Paul Industries designs and installs data center liquid cooling in California: coolant distribution units, technology and facility water loops, rack manifolds, rear door heat exchangers, dry coolers and trim chillers for dense racks in Silicon Valley colocation suites, East Bay and Livermore research computing, and Southern California halls. California’s mild coastal climate lets a warm-water loop reject heat without compressors for most of the year, and the state’s power limits and 2027 efficiency reporting give every kilowatt saved a visible value. Paul Industries builds the whole liquid path from the cold plate connection to the heat rejection yard.
Request a quote or call 201-450-8280
Why data center liquid cooling in California pays off sooner than elsewhere
Data center liquid cooling in California pays off sooner than in most states because three conditions meet: industrial electricity averaging 21.53 cents per kilowatt-hour against 8.13 nationally (EIA, 2024), utilities in Santa Clara and other data center districts that limit how much power a building can draw, and a climate in which warm facility water can be cooled by outdoor air for most hours of the year. Liquid cooling moves heat at higher temperatures than air does, which is what lets a California plant switch off its compressors so often.
Power allocation is the deciding factor in Silicon Valley. Silicon Valley Power, Santa Clara’s city-owned utility, reported in 2026 that data centers make up about 55 percent of its power use, and new capacity arrives slowly. In that setting, a lower facility overhead converts directly into rack power. As an illustration, a 1-megawatt hall whose power usage effectiveness falls from 1.35 to 1.15 frees 200 kilowatts, about 1.75 million kilowatt-hours a year, worth roughly $377,000 at the state’s 2024 industrial average; the figure rests on stated assumptions, and Paul Industries runs the same calculation with the owner’s tariff and load.
The general principles of cold plates, coolant distribution and immersion are set out on the data center liquid cooling page. This page deals with the California side: the climate a loop rejects heat into, the code rules that touch the plant, the seismic restraint that every manifold needs, and the reporting that will make a hall’s efficiency public.
Liquid-cooled computing already running in California
Data centers in California already use liquid cooling on systems as large as an exascale supercomputer, and colocation operators across the Bay Area and Southern California are adding it to suites as rack densities rise past what air can carry. Two federal laboratories show how far the approach has gone.
El Capitan, at Lawrence Livermore National Laboratory in Livermore, is a direct liquid-cooled exascale system that entered the November 2024 TOP500 list in first place at 1.742 exaflops and held that place on the November 2025 list. At Lawrence Berkeley National Laboratory, the National Energy Research Scientific Computing Center in Shyh Wang Hall runs with no mechanical chillers at all: wet cooling towers deliver water at 75 degrees Fahrenheit or colder all year to the liquid-cooled systems, according to the center’s own published material. Those buildings are market facts about what the Bay Area climate allows, and they set a benchmark that commercial owners now study.
Commercial adoption follows the hardware. Accelerated servers with cold plates arrive from the factory expecting a facility water supply within a stated temperature and flow range, so colocation providers in Santa Clara, San Jose, Los Angeles and San Diego are retrofitting halls with coolant distribution units and facility water branches, and new buildings are designed for liquid from the first drawing.
How does data center liquid cooling work in a California hall?
Data center liquid cooling works by carrying heat away from processors in a sealed coolant loop, handing it to a facility water loop in a coolant distribution unit, and rejecting it outdoors through dry coolers, cooling towers or chillers. The coolant that touches the hardware never mixes with the building’s water, and the coolant distribution unit controls its temperature, pressure, flow and filtration.
The technology cooling loop runs from the coolant distribution unit through rack manifolds and quick disconnects to the cold plates. The facility water loop runs from the coolant distribution unit to the heat rejection plant. In California’s coastal climate the facility loop can run warm, often in the 80s or 90s Fahrenheit, which is the temperature range where dry coolers carry the load on most days. Inland sites such as Sacramento and Riverside see July normal highs of 94 and 93 degrees, so their dry coolers either run at higher loop temperatures or get help from trim chillers or adiabatic assist during the hottest hours.
| ASHRAE class | Facility water supply limit | Typical heat rejection | California fit |
|---|---|---|---|
| W17 | 17 C (63 F) | Chillers with economizer support | Any region; compressors run often in Southern California summers |
| W27 | 27 C (81 F) | Cooling towers or dry coolers with chiller backup | Bay Area and coastal Southern California for most hours |
| W32 | 32 C (90 F) | Dry coolers or towers, chillers for peaks only | Silicon Valley, East Bay and San Diego with few chiller hours |
| W40 | 40 C (104 F) | Dry coolers without compressors in most climates | Most of California, including inland valleys outside heat waves |
| W45 and W+ | 45 C (113 F) and above | Dry coolers; heat warm enough to reuse | Statewide, including candidate heat reuse projects |
The class is set by the server maker’s facility water specification, not by the building owner’s preference, so the first design step is to collect the hardware requirements for every rack type that will land in the hall. Paul Industries designs the facility loop to the warmest supply temperature the hardware accepts, because each degree upward adds economizer hours in California.
