Verdict (for procurement): Coolnetpower is a credible shortlist candidate when you want one vendor that can cover liquid-cooling building blocks (CDU, cold plate, immersion options, rear-door heat exchangers) and broader data center infrastructure—but your final “price per kW” will be quote-driven. Treat lead times, spares, and on‑site response as contract items that must be written into the MSA/SLA.
Key Takeaway: For liquid‑cooled AI/HPC colocation, “price per kW” is rarely a single number. It’s a stack: facility power + redundancy + liquid loop topology + commissioning + spares + service commitments.
Who this is for (and who it isn’t)
This review is for you if you’re:
A distributor or SI scoping 10–100+ kW racks (or pods) and need a procurement-ready way to talk about pricing, lead times, integration risk, and acceptance testing.
Trying to standardize a quote package (BOM + milestones + acceptance tests) for liquid-cooled deployments.
This review is not for you if you’re:
Looking for a fixed public “liquid cooling colocation rate card.” Real colocation pricing varies by site, redundancy tier, and contract structure.
Only buying small quantities with no need for integration, commissioning, or service terms.
What Coolnetpower covers (published scope)
Coolnetpower publishes a liquid cooling portfolio that includes CDUs, immersion, and rear‑door heat exchangers under its liquid cooling portfolio.
For compliance packs and vendor credibility statements, Coolnetpower’s certifications and company background reference ISO9001, ISO14001, CE, and CCC (request current certificates in your due diligence pack).
Liquid cooled AI colocation pricing: what drives $/kW
If you’re pricing liquid‑cooled AI/HPC colocation, the fastest way to avoid later surprises is to turn “$/kW” into a traceable price stack.
1) Facility baseline: directional $/kW/month (illustrative)
Even for standard air‑cooled deployments, many US colo deals are quoted as $/kW/month. QuoteColo publishes directional US market ranges in its Multi‑Rack Colocation Pricing guide.
Pro Tip: Pair $/kW with a “committed vs actual draw” table. QuoteColo’s average cost per rack ranges by density is a useful cross-check when stakeholders insist on seeing pricing as $/rack.
Illustrative example range (US, multi‑rack; not quotes):
Standard density (5–8 kW/rack): ~$95–$185 per kW/month
High density (10–20 kW/rack): ~$130–$230 per kW/month
2) Density reality check (so you pick the right cooling topology)
A practical non-vendor reference is DataCSI’s density framing: its cooling density bands guidance summarizes when air breaks down and when hybrid/liquid becomes necessary.
A simplified planning lens:
~20–30 kW/rack: rear‑door and row-based can still work (site‑dependent)
~30–60 kW/rack: hybrid approaches become common
~60–100+ kW/rack: direct-to-chip liquid cooling colocation designs become hard to avoid
3) The liquid‑cooling “delta”: price stack template you can reuse
Use this template in procurement reviews. It makes it obvious which part is a monthly service, which part is a one-time project, and which part is a risk item.
Cost element | What it includes | Typical pricing method | What to clarify in the quote |
|---|---|---|---|
Facility power + space | Power commit, space/cabinet/cage, base cooling allowance | $/kW/month or $/rack/month | Minimum commits, overage rules, what “included cooling” means |
Redundancy tier | A/B feeds, N+1 vs 2N, generator/UPS tier | uplift % or higher $/kW | What is guaranteed vs best-effort |
Liquid loop topology | RDHx vs D2C vs immersion; rack vs row CDUs | BOM / project | Who owns the CDU and secondary loop |
Commissioning & acceptance | FAT/SAT, leak tests, alarms integration, MOPs | one‑time services | Pass/fail criteria; rollback procedure |
Monitoring integration | BMS/DCIM alarms, telemetry points, reporting | project/services | Protocols, data points, escalation ownership |
Spares & service | Onsite spares kit, response times, annual PM | recurring services | SLA terms, exclusions, regional coverage |
⚠️ Warning: If a proposal doesn’t state where responsibility changes hands (facility water vs rack loop, alarm ownership, leak response), you don’t have a price—you have a dispute scheduled for later.
