The Budget That Looked Great on Paper
I manage procurement for a mid-sized colocation provider. Last year, during our Q4 planning cycle, I put together a cooling infrastructure upgrade budget that I was actually proud of. We needed to replace two aging York chillers and expand our capacity by 400 kW. I did the math. I got quotes from three vendors. I had a clear winner on cost.
Then the project went live.
Within three months, we were 22% over budget. Not on the chillers—those came in exactly at quote. The overrun was in places I hadn't modeled: field wiring issues with the new temperature sensors, a fan motor failure on an auxiliary unit that wasn't part of the main bid, and three service calls to recalibrate the building automation system's communication with the Metasys controller. I'd tracked every invoice. I still missed it.
The Surface Problem: 'The Chiller Is Too Expensive'
When most facility managers or finance teams look at a data center cooling upgrade, the first thing they do is compare chiller prices. It makes sense—it's the biggest single line item. A 500-ton centrifugal chiller from York can run anywhere from $250,000 to $400,000 depending on efficiency rating and configuration. That's where the sticker shock is.
So they do what I did. They optimize that one number. They negotiate the chiller price down by 8%, feel good about the savings, and move on.
Here's the thing: that focus on the chiller is rational, but it's incomplete. The chiller is maybe 35-40% of the total project cost in a typical retrofit. The rest—controls, sensors, VFDs, ductwork, commissioning labor—is where the budget gets eaten alive.
The Deeper Issue: Sensor and Control System Complexity
The surprise wasn't the chiller performance. It was how much time we lost to temperature sensor integration. We installed 47 new Johnson Controls temperature sensors across the facility. On paper, they were all the same model. In reality, three of them failed to communicate properly with the Metasys front end because the wiring run was too long and introduced signal noise. We had to pull new cable for two of them. That cost $1,700 in labor and materials I hadn't budgeted for.
I don't have hard data on how often this happens industry-wide, but based on our last five projects, I'd guess it affects about 10-15% of sensor installs in retrofit environments. It's not the sensor's fault. It's that the installation conditions in an existing building are never as clean as the spec sheet assumes.
The AC Fan Motor That Wasn't in the Scope
Another hidden cost: the AC fan motor on a CRAC unit adjacent to our new chiller. The old motor was rated for 15 years of continuous operation. It failed at year 12. Not unusual, but the timing was terrible. We had to order a replacement motor—a specific variable-speed model compatible with the VFD—on an emergency basis. Rush shipping cost us an extra $350. The motor itself was $1,200. Total: $1,550 we hadn't planned for.
A small number, but when you stack three or four of these 'surprises' on top of each other, you're suddenly looking at $8,000 to $12,000 in unplanned spend. On a $600,000 project, that's 2%. Annoying but manageable. On a $150,000 project, that's 8%. That's painful.
The Real Cost: Installation and Integration Labor
Every spreadsheet analysis I did pointed to a specific vendor—they were 12% cheaper on the chiller than the other two quotes. Something felt off. Their responsiveness during the bidding process was slow. Emails took two days to answer. I went with my gut and chose the slightly more expensive vendor. Later, I learned the cheaper vendor had a reputation for poor commissioning support.
What I'm getting at is this: the biggest hidden cost isn't equipment failures. It's integration complexity. Modern data center cooling isn't just a chiller anymore. It's a chiller connected to a VFD, controlled by a building automation system, monitored by temperature sensors, and optimized by an AI-driven platform. Each of those connections is a potential failure point—and a potential cost overrun.
Based on the projects I've tracked over the past six years—about 18 major HVAC upgrades across three facilities—the ratio is roughly:
- 45% equipment (chiller, pumps, fan motors)
- 25% controls and sensors
- 20% installation labor (including unexpected rework)
- 10% project management and commissioning
Most procurement managers focus on the first 45%. The other 55% is where the real budget risk lives.
The Vendor Who Said 'This Isn't Our Strength'
I remember a conversation with a sales engineer from one of the major controls vendors—I won't say which—who told me point-blank: 'We're great at the chiller, but if you need a specialized air dryer for your compressed air system, you're better off calling a specialist.' I respected that. The vendor who said 'this isn't our strength—here's who does it better' earned my trust for everything else. We ended up buying the chiller from them anyway.
That's the expertise boundary a lot of vendors don't want to admit exists. 'One-stop shop' sounds great in a marketing brochure. In practice, generalists struggle with the deep integration work that makes a modern cooling system actually efficient.
So What Actually Works?
I'm not saying you should avoid the big chiller vendors. I'm saying you need to model the total cost of installation, not just the sticker price. After getting burned on hidden fees twice—once on a sensor retrofit, once on a fan motor replacement—I built a simple cost calculator for our team. It includes line items for:
- Cabling and wiring (assume 10% overrun)
- Sensor calibration and testing (1-2 hours per sensor)
- Spare fan motors for critical units (stock one per 10 units)
- Commissioning support from the controls vendor (not included in many quotes)
That 'slow to reply' vendor from earlier? Turns out 'slow to email' was actually a preview of 'slow to deliver replacement parts.' We dodged that bullet.
I still don't have perfect data on exactly how much these hidden costs add up to across the industry. But anecdotally, on our last three projects, the delta between 'budgeted' and 'actual' was between 7% and 18%. The worst was a project where the installation contractor under-quoted by 40% on labor then tried to change-order us after the chiller was on the roof.
The solution isn't to buy a cheaper chiller. It's to buy a chiller from a vendor who understands that the installation is more complicated than the spec sheet suggests—and prices accordingly.
