Vietnam
Manufacturing engine
Precision metal fabrication and volume assembly at Legend's ASEAN Advanced Manufacturing Centre — Viethan Electric, Hanoi and Nam Dinh.
AI infrastructure manufacturing & deployment partner
Engineering the physical infrastructure behind AI.
The rack is our platform, not our whole proposition. Legend engineers, manufactures and deploys every physical layer an AI hall needs — as one system, or as the single layer you are missing.
Designed for AI
Manufactured in Vietnam
Delivered globally from Singapore
- Manufacturing since
- 2000
- Plant area
- 8,000 m²
- Infrastructure layers
- 9
- Rack configurations
- 8
Why now
AI is scaling fast. Physical infrastructure is now the constraint.
Enterprises have moved from experimenting with AI to running it at scale — and that shift lands on racks, power, cooling and deployment capacity.
89%
of organisations report regular AI use in at least one business function
44%
say AI is scaling across the enterprise, up from 38% a year earlier
60%
expect to increase AI investment over the coming year
28%
say AI already consumes more than 10% of enterprise ICT budgets
Source: McKinsey Global AI Survey, 2025.
The market is moving from buying AI software to building the physical infrastructure needed to deploy AI at scale. That is precisely where Legend operates.
The model
Vietnam builds it. Singapore engineers it.
Manufacturing economics sit in Vietnam. Contracting confidence and engineering access sit in Singapore. Customers get both, four hours apart.
Singapore
Commercial and engineering hub
Global sales, application engineering, product management, regional inventory and an experience centre — rule of law, IP protection and open trade behind every contract.
Global
Deployment
Hyperscalers, data-centre operators, EPC and M&E contractors, system integrators and OEM partners across APAC, the Middle East, Europe and North America.
What we deliver
Nine layers, one partner, one specification.
Take the complete platform, or take only the layer you are missing and integrate it with what you already run.
House the hardware
- Rack 2.0 m frame, 42U and 48U
- Containment hot- or cold-aisle
- Cable management your structured cabling standard
Power and cool it
- Power PDU and busway integration
- Cooling air, hybrid or contained
- Liquid cooling RDHx, manifold, CDU-connectable
Run and deploy it
- Monitoring streams into your DCIM
- Security door access logging, leak detection
- Deployment delivered and commissioned on site
Take the complete platform
We engineer, build and deliver every layer as one system — one partner, one specification, one delivery to site.
Or take only the layers you need
Any layer can be supplied on its own and integrated with the racks, power and cooling you already run.
Manufacturing capability
8,000 m² built to data-centre tolerances.
Viethan Electric Co., Ltd — established 2000, Hanoi and Nam Dinh. Punching, welding, powder coating and assembly under one roof. A real factory, not a trading operation.
Engineering
Precision fabrication and finishing to data-centre tolerances.
Quality control
Consistent structural and finish standards across every production run.
Production scalability
Capacity to move from prototype to high-volume manufacturing.
Customisation
Configuration-driven manufacturing, not one-size-fits-all.
Rapid production
Fast quotation-to-delivery cycles — weeks, not quarters.
Cost efficiency
Competitive ASEAN economics without compromising engineering.
Photography: Viethan Electric Co., Ltd production facility, Nam Dinh, Vietnam.
Product architecture
One platform. Eight configurations. No redesign.
70–80% common components across all eight configurations. Find your cooling method and your rack density, then select the configuration that meets it.
Scroll the grid sideways to see every density band
Immersion-readysealed / tank
Liquid at the rackRDHx / manifold
Liquid-readyCDU-connectable
Air-cooledcontained aisle
Rack density per cabinet
Up to 10 kW
10–20 kW
20–40 kW
40–80 kW
80–150 kW
Select a configuration
Confirmed production SKUs: AI-R1 → VL-R42-8012 (Enterprise 42U) · AI-R4 → VL-R48-AI-SMART (Hyperscale 48U, RDHx-ready) · AI-R8 → VL-R48-IMM (immersion concept, 2027 roadmap).
Density bands are indicative planning figures, not commitments: regional operators are designing Southeast Asian halls for 10–150 kW per rack, and GPU environments with direct liquid cooling typically run 20–80 kW. AI-R8 is in development. Configure once, manufacture repeatedly, deploy globally.
Our products
One platform, from the edge to hyperscale.
The 42U and 48U racks share the same 2.0 m frame, so aisle planning and logistics do not change between them. The compact edge cabinet drops to 12U for branch sites and comms rooms.
Compact edge
VL MINI 12
600 mm W × 1000 mm D · 12U
- Passive airflow
- Standard power
- Basic monitoring
2000 mm
Enterprise rack 42U
VL R42-3-100MW
800 mm W × 1200 mm D
- Hybrid cooling
- Dual PDUs
- Rear-door heat exchanger
2000 mm
Hyperscale AI rack
VL R48 AI SMART
800 mm W × 1200 mm D
- Liquid cooling
- High density
- Flow metering and door access logging
Inside the rack
Water that recycles. Sensors that keep proving it.
The same sensors that prove performance at handover keep proving it for the life of the rack.
Closed-loop liquid cooling
AI-R3 and AI-R4 connect into a closed water loop — the principle behind district cooling, scaled down to a single cabinet. Heat is recovered at the rack instead of exhausted to the room.
Supply chilled water from facility CDU
AI-R3 / AI-R4 rack heat captured at the cabinet flow metered
Return warmer water back to CDU and heat rejection
Illustrative flow. Actual supply and return temperatures depend on CDU specification, confirmed at quotation against SS 697:2023 / SS 715:2025.