Lead times (typical/variable ranges, labeled as typical)
Because you requested planning ranges rather than exact promises, here’s a procurement-friendly way to discuss lead time without overcommitting.
Typical (illustrative) planning ranges:
CDU + manifolds + quick disconnects: ~8–20 weeks
Rear‑door heat exchanger hardware: ~6–16 weeks
Skid/pod plumbing + sensors + leak detection: ~6–18 weeks
Onsite commissioning window: days to weeks (depends on rack count, test scope, and change windows)
What to request in the quote package:
a milestone plan (order → factory test → ship → site‑ready → commissioning)
what’s stocked vs built-to-order
expediting options (and what expediting does not fix—e.g., site readiness)
Integration: PDU/UPS, monitoring, and what “done” looks like
Liquid‑cooled deployments fail in predictable places: unclear boundaries, missing alarms, and weak maintenance procedures.
Below is a checklist built for SIs and procurement.
PDU/UPS power-path questions
Can the CDU be powered from dual sources (A+B) where available? If not, what’s the failure mode?
What’s the CDU load (kW) and breaker plan per row/pod?
What’s the UPS ride‑through expectation for pumps/controls (seconds vs minutes)?
Monitoring and alarms
Which alarms are exported to BMS/DCIM (leak detection, flow, pressure, temperature, pump status)?
Who is on call for which alarm class—and what’s the escalation path?
Mechanical integration
Facility water interface: what supply/return temperatures and flow are required?
Do you need dripless quick disconnects and isolation valves for service windows?
Water quality plan: filtration, corrosion control, sampling responsibility.
Commissioning & MOPs
Are there written MOPs for filling, bleed, leak test, and rollback?
Are maintenance windows pre‑approved for staged cutovers?
If you want a vendor-published starting point to anchor scope language, Coolnetpower’s AI/HPC solution overview is the cleanest page to cite for how they position high‑density environments (treat performance numbers as proposal‑level items to validate).
Reference projects you can cite (published)
Coolnetpower publishes project references in its project gallery. One example with concrete scope detail is the Dubai cold aisle modular data center reference project, describing an integrated deployment (cabinets, in‑row cooling, modular UPS, power distribution) with N+1 redundancy.
If you need liquid‑cooling‑specific references for BOFU validation, request:
the cooling topology (RDHx vs D2C vs immersion)
rack/pod density band
commissioning scope and acceptance tests
which parts were Coolnetpower‑supplied vs SI‑integrated
Spares and on‑site SLA terms (how to structure the contract)
Public pages rarely define enforceable SLAs. Assume service is contract-defined.
Use this clause checklist to make support comparable across vendors:
SLA / spares item | What good looks like | What to ask for |
|---|---|---|
Response time | defined by severity (P1/P2/P3) | hours for remote, hours/days for onsite |
Onsite coverage | named regions + hours | coverage map + local partner details |
Spares kit | aligned to failure modes | pumps, sensors, connectors, controller modules |
RMA process | clear turnaround and loaners | advance replacement policy |
Preventive maintenance | scheduled + documented | PM checklist + frequency |
Acceptance & warranty | documented pass/fail tests | SAT scripts + warranty start condition |
Bottom line
Coolnetpower is worth shortlisting for liquid‑cooled AI/HPC colocation projects when you value integrated scope and want a vendor that can cover multiple cooling approaches.
The buyer-critical step is to force the quote into a structured package: price stack + lead-time milestones + integration boundaries + commissioning acceptance + SLA/spares.
Next step
To request a quote package you can compare cleanly, use the Coolnetpower Contact page and ask for:
a BOM aligned to your density band (for example, 30–60 kW hybrid vs 60–100+ kW D2C)
a milestone schedule with typical lead-time ranges by subsystem
a draft SLA/spares appendix you can attach to the MSA