Data center liquid cooling vs air cooling under the California Energy Code
Data center liquid cooling vs air cooling looks different in California because the Energy Code already regulates the air side of a computer room. Section 140.9(a) of the 2025 Energy Code requires an integrated economizer, caps total fan power at 27 watts per thousand Btu per hour of net sensible cooling, and requires hot or cold aisle containment where air-cooled racks exceed 10 kilowatts of design load in a room.
Liquid cooling changes how those rules land. A water economizer under the code must carry the full load at 50 degrees Fahrenheit dry bulb or 45 degrees wet bulb, a test that a warm-water loop passes with wide margin, and the fans that remain in the room handle only the share of heat that still leaves through air. Hybrid halls keep both systems: cold plates take most of the processor heat, and air handling sized for the remainder removes heat from memory, power supplies and storage. Paul Industries sizes the air side for that residual load, with containment where the air-cooled share of a room still exceeds the code threshold.
| Factor | Air cooling | Liquid cooling |
|---|---|---|
| Practical rack density | Limited by airflow and aisle containment | Far higher per rack with cold plates or immersion |
| Facility temperature | Chilled water typically in the 50s or 60s Fahrenheit | Warm water, often in the 80s or 90s, set by the hardware class |
| Compressor hours in coastal California | Moderate, reduced by economizers | Few or none in many years |
| Fan energy | Large; capped by the code’s fan power limit | Smaller, for the residual air load only |
| Water use | Depends on towers or evaporative air handlers | Very low when dry coolers reject the heat |
| Seismic detailing | Air handlers, ducts and chilled water piping | Adds coolant distribution units, manifolds and hoses in the white space |
Data center liquid cooling options for California rooms
The data center liquid cooling options for California rooms are direct-to-chip cold plates with liquid-to-liquid coolant distribution units, cold plates with liquid-to-air units for rooms that lack facility water, rear door heat exchangers, and single-phase immersion tanks. Each one suits a different mix of building, tenant and hardware.
Liquid-to-liquid coolant distribution units on a facility water loop give high efficiency and high density, and they are the usual choice for new halls and for retrofits where facility water can be extended to the room. Liquid-to-air units reject the rack heat into the room’s existing cooling, which lets a tenant run a few cold plate racks in an air-cooled suite without new piping, at the cost of loading the room’s air handlers. Rear door heat exchangers remove rack exhaust heat with chilled or warm water at the back of each cabinet, which suits mixed suites with moderate densities. Immersion tanks hold whole servers in dielectric fluid, suit high-density clusters on slab floors, and need floor loading checks, fluid handling procedures and the data center immersion cooling scope. Paul Industries supplies and installs each option and often combines two in one hall.
Liquid cooling data center design for seismic California
A liquid cooling data center design in California has to keep coolant contained through an earthquake, because a sheared hose or manifold over a row of servers causes far more damage than the shaking itself. Seismic loads on that equipment come from ASCE 7-22, the standard the 2025 California Building Code has referenced since January 1, 2026, and its Chapter 13 treats coolant distribution units, manifolds and facility water piping as nonstructural components with their own design forces.
Paul Industries anchors coolant distribution units to the floor or slab to their calculated forces, braces facility water mains and branches, and provides flexible connections where piping crosses from the building structure to raised floors, rack rows or equipment on isolators. Rack manifolds are secured to the rack frame, hoses are routed with slack for relative movement between racks, and leak detection cable runs under every manifold and coolant distribution unit, with alarms that close isolation valves on a confirmed leak. Overhead piping above live racks gets drip trays where the owner prefers, and every joint above the white space is pressure tested before coolant is introduced.
Coolant, fill water and filtration
Technology cooling loops usually run on a propylene glycol and water mixture or a treated water solution specified by the hardware maker, and the quality of the make-up water matters because dissolved minerals and particles foul cold plate microchannels. Paul Industries fills and flushes technology loops with deionized or purified water to the hardware maker’s specification, adds the specified inhibitor package, and installs filtration at the coolant distribution unit to the cold plate maker’s micron rating. Sites that also run process or laboratory water treatment can draw on the approaches described on ultrapure water systems in California.