Monitoring and DCIM
Sensors throughout each rack stream live data to your DCIM or monitoring platform of choice — continuously, not just at commissioning. We do not ask you to adopt our platform.
- Inlet air temperatureFront-of-rack, zoned top, middle and bottom
- Outlet air temperatureRear exhaust, delta-T monitoring
- Relative humidityTracked within the SS 697:2023 envelope
- Coolant supply and returnTemperature and flow at the CDU manifold
- Power and current drawPer PDU, per outlet where fitted
- Access and leak detectionDoor events logged; coolant-loop leak sensing
Open architecture
AI infrastructure without vendor lock-in.
Our rack is the open frame in the middle. Standard dimensions, standard mounting, engineered to accept what you already buy.
Standard serversany rack-mount server, any manufacturer
Standard PDUsyour preferred power brand
Cooling systemsair, containment or hybrid
The Legend AI Rack Platform
Open frame · standard dimensions · standard mounting
Liquid cooling and CDUsthird-party loops and manifolds
Monitoring and DCIMintegrates with your platform
Cable managementyour structured cabling standard
No single-vendor stack. Your infrastructure investment stays flexible as it grows.
Responsible by design
Engineered so AI infrastructure doesn't burden the grid.
Responsible infrastructure isn't a certificate on a wall. It is designed into every rack before it ships.
Precision nesting cuts scrap and embodied carbon per rack.
Specified for Southeast Asian heat and humidity from day one.
Every 1 °C safely raised saves an estimated 2–5% in cooling energy.
Materials and finishes engineered to standard from the first production run.
Upgradeable racks extend service life and cut e-waste.
82%
of Singapore's ICT-sector energy use is data centres. Our job is to keep that number from climbing as AI racks scale.
Retrofit programme
Can your data centre support your next generation of AI?
Four questions decide whether you retrofit or rebuild. Two or more "no" answers usually means retrofit.
01
Rack density
What is your peak load per rack today, and for your next AI cluster?
Above roughly 15 kW per rack, air alone stops being viable.
02
Cooling path
Is chilled water or a CDU loop available to the hall, or is it air only?
Liquid-ready design is expected beyond roughly 30 kW per rack.
03
Power train
Are your PDUs and busway rated for planned density, not today's?
AI racks commonly draw two to four times legacy rack loads.
04
Floor and space
Can the slab and aisle clearances take a 2.0 m, fully loaded rack?
Plan for roughly 1,200–1,500 kg per populated rack.
Book a Legend site assessment
Density, cooling, power and floor-loading report within two weeks.
Who we serve
We supply the companies that build for the hyperscalers.
We don't compete for the hyperscaler's purchase order. Demand starts with the AI companies, but racks are bought by the four groups that specify and build the halls.
Where demand starts
AI and hyperscale companiesOwns the demand. Hardest to reach directly.
Where racks are bought
Data-centre operatorsRuns the halls the racks go into.
Where racks are bought
EPC and M&E contractorsDelivers the physical build.
Where racks are bought
System integratorsSpecifies and assembles the compute environment.
Where racks are bought
Solution providers and distributorsCarries stock and reaches the long tail.
Legend supplies every stage: racks · cooling-ready design · power and PDU integration · retrofit · OEM manufacturing.
Legend brand
A premium regional and global AI infrastructure brand, delivered under our name with Singapore application engineering behind it. Best for contractors and integrators.
OEM and private label
Manufactured under your brand, to your specification, from the same production line and the same quality system. Best for distributors, solution providers and emerging data-centre brands.
Why Legend
Weeks, not quarters — and no lock-in.
Large enough to manufacture at volume. Small enough to answer you the same day.
Speed
Fast quotation and delivery cycles, with no approval chain between you and the factory.
Flexibility
Configuration-driven customisation, without bespoke-engineering cost.
Cost
ASEAN manufacturing economics, with no compromise on quality.
AI specialisation
Purpose-built for AI density, not adapted legacy enclosures.
Vendor neutrality
Works across ecosystems, server brands and cooling platforms.
Also in the package: OEM manufacturing · Singapore application engineering · retrofit expertise · regional supply across Vietnam, Singapore and ASEAN · direct access to the engineering team.
Trusted by
Vietnam's telecom, power and banking leaders.
Two decades of Viethan Electric manufacturing for institutions that cannot afford downtime.
Technology
FPT
Vietnam's largest IT services group
Telecom
Viettel
National telecom operator
Power
EVN
State electricity utility
Banking
Vietcombank
Top-tier commercial bank
Banking
VietinBank
Top-tier commercial bank
20+ yearsof panel and enclosure manufacturing
3 critical sectorstelecom, power and banking
Same factorynow building Legend AI racks
Proven lines: switchgear enclosures, electrical distribution, circuit-breaker boxes, military cabinetry, EV-charging stations — and now data-centre racks and customised build-to-order.
Let's talk
Let's build the infrastructure behind your AI.
Start with a site assessment, a sample rack, or an OEM conversation. Engineered in Singapore. Manufactured in Vietnam. Deployed globally.
Singapore
Global AI Rack Commercial & Engineering Centre
Sales, application engineering, product management and regional spares.
- Rikado Lee
- +65 9181 0818 · RLee@vhlegend.com
- Richard Ang
- +65 8910 0965 · RAng@vhlegend.com
Vietnam
ASEAN Advanced Manufacturing Centre — Viethan Electric Co., Ltd
Road N4, Hoa Xa Industrial Zone, Nam Dinh Province, Vietnam
- Tel
- 0228 3670835
- Atuan@vhlegend.com
- Web
- www.vhlegend.com