts to distill what's worth paying attention to and why it matters.