SuperX launches modular AI data center solution with 6-month deployment

Super X AI Technology Limited (NASDAQ: SUPX) announced the launch of its SuperX Modular AI Factory, a prefabricated data center solution designed to reduce deployment time to under six months, according to a company statement.

The modular system integrates compute, cooling, and power components that are pre-fabricated and tested before on-site assembly. Each 20MW module can support up to 144 NVIDIA GB200 NVL72 systems across six SuperX NeuroBlock core compute units, requiring approximately 6,000 square meters of physical space.

The solution uses High-Voltage Direct Current technology to achieve power efficiency above 98.5% and targets a Power Usage Effectiveness ratio of 1.15, compared to traditional air-cooled systems that typically achieve 1.5 PUE, based on company specifications.

Key components include the SuperX NeuroBlock compute unit supporting up to 24 NVIDIA GB200 systems with 3.5MW power capacity, the SuperX CryoPod cooling system with dry cooler or chiller options, and energy storage systems for renewable power integration.

The modular architecture allows customers to start with a single module and scale incrementally. The company states this approach reduces initial capital expenditure compared to traditional data center construction, which typically requires 18 to 24 months for completion.

SuperX, headquartered in Singapore, provides AI infrastructure solutions including high-performance servers, HVDC solutions, and liquid cooling systems for enterprises, research institutions, and cloud computing deployments globally.

You May Also Be Interested In

• Microsoft board member Reid Hoffman to step down at 2026 meeting
• SuperX to debut 1.6T optical modules at Interop Tokyo 2026
• USD.AI provides $98.1M debt financing for Duos edge AI GPU deployment