AaronAlphaX
Aerial view of a liquid-cooled data center at night
③ Showcase · Consulting

Compute Strategy & Liquid-Cooling DeploymentGlobal Compute Strategy & China Liquid Cooling Deployment

Global Compute Strategy Consulting for enterprises and agents planning AI capacity — from site and cooling choices to Liquid Cooling Data Center Deployment and the supporting services around it.

Why liquid cooling

Rising chip and memory power is pushing AI racks past what air can cool; ASHRAE TC 9.9 recommends that new data centers plan for liquid cooling. We help you decide where, how and with whom to deploy it.

Our Compute Infrastructure Consulting covers cold-plate and immersion options, coolant distribution, facility integration and vendor selection — including China Liquid Cooling Solution suppliers where they fit your requirements — through to Global Compute Implementation.

Global Compute Strategy ConsultingCompute Infrastructure ConsultingChina Liquid Cooling SolutionLiquid Cooling Data Center DeploymentDynamic Data Center ShowcaseCompute Infrastructure Case StudiesWhitepapersGlobal Compute ImplementationSupporting ServicesData Center God-View Showcase

Data Center God-View Showcase

A dynamic, top-down view of an AI data hall — the format we use to walk stakeholders through layout, liquid distribution and capacity.

Aerial view of a liquid-cooled data center at night
Illustrative visual · not a client facility
Rows of server racks in a large data centre
Illustrative photography
Close-up of server cooling fans
Illustrative photography

Consulting & supporting services

Global Compute Strategy Consulting

Capacity planning, region and site selection, build-vs-buy and partner strategy.

Compute Infrastructure Consulting

Rack density, power and cooling architecture reviews for AI workloads.

Liquid Cooling Solution Selection

Cold plate, immersion, CDU and heat-reuse options compared against your site.

Liquid Cooling Data Center Deployment

Deployment planning, supplier coordination and commissioning support.

Case Studies & Whitepapers

Public standards and whitepapers from OCP and ASHRAE as the reference base.

Supporting Services

Supplier matching (②), API capacity (①) and written agreements around the project.

From strategy to deployment

Assess

Workloads, density targets, sites and constraints.

Design

Cooling architecture and reference designs.

Source

Vendor shortlist and screening via ②.

Deploy

Implementation, commissioning and handover support.

Engagement path

From readiness to commissioned capacity

  1. 01

    Experience

    An infrastructure readiness snapshot: site, power, cooling and density constraints reviewed against your plans.

  2. 02

    Full Solution

    A tailored architecture and proposal: power pathway, cooling design, partial-load requirements and a vendor RFP with deal-breaker screen.

  3. 03

    Integrate & Scale

    Owner-side oversight through deployment and commissioning, colocation sourcing and interim API capacity, with regular reporting.

Engagement tiers

Consulting tiers: Essential · Professional · Enterprise

01

Essential

Readiness snapshot

Scope
Readiness review and decision memo
Support
Workshop plus email follow-up
SLA
Agreed delivery dates
Regions
One site or region
Integration
Recommendations only
Request proposal
Most requested02

Professional

Architecture & RFP

Scope
Power and cooling architecture with vendor RFP
Support
Named lead engineer
SLA
Written deliverables and acceptance criteria
Regions
Multiple sites or regions
Integration
Vendor evaluation and recommendation
Request proposal
03

Enterprise

Owner-side advisor

Scope
End-to-end program through commissioning
Support
Dedicated team and bi-weekly reporting
SLA
Written SLA
Regions
US · GCC · Southeast Asia colocation options
Integration
Deployment oversight plus interim API capacity via ①
Request proposal

No public price list: every proposal is scoped and quoted in writing.

Insights · Whitepapers

Decision briefs from the field.

Two anonymized briefs showing how we structure engagements. Key points and outcomes only — the working methods stay in the engagement.

