Compute Center Construction

From national centers to distributed compute networks.

Project delivery for national-scale compute centers, regional and industry infrastructure, distributed systems, and custom hardware and software. Align facilities, clusters, networks, storage and operating platforms around workloads and measurable acceptance criteria.

Discuss your project

Illustrative infrastructure scene, not a photograph of the Ghana project.

Three ways we support your build

National-scale centers

Site assessment, capacity planning and phased construction for public services, research and industry, integrating power, cooling, networks, storage and dedicated inference platforms.

Distributed compute

Connect central, regional and edge resources through resource pools, scheduling, fault isolation and remote operations, with energy coordination, metering and staged expansion.

Custom hardware & software

Specify GPU/CPU servers, networks and storage around real workloads. Integrate model runtimes, APIs, tenant management, billing, security audit and business interfaces.

Compute centers: from planning to rack integration

Plan compute equipment, cooling, power and digital operations together, with interfaces, capacity and acceptance tied to real workloads.

Digital twins and spatial planning
Relate room models to rack positions, equipment lists and operations views to review layouts, construction interfaces and changes.
Open full reference image (new tab)

Rack integration and cooling
Plan rack support around server power, cooling and maintenance access. Confirm the cooling design through compatibility checks and site tests.
Open full reference image (new tab)

Power-distribution architecture reference
The original names a Visionbay.ai 800 VDC reference. Use it to discuss upstream supply, redundant conversion and rack distribution; selection, permits and protection require project validation.
Open full reference image (new tab)

Server modules and system integration
Check compute, network, storage and thermal interfaces from accelerator modules to complete servers. Models and delivery configurations follow procurement and acceptance records.
Open full reference image (new tab)

User-supplied third-party technical references, not completed Xinan projects or proprietary product photographs. Original marks are retained; no partnership or endorsement is implied. English text and values such as 800 VDC, 4 MW and 16 MW belong to the illustrated reference only, not a project commitment.

Explore DCP compute-energy coordination

02 / DELIVERY

From defined requirements to verified handover

Scope & feasibility

Define users, workloads, data and service targets. Establish land, power, connectivity and funding conditions.

Design & validation

Develop architecture, equipment lists and interface boundaries. Validate capacity, power and cooling with pilot tests and site data.

Procurement & integration

Advance procurement, construction and integration through approval and contract gates, tracking supply, quality and configurations.

Acceptance & handover

Coordinate factory, site and integrated systems tests. Deliver as-built records, training and pilot-operation validation.

Scope is defined by contract. Qualified parties separately handle local design sign-off, permits, utility connections and equipment export/import requirements. Facility ratings, performance and availability remain subject to design review and acceptance testing.

Ghana Star Initiative

National compute center pilot · Planning case

A modular compute center concept for dedicated public-sector inference, enterprise APIs, research and interruptible batch workloads in Ghana. The plan coordinates facilities, energy and operating systems through staged validation and conditional release gates.

Figures are design capacities from plan V4.4, not purchased, completed or operational capacity. Project-specific approvals, contracts and final configurations require verification at implementation gates.

1,024

Planned GPUs

128 eight-GPU servers, including reserves

128

Planned servers

AMD-preferred route; selection to be finalized

4 × 32

Compute modules

32 servers per module; staged validation

1.736

MW planned IT load

Design value, not measured operation

01
Compute & data
Four compute modules; training/RDMA, storage and out-of-band networks; shared flash, object storage and backup.
02
Facilities & energy
Building and MEP systems, A/B UPS paths, cooling, fire protection, security and metering. Solar and storage are separate optional packages.
03
Services & operations
Tenant portal, model deployment, scheduling, billing and security audit, with compute-energy controls and operational validation.

Roles described in the plan: DROLMA GROUP LLC handles core supply and final technical confirmation; Hangzhou Xinan Intelligent Technology Co., Ltd. handles China-side supporting engineering and construction organization; the Ghana project company handles land, utilities, approvals and operations. Executed contracts and locally signed documents determine final responsibilities.

Source: Ghana Star Initiative complete project plan V4.4, baseline 2026-09-08, pp. 3–6, 12, 63 and 71. Planning summary only; full financing, pricing and cost-model appendices are not published here. This is not a claim of government endorsement, completed delivery or guaranteed returns.

Discuss your project

Share your location, users, workloads, target capacity and available site conditions so we can discuss scope, customization and the next validation steps.

Discuss your project

Loading page…

Compute Center Construction

From national centers to distributed compute networks.

Discuss your project

Three ways we support your build

National-scale centers

Site assessment, capacity planning and phased construction for public services, research and industry, integrating power, cooling, networks, storage and dedicated inference platforms.

Distributed compute

Connect central, regional and edge resources through resource pools, scheduling, fault isolation and remote operations, with energy coordination, metering and staged expansion.

Custom hardware & software

Specify GPU/CPU servers, networks and storage around real workloads. Integrate model runtimes, APIs, tenant management, billing, security audit and business interfaces.

Compute centers: from planning to rack integration

Plan compute equipment, cooling, power and digital operations together, with interfaces, capacity and acceptance tied to real workloads.

Explore DCP compute-energy coordination

02 / DELIVERY

From defined requirements to verified handover

Scope & feasibility

Define users, workloads, data and service targets. Establish land, power, connectivity and funding conditions.

Design & validation

Develop architecture, equipment lists and interface boundaries. Validate capacity, power and cooling with pilot tests and site data.

Procurement & integration

Advance procurement, construction and integration through approval and contract gates, tracking supply, quality and configurations.

Acceptance & handover

Coordinate factory, site and integrated systems tests. Deliver as-built records, training and pilot-operation validation.

Ghana Star Initiative

National compute center pilot · Planning case

1,024

Planned GPUs

128 eight-GPU servers, including reserves

128

Planned servers

AMD-preferred route; selection to be finalized

4 × 32

Compute modules

32 servers per module; staged validation

1.736

MW planned IT load

Design value, not measured operation

01
Compute & data
Four compute modules; training/RDMA, storage and out-of-band networks; shared flash, object storage and backup.
02
Facilities & energy
Building and MEP systems, A/B UPS paths, cooling, fire protection, security and metering. Solar and storage are separate optional packages.
03
Services & operations
Tenant portal, model deployment, scheduling, billing and security audit, with compute-energy controls and operational validation.

Discuss your project

Share your location, users, workloads, target capacity and available site conditions so we can discuss scope, customization and the next validation steps.

Discuss your project

Compute Center Construction

Three ways we support your build

National-scale centers

Distributed compute

Custom hardware & software

Compute centers: from planning to rack integration

Digital twins and spatial planning

Rack integration and cooling

Power-distribution architecture reference

Server modules and system integration

From defined requirements to verified handover

Scope & feasibility

Design & validation

Procurement & integration

Acceptance & handover

Ghana Star Initiative

Compute & data

Facilities & energy

Services & operations

Discuss your project

Selection and delivery

Hardware and everyday applications

Compute halls and cooling

Firmware development bench

Compute Center Construction

Three ways we support your build

National-scale centers

Distributed compute

Custom hardware & software

Compute centers: from planning to rack integration

Digital twins and spatial planning

Rack integration and cooling

Power-distribution architecture reference

Server modules and system integration

From defined requirements to verified handover

Scope & feasibility

Design & validation

Procurement & integration

Acceptance & handover

Ghana Star Initiative

Compute & data

Facilities & energy

Services & operations

Discuss your project