Back to News
ProductAI Understanding briefing

Delta Electronics unveils AI modular data center with 800 VDC in‑row power

Delta Electronics announced a prefabricated AI Modular Data Center that combines 800 VDC in‑row power, liquid‑to‑air cooling and a microgrid solution featuring solid‑state transformers and solid‑oxide fuel cells, targeting high‑density AI workloads in Southeast Asia.

4 min readRead the linked source
Source-provided image accompanying Delta Electronics unveils AI modular data center with 800 VDC in‑row power
Source referenceSource recorded
Publisher
us.acrofan.com
Source link
us.acrofan.comhttps://us.acrofan.com/detail.php?number=1021887
Source type
Linked source — primary-source status has not been established.
ContextUnderstand this in 60 seconds

Start here

Key terms

Transformer
A neural architecture that uses attention to model relationships across sequences in parallel.
Compute
The processing resources required to train and run models, often measured in FLOPS or GPU hours.
Latency
The time between sending a request and receiving the model's output.
Test yourselfFuture of AI Quiz

What happened

Delta Electronics introduced its AI Modular Data Center (AI MDC) at Data Center World Asia 2026, featuring an 800 VDC in‑row power architecture delivering up to 660 kW of output power with 98 % AC‑DC efficiency. The system includes a 3.6 MW HVDC in‑row coolant distribution unit, liquid‑to‑air CDU technology, and a prefabricated containerized data‑center option that can pack 60 kW per rack within a single parking‑space footprint. The announcement also highlighted a Microgrid Solution that integrates renewable generation, energy storage, solid‑oxide fuel cells (SOFC) and solid‑state transformers (SST) to provide resilient, on‑site power for AI “factories.”

At the Data Center World Asia 2026 exhibition in Singapore, Delta Electronics showcased its AI Modular Data Center, a prefabricated solution designed for rapid deployment of high‑density AI workloads. The system’s core is an 800 VDC in‑row power architecture that delivers up to 660 kW per rack, with an advertised AC‑DC conversion efficiency of up to 98 %. The power module includes a 3.6 MW high‑voltage direct current (HVDC) in‑row coolant distribution unit (CDU) to manage thermal loads generated by dense GPU clusters.

The announcement also detailed a liquid‑to‑air CDU system that can achieve a Power Usage Effectiveness (PUE) as low as 1.19, indicating very low overhead for cooling. In addition, Delta introduced a containerized data‑center (CDC) variant capable of delivering 60 kW per rack within the footprint of a single parking space, emphasizing modularity and space efficiency.

Complementing the power and cooling hardware, Delta presented a Microgrid Solution that combines on‑site renewable generation, battery storage, solid‑oxide fuel cells (SOFC) and solid‑state transformers (SST). This architecture is intended to provide continuous, high‑quality power even during grid outages, reducing reliance on external utilities and improving overall resilience for AI workloads.

Source details: us.acrofan.com ↗

Why it matters

The launch addresses a pressing challenge for AI operators: the need for power‑dense, energy‑efficient infrastructure that can keep pace with rapidly scaling workloads. By moving power conversion closer to the rack with 800 VDC in‑row delivery, Delta claims to reduce conversion losses and improve overall PUE to as low as 1.19, which could translate into significant operational cost savings and lower carbon footprints. The integrated microgrid, featuring SSTs and SOFCs, offers a pathway to greater resilience against grid interruptions—a growing concern as AI data centers consume ever‑larger shares of regional electricity. If adopted widely, these solutions could reshape data‑center design standards in Southeast Asia, a region that is emerging as a hub for next‑generation AI infrastructure.

AI workloads are increasingly power‑hungry and sensitive to , making traditional AC‑based power distribution less efficient for next‑generation models. By delivering power at 800 VDC directly to the rack, Delta’s solution reduces the number of conversion stages, cutting energy loss and potentially lowering operating expenses for data‑center operators.

The integrated microgrid approach aligns with broader sustainability goals. Solid‑oxide fuel cells can run on a variety of fuels and provide steady baseload power, while solid‑state transformers reduce conversion losses compared with conventional technology. Together, these components could help data‑center operators meet regional carbon‑reduction targets.

Southeast Asia is positioning itself as a strategic location for AI due to favorable climate, lower labor costs, and growing demand for AI services. A turnkey, high‑efficiency infrastructure offering could accelerate the region’s capacity expansion, influencing global AI supply chains.

Interactive Mechanism

Interactive Mechanism: How It Actually Works

Explore the underlying technology behind this development interactively.

Model Parameter Size:8B Parameters
VRAM Required5.5 GBGPU memory footprint
Target HardwareMacBook / Single GPUDeployment tier
Privacy100% Air-GappedLocal device capability
Core takeaway: Small, quantized models (3B–8B) now run directly inside smartphones and laptops with complete data privacy, while mammoth 400B+ models remain the domain of datacenter clusters.
Interactive Concept Check+10 Points
Future of AI Quiz

What did neural scaling law research (e.g. Kaplan et al., 2020) observe?

What to watch next

Key factors to monitor include the pricing and availability of the AI MDC and its microgrid components, which the press release did not disclose. Adoption will depend on how quickly major cloud and AI service providers in the region can integrate the 800 VDC architecture with existing server hardware, especially NVIDIA’s Vera Rubin GPUs. Competitors such as Submer, Schneider Electric and other modular‑data‑center vendors may respond with comparable high‑voltage solutions, potentially driving a broader industry shift toward DC‑centric designs. Regulatory developments around renewable integration and grid interconnection in Southeast Asian markets could also influence deployment timelines.

Pricing and commercial availability remain undisclosed; market uptake will hinge on competitive pricing relative to existing AC‑based solutions.

Compatibility with NVIDIA’s Vera Rubin GPUs and other emerging AI accelerators will be critical; any integration challenges could slow adoption.

Regulatory approvals for on‑site generation and microgrid operation vary across Southeast Asian jurisdictions and may affect rollout speed.

Competitor responses, especially from firms already offering modular DC power solutions, could drive price competition and spur further innovation in the sector.

Related guides & quizzes

Future of AIAI TrainingAI Models ExplainedTest what you know — try a free AI quizLook up an AI term in our glossaryFollow the AI model release tracker
Found this useful?