Heat rejection for California liquid loops
Heat rejection for California liquid loops is usually a dry cooler yard with trim chillers or adiabatic assist for heat waves, because warm facility water lets air cool the loop most of the year without using water. Bay Area and coastal Southern California sites may need trim cooling only for a handful of afternoons; Sacramento, Fresno and the Inland Empire need more, and El Centro-type desert heat needs mechanical cooling sized for long periods.
Where water is available and acceptable to the community, cooling towers run the loop at lower temperatures with less fan power, as Berkeley Lab’s chiller-free plant demonstrates. The choice now carries reporting weight: under AB 1577, signed in September 2026, qualifying data centers report energy and water use, power usage effectiveness and water usage effectiveness to the California Energy Commission, with provisions generally operative from January 1, 2027. A dry cooler plant reports a low water figure, a tower plant reports lower energy, and Paul Industries models both for the site before the owner commits. The air-side plant, economizer rules and water disclosure law are covered in more depth on data center cooling in California.
Dry cooler yards in California also need attention to sound and to what settles on the coils. Large fan arrays near homes in San Jose, Hayward or the Inland Empire run against city noise limits at the property line, so fan selection, speed control at night and acoustic screening are part of the design. Inland yards collect dust, pollen and, in fire seasons, ash on their coil faces, which raises approach temperature and cuts capacity just when heat waves arrive. Paul Industries specifies coil spacing and access for cleaning, adds coil cleaning to the maintenance plan before each summer, and selects dry cooler capacity with a fouling margin so the loop holds its supply temperature through the hottest week of the year.
Bringing facility water into an occupied California suite
Bringing facility water into an occupied California suite for liquid cooling means working in a room full of live equipment, and the work is planned around the tenants who stay online. The usual sequence is a facility water branch with isolation valves installed from outside the suite or during a planned window, coolant distribution units set and anchored, overhead or underfloor distribution to rack rows, and manifold connections made rack by rack as hardware arrives.
Paul Industries connects new branches to live chilled water or condenser headers with hot taps or line stops so the existing plant never loses redundancy, pressure tests and flushes new piping before it is opened to the system, and commissions each coolant distribution unit with its controls, alarms and leak detection before the first rack is connected. The integrated tests that prove the expanded plant are described on data center commissioning in California, and the wider mechanical package for new halls on data center mechanical contractor in California.
Where does the landlord’s plant end and the tenant’s loop begin?
The landlord’s plant usually ends at a valved facility water connection in or beside the suite, and the tenant’s scope begins at the coolant distribution unit, but California colocation leases draw that line in several places, and the drawing set has to show it clearly before work starts. Three boundary arrangements are common. In the first, the operator provides facility water to a set of valved stubs and the tenant supplies coolant distribution units, manifolds and hoses. In the second, the operator owns and maintains the coolant distribution units and delivers technology loop water to each rack at an agreed temperature and flow. In the third, used for whole-hall leases, the tenant takes everything from the heat rejection yard inward.
Each arrangement changes who sets the water temperature, who answers a leak alarm at three in the morning and who holds the records an auditor asks for. Paul Industries works to whichever boundary the lease sets, labels the demarcation valves in the field, and hands over separate records for the operator’s plant and the tenant’s loop so maintenance responsibility is never in doubt. Where an owner is still writing the lease, Paul Industries describes the cost and control implications of each option so the commercial terms match the engineering.
Startup checks before the first liquid-cooled rack goes live
Startup checks before the first liquid-cooled rack goes live prove that the loop is clean, tight and controlled, because debris or air left in a technology loop reaches cold plates within minutes of startup. The sequence Paul Industries follows starts with a hydrostatic or pneumatic pressure test of all new piping, then a high-velocity flush through temporary bypasses until strainer and filter checks run clean, then fill with the specified fluid and air removal at every high point.
With fluid in the loop, the coolant distribution unit’s pumps, valves and sensors are checked against their setpoints, alarms are triggered and confirmed at the building management system, leak detection is tested by wetting each sensing cable, and the unit is run against a load bank or the first hardware at partial load. Supply temperature, differential pressure and flow are recorded at each step and form part of the turnover file with the fluid chemistry report.