Illustrative engagement briefAnonymized · no client dataCompute strategy · ③

500 kW Modular AI Inference Build — Power & Liquid-Cooling Decision Brief

An AI inference and LLM research environment planned at 500 kW over two to three years, starting with B300-class systems and planned to take Rubin-generation systems later.

01.1

Phase the build, not the ambition

Pathways, space and interfaces are sized for the full build-out; pods are added phase by phase without outages to live systems, and next-generation readiness is re-confirmed against vendor roadmaps at each phase.

01.2

Keep the 800 VDC door open

Three power options are assessed: 415/480 V AC today; AC today but 800 VDC-ready; or early 800 VDC adoption. NVIDIA has signalled 800 VDC as the direction for next-generation AI factories, so the output is a recommendation with its rationale and decision triggers for revisiting it.

01.3

Choose the cooling split at 120–140 kW per rack

Direct-to-chip liquid cooling with in-row or in-rack CDUs is compared with a hybrid liquid-plus-air design, using the actual system SKUs: liquid/air heat split, residual air-side capacity, CDU modularity by phase and facility-water temperature strategy.

01.4

Contract for partial-load efficiency

Day-1 load is a small fraction of design load, so efficiency requirements are set for low load as well as full load — UPS and CDU behaviour and how redundancy is structured — and written into the vendor RFP.

Size the pathways for 500 kW. Size the spend for Day 1.

Exhibit 1Phased build-out to 500 kW
Phase 1Initial B300-class pods live; infrastructure sized for full build-out where cost-effective
Phase 2Additional pods added without outage to Phase 1 systems
Phase 3Up to 500 kW; Rubin-generation readiness confirmed

Phase sizes are set per engagement; bars are illustrative. Source: AaronAlphaX illustrative engagement brief.

Exhibit 2Power pathway options
OptionWhat it keeps openMain risk to manage
A · 415/480 V ACMatches current B300-class requirements and supply chainLater move to DC may mean rebuilding the electrical room
B · AC today, 800 VDC-readySpace, pathways and interfaces provisioned to convert laterUpfront provisioning must be justified by decision triggers
C · Early 800 VDCNative DC from day oneEquipment maturity, vendor support and timing

Source: AaronAlphaX illustrative engagement brief; NVIDIA Technical Blog, “NVIDIA 800 VDC Architecture Will Power the Next Generation of AI Factories” (May 2025).

Engagement outcomes (deliverables)

  1. Requirements baseline and capacity model
  2. Power and cooling architecture recommendation
  3. Vendor RFP package with deal-breaker criteria
  4. Vendor evaluation and recommendation
  5. Colocation shortlist and optional interim API capacity

Illustrative engagement briefAnonymized · no client dataAI search visibility

GEO / AI Search Visibility — 90-Day Program Framework

How a brand improves how accurately and how often it appears in AI-generated answers across ChatGPT, Perplexity, Google Gemini and Google AI Overviews — measured against an agreed set of buyer-intent prompts.

02.1

Baseline before you build

An AI visibility audit records whether the brand is mentioned, how it is described and which sources are cited, compared with a small set of named competitors.

02.2

Fix the entity, then the content

Structured-data and brand-fact fixes help AI systems recognise the brand as an entity; a content and citation-source plan then targets the owned and third-party sources the category relies on.

02.3

Report trends, not promises

The agreed prompt set is measured every month against the Day-0 baseline. Visibility depends on third-party AI platforms, so outcomes are reported, never guaranteed.

Success is a measured trend against the Day-0 baseline — not a guaranteed placement.

Exhibit 3The 90-day program
Days 1–30 · BaselinePrompt set agreed; audit and baseline delivered; entity and content plans approved
Days 31–60 · BuildPriority entity fixes live; first content and source actions; first tracking report
Days 61–90 · Expand & measureSecond content and source cycle; Day-90 report against baseline; next-phase recommendation

Source: AaronAlphaX illustrative engagement brief.

Plan your next AI data hall.

Book a strategy consultation →