VRX Technologies
Modular data center cooling units installed in the field

Modular Data Center Cooling

Cooling at the speed of compute.

Add cooling capacity when IT capacity is energized — not years before it is needed. A phased approach to schedule, capital and risk.

Modular data center cooling: cooling at the speed of compute, a phased approach to schedule, capital and risk

01 / Two ways to build

The same capacity. Two very different projects.

A conventional central plant is a large plant, a large piping network and a major day-one commitment. A modular plant is a repeatable block, added as load arrives. The ultimate tonnage can be identical; the schedule and the capital profile are not.

Conventional central plant with large piping network compared with modular, incremental cooling blocks added as load arrives
Two very different ways to build cooling

02 / The core idea

Don’t buy 100% of the cooling plant before you have 100% of the IT load.

Match cooling investment to actual data center growth: usable capacity sooner, mechanical spending deferred until it is needed, and less stranded plant if density or technology changes.

The core idea: faster capacity, deferred CapEx and lower stranded risk by matching cooling investment to data center growth
Faster capacity · deferred CapEx · lower stranded risk
Cooling commissioned in phases of 25, 50, 75 and 100 percent, synchronized with power, data halls and rack deployment
Commission capacity in useful increments

03 / Capital

CapEx follows demand instead of leading it.

A central plant releases its full mechanical budget before the first rack draws power. Modular capacity is released in blocks as data halls are energized and start producing revenue. Deferring spend is not the same as guaranteeing a lower total installed cost — that depends on the project.

Illustrative cumulative cooling CapEx: traditional plant paid early versus modular capacity paid as needed
CapEx follows demand — illustrative phasing, not a financial forecast
Illustrative 6,000-ton campus: one central plant versus twenty 300-ton blocks released in four phases
Illustrative 6,000-ton campus — one commitment or twenty repeatable blocks

04 / Risk

Modularity is also a risk-management tool.

A giant plant commits capital before utilization is proven and locks in early design assumptions. Incremental blocks keep spending aligned with load, let later phases adapt to density or technology changes, and keep one equipment delay from holding the whole campus.

Giant plant up front versus incremental modular cooling: optionality has value
Optionality has value
Buy cooling when the capacity creates value: spend follows load in 25, 50, 75 and 100 percent steps
From “How large a plant do we need at full buildout?” to “What is the next useful block of cooling?”

05 / Part load

A 60 MW data center is not one 60 MW load.

The largest part-load exposure is usually the multi-year buildout, not a daily swing. With three 6,000-ton chillers, one machine is a third of the plant. HydroCore stages in small increments — 60 skids of 16 variable-speed modules — so the active compression can stay close to its efficient operating zone as white space is energized.

Illustrative 18,000-ton campus buildout: staging granularity of 33 percent for three 6,000-ton chillers, 17 percent for six 3,000-ton chillers and about 0.10 percent for HydroCore
Staging resolution — illustrative 18,000-ton campus

06 / Architecture

Aligned on direction. Different in architecture.

The industry has validated modular hydronic plants. HydroCore goes further: cell-sized capacity steps, factory-packaged pumps and hydronics, hierarchical controls from cell to fleet, fault isolation down to the cell, and DewCycle self-cleaning built in.

Packaged modular hydronic plants compared with HydroCore across capacity granularity, hydronics, controls, redundancy and maintenance
Packaged modular hydronic plants vs. HydroCore

07 / DewCycle / Patent pending

Clean coils. Cooling stays online.

When dew point permits, the BMS self-rinses one HydroCore at a time using coil-generated condensate while the remaining fleet keeps cooling. Performance is measured before and after each cycle. Results depend on regional dust and pollen, coil condition and available dew-point windows; manual service may still be needed.

HydroCore DewCycle: one of twenty modules self-rinsing with coil-generated condensate while nineteen remain in cooling
DewCycle — concept sequence

08 / Sound & Community

Sound is the third constraint.

Water and power get the headlines. At night, community noise can decide how well a data center fits its neighborhood. Cooler night air and fleet diversity let HydroCore spread the load across more modules at lower speed after dark.

HydroCore night strategy: sound is the third constraint, with a Silent 3 screening estimate at 30 feet
HydroCore night strategy

The ≈42 dBA at 30 ft shown above is a night-time Silent 3 screening estimate at reduced module loading. The published HydroCore 1 MW acoustic basis is the design reference for rated and quiet-mode sound levels.

09 / Outside air / Cascade

A separate response to the outside-air load.

Cascade addresses the campus outside-air requirement in hot, dry climates. It complements the HydroCore cooling platform with a dedicated approach to conditioning 100% outside air.

Follow the air: 115°F → 80°F → 58.1°F → 57.3°F → 60.7°F. The final rise is fan heat. Illustrated full-assist Pinal County design point; performance depends on operating conditions.

10 / Written for the reviewer

Claims your engineer can take to the AHJ.

Our platform supports the principal U.S. data center energy‑code pathways — ASHRAE 90.1, ASHRAE 90.4 and California Title 24 computer‑room provisions.

We will not tell you a product is compliant. Adoption, amendments and local enforcement vary, and every project needs its own engineering validation and authority review. What we will do is hand your design team the analysis in the form they have to defend it in.

Energy-code pathway analysis90.1 / 90.4 / T24
Water and acoustic basisProject-specific
Cost and O&M modeling10-yr lifecycle
Enhanced Greek pottery artwork depicting Hephaestus at work
Hephaestus — engineering judgment, supported by documented analysis

Bring us the constraint that is holding the project up.

A load profile, a water restriction, a noise ordinance or a phasing plan. You will hear back from someone who can read it and tell you what the platform does about it.