STOCK TITAN

3 E Network Unveils Mikkeli AI Data Center Blueprint Designed for NVIDIA Vera Rubin Architecture

3 E Network details a Finland AI data center blueprint built for NVIDIA Vera Rubin-class power, cooling and 1.6T network demands.

(Moderate)
(Positive)
Tags
AI

3 E Network Technology Group (MASK) announced the core engineering blueprint for its planned AI data center in Mikkeli, Finland, designed to support NVIDIA’s next-generation Vera Rubin DSX architecture standards. The facility’s civil, mechanical, electrical and environmental systems are engineered for high thermal density, liquid cooling and 1.6T lossless network topologies to accommodate future Vera CPU, Rubin GPU Superchip and HBM4-based systems.

The design features direct-to-chip liquid cooling with Coolant Distribution Units, extensive routing capacity for 6th-generation NVLink and 1.6T scale-out networks, structural reinforcement for near two-ton liquid-cooled racks, and a hybrid power architecture that supports both current HGX/MGX nodes and future 48V DC power shelves with rack-level Battery Backup Units. The blueprint separates long-lived MEP infrastructure from compute hardware and underpins a dual-track commercialization plan using HGX clusters for large-model training and MGX clusters for inference and agile, mixed workloads.

Loading...
Loading translation...

Positive

  • None.

Negative

  • None.
Argus 15 min delay
+2.38% vs previous close $1.29 last price 1725.2x rel. volume Open Argus
Details

Market Reaction – MASK

+14.8% Peak in 13 min
$1.23 $1.55 Day Range
$4.64M Market Cap

Following this news, MASK has gained 2.38%, reflecting a moderate positive market reaction. Argus tracked a peak move of +14.8% during the session. Our momentum scanner has triggered 26 alerts so far, indicating elevated trading interest and price volatility. The stock is currently trading at $1.29. Trading volume is exceptionally heavy at 1725.2x the average, suggesting very strong buying interest.

Data tracked by StockTitan Argus (15 min delayed). Upgrade to Gold for real-time data.

Market Context

On Sep 8, MASK’s prior AI storage strategy drew -0.76% over 24 hours, providing a related market dat...
Analysis

On Sep 8, MASK’s prior AI storage strategy drew -0.76% over 24 hours, providing a related market datapoint as this blueprint further specified the Finland data-center infrastructure.

Key Figures

Network scale: 1.6T Facility lifespan: 10-15 years Chip iteration cycle: 1-2 years +3 more
Network scale
1.6T
Planned scale-out and high-density interconnect infrastructure
Facility lifespan
10-15 years
Typical facility lifespan cited in the blueprint
Chip iteration cycle
1-2 years
Typical AI chip iteration cycle cited in the blueprint
Single-rack thermal load
Tens to over a hundred kilowatts
Expected with Vera CPU, Rubin GPU Superchip and HBM4
Computing rack weight
Near two tons
Anticipated weight of liquid-cooled computing racks
Pre-training scale
Trillion-parameter
AI enterprise and foundational model training tasks

Previous AI Reports

1 past event · Latest: Sep 08
Same Type 1 event
  1. Sep 08

    AI storage strategy

    24h Move
    -0.8%

    Outlined AI storage strategy and identified Finland data center as deployment and feedback hub.

24h Move is the share-price change in the day after each event; other market factors may also have contributed.

Key Terms

power usage effectiveness, infiniband, retrieval-augmented generation
3 terms
power usage effectiveness technical
"continuously optimize Power Usage Effectiveness under high loads"
Power usage effectiveness (PUE) is a simple ratio that compares the total energy a facility uses (cooling, lighting, power losses) to the energy used directly by computing equipment; a lower number means more of the power is doing productive work. For investors, PUE acts like a fuel-efficiency rating for data centers — better PUE usually means lower operating costs, smaller environmental footprint, and a competitive advantage when scaling operations or meeting sustainability expectations.
infiniband technical
"Spectrum-X Ethernet and advanced InfiniBand architectures"
Infiniband is a high-speed data transport technology used inside data centers to move large amounts of information quickly and with very little delay, often used for servers, storage and computing clusters. Investors should care because it acts like a multilane expressway for data—companies that build, use or support such fast networks can gain competitive advantages in cloud services, high-performance computing and AI workloads, which can affect costs and revenue potential.
retrieval-augmented generation technical
"compute power in Retrieval-Augmented Generation"
An AI method that combines a conversational language system with live access to external documents or databases, so the AI first fetches relevant facts and then uses them to form its answer. Think of it as an assistant that checks a file cabinet for source papers before replying, which helps reduce mistakes and reveal evidence. For investors it matters because it can produce more accurate, verifiable summaries of filings, news and research, speeding due diligence while still depending on the quality of the underlying data.

AI-generated analysis. How Rhea-AI works. Not financial advice.

See more from StockTitan in Google Search and AI answers. Adds StockTitan as a preferred source · opens Google
Add on Google
  • Future-Ready Infrastructure: Natively designed for next-gen thermal density and 1.6T lossless networks, supporting long-term capital expenditures.
  • Dual-Track Monetization: Featuring initial deployment of advanced HGX and MGX mixed clusters for AI training and inference.

HONG KONG, Sept. 11, 2026 (GLOBE NEWSWIRE) -- 3 E Network Technology Group Limited (Nasdaq: MASK) (the “Company” or “3 E Network”), a business-to-business (“B2B”) information technology (“IT”) business solutions provider, committed to becoming a next-generation artificial intelligence (“AI”) infrastructure solutions provider, today unveiled the core engineering blueprint for its key AI data center in Mikkeli, Finland. The project’s physical infrastructure (civil, mechanical, electrical, and environmental control) is natively designed to provide support for NVIDIA’s next-generation Vera Rubin DSX system-level architecture standards. To support initial commercialization, the Company plans to deploy