Pricing a California liquid cooling project
Data center liquid cooling cost in California is driven by the number and capacity of coolant distribution units, the length and size of facility water distribution, rack manifold and hose counts, heat rejection equipment, seismic anchorage and bracing, leak detection and controls, and whether the work is new construction or a retrofit in an occupied suite. Prices for the coolant distribution units and dry coolers also respond to manufacturer lead times.
| Driver | Effect on cost | Information that firms up the price |
|---|---|---|
| Rack count and heat load per rack | Sets coolant distribution unit capacity and manifold count | Hardware models, kilowatts per rack, liquid and air split |
| Facility water class | Warmer supply shrinks or removes chiller capacity | Server maker’s facility water specification |
| Distribution route | Overhead, underfloor or in-row changes labor and bracing | Room drawings, ceiling and floor heights |
| Heat rejection | Dry coolers, towers or trim chillers, plus yard or roof structure | Site plan, roof capacity, water source |
| Seismic restraint | Anchors, bracing and flexible connections under ASCE 7-22 | Building structure, floor type, site location |
| Occupied suite work | Hot taps, night shifts and phased connections | Tenant rules, allowed outage windows, redundancy |
Paul Industries gives a budget range from comparable scope once rack counts and hardware are known, then a firm price after a site survey. Owners choose supply only of coolant distribution units, manifolds and dry coolers, design-build from a rack and efficiency target, or complete installation from the facility water tap to the last quick disconnect.
Who installs data center liquid cooling in California?
Paul Industries installs data center liquid cooling in California, designing, supplying, installing, testing and servicing coolant distribution units, facility and technology water loops, rack manifolds, rear door heat exchangers, immersion systems, dry coolers and trim chillers. Paul Industries is headquartered at 256 Shore Drive in Kilmarnock, Virginia, and takes liquid cooling work in Santa Clara, San Jose, Fremont, the Livermore Valley, Sacramento, Los Angeles, the Inland Empire and San Diego, as well as in all 50 states.
The piping that carries facility water follows the same welding, testing and documentation practice described on process piping in California. Owners with halls on more than one coast can compare this page with data center liquid cooling in Virginia, the headquarters state, and data center liquid cooling in Florida. The West Coast page describes work in the other Pacific states.
Next step for liquid-cooled racks in California
Paul Industries builds liquid cooling for California racks from the facility water connection through the coolant distribution unit and manifolds to the heat rejection yard, with seismic bracing, leak detection, flushing, fill and commissioning records in one package. Send the hardware models and kilowatts per rack, the facility water specification, room drawings and the site’s heat rejection options. Call 201-450-8280 or request a quote. Other process, water and piping systems Paul Industries builds statewide appear on the California page.
Standards referenced: 2025 California Energy Code (Title 24, Part 6), Section 140.9(a) · 2025 California Building Code (Title 24, Part 2) and ASCE 7-22, Chapter 13 · 2025 California Mechanical Code (Title 24, Part 4) · ASHRAE TC 9.9 liquid cooling classes (W17 to W+) · ASME B31.9 Building Services Piping · AB 1577 (2026 data center energy and water reporting)
Frequently asked questions
Who does liquid cooling for data centers in the East Bay and the Livermore Valley?
Paul Industries does liquid cooling for data centers in Oakland, Hayward, Fremont, Livermore and Pleasanton, from coolant distribution units and rack manifolds to dry coolers and facility water piping. The East Bay’s marine climate keeps warm-water loops on free cooling for most of the year, so Paul Industries designs those plants around dry coolers with trim cooling sized for the few hot afternoons.
Does Paul Industries install liquid cooling in Santa Clara and San Jose colocation suites?
Yes. Paul Industries extends facility water into occupied suites in Santa Clara and San Jose, sets and anchors coolant distribution units, and connects rack manifolds as tenant hardware arrives. Work in live buildings follows the operator’s method of procedure, with hot taps on existing headers so the building keeps its redundancy throughout.
Who installs liquid cooling for dense racks in Los Angeles and San Diego?
Paul Industries installs liquid cooling for colocation and enterprise rooms in Los Angeles, Orange County and San Diego, including buildings converted from office or industrial use. Older structures need floor loading checks for coolant distribution units and immersion tanks, and roof or yard space for dry coolers, so Paul Industries surveys structure and routing before the design is fixed.
Are data centers water cooled in California, or air cooled?
Both, and many halls now use each. Most existing California rooms are air cooled with chilled water or direct expansion units, while new and retrofitted halls for accelerated computing add liquid cooling for the densest racks. Some research centers, including Berkeley Lab’s NERSC, cool liquid-cooled systems with tower water and no chillers at all.
What facility water temperature should a Silicon Valley liquid-cooled hall run?
The warmest the hardware allows. Server makers specify the facility water supply range for their cold plates, often using ASHRAE classes such as W32 or W40. In Silicon Valley, where San Jose’s July nights average 58 degrees Fahrenheit, a loop in the 80s or 90s rejects heat through dry coolers with few or no chiller hours, which lowers energy use and frees power under the building’s allocation.
Can a warm-water liquid loop satisfy the California computer room economizer rule?
A water economizer under Section 140.9(a) must carry the full expected load at 50 degrees Fahrenheit dry bulb or 45 degrees wet bulb and below. A warm-water loop served by dry coolers or towers passes that test with wide margin because it rejects heat at far higher temperatures than those thresholds. The air side of the room still follows the code’s fan power and containment rules for its share of the load.
How is liquid cooling piping braced for earthquakes in a California hall?
Coolant distribution units are anchored to the slab or floor structure for the forces set by ASCE 7-22 Chapter 13, which the 2025 California Building Code adopts. Facility water mains and branches get sway bracing, flexible connections are used where piping meets racks or isolated equipment, and hoses at rack manifolds are routed with slack. Leak detection under manifolds closes isolation valves on a confirmed leak.
What does California's 2026 reporting law mean for a liquid-cooled data center?
AB 1577, signed in September 2026, sets up reporting to the California Energy Commission that covers energy and water use, power usage effectiveness and water usage effectiveness, with provisions generally operative from January 1, 2027. Liquid-cooled halls with dry coolers usually report favorable figures on both measures, and Paul Industries provides metering points in the cooling plant so the reported numbers come from measurement.
Is glycol needed in a liquid loop serving a coastal California hall?
The technology loop usually carries the coolant the hardware maker specifies, often a propylene glycol solution or treated water. The facility loop needs freeze protection only where dry coolers or outdoor piping see freezing temperatures, which in California means mountain and high desert sites and occasional cold snaps in the Central Valley. Coastal facility loops often run on treated water with drain-down or heat tracing for exposed sections.
Can liquid cooling heat be reused at a California campus?
At warm-water classes such as W40 and W45, return water is hot enough to preheat domestic water, warm buildings or serve absorption equipment, and California’s high energy prices improve the payback. Reuse works best when a heat user sits beside the hall, such as a campus or laboratory building. Paul Industries evaluates the load match and piping cost before recommending it.
What water is used to fill a technology cooling loop?
Technology loops are filled with deionized or purified water mixed with the inhibitor or glycol package the hardware maker specifies, because minerals and particles in tap water foul cold plate microchannels. Paul Industries flushes the new loop until it runs clean, fills it to specification, records the fluid chemistry at turnover and installs filtration at the coolant distribution unit.
How long do coolant distribution unit pumps need to run on backup power?
Coolant pumps must keep moving coolant for as long as the processors produce heat, so they run on the same uninterruptible power as the servers or on a dedicated UPS sized to carry them until generators pick up the load. Cold plates hold very little coolant, so a pump stop raises chip temperatures within seconds. Paul Industries confirms the power path for every coolant distribution unit during design.
Does Paul Industries supply coolant distribution units without installing them?
Yes. Paul Industries offers supply only of coolant distribution units, rack manifolds, dry coolers and trim chillers, design-build of a complete liquid cooling plant from a rack and efficiency target, and full installation including facility water piping, seismic bracing, leak detection, flushing, fill and commissioning. Many owners start with supply only for a pilot row and move to full installation for the hall.
What sets per-rack liquid cooling pricing in a California suite?
Per-rack cost depends on rack heat load, the liquid cooling method, coolant distribution unit sharing, distribution length, heat rejection and whether the suite is occupied. Seismic bracing and leak detection add more in California than in many states. Paul Industries prices from rack counts, hardware specifications and room drawings, giving a budget range first and a firm price after a survey.
What maintenance does a California liquid cooling system need after turnover?
Routine work covers coolant sampling and chemistry adjustment, filter changes at the coolant distribution unit, pump and sensor checks, leak detection tests, quick disconnect inspection and dry cooler coil cleaning, which matters more where dust or wildfire ash settles on coils. Paul Industries offers that program under a service contract with records the operator keeps for tenants and auditors.
Do immersion tanks need anything special in a California building?
Immersion tanks are heavy when filled, so floor loading and slab design come first, and they need seismic anchorage like any other tank of their weight. Fluid storage, spill containment, handling procedures and fire code review of the dielectric fluid follow. Paul Industries designs the anchorage, facility water connection and heat rejection for the tanks and installs them with the rest of the plant.
Which documents start a liquid cooling design for a California data hall?
Send the hardware models and kilowatts per rack, the liquid and air split, the server maker’s facility water specification, room drawings with floor and ceiling details, the site’s heat rejection options and any existing plant drawings. For occupied suites, add the operator’s rules for work windows. Paul Industries returns a design outline and budget range.
Bringing liquid-cooled racks into a California hall?
Send the hardware models, kilowatts per rack, facility water specification and room drawings. Call 201-450-8280 or request a quote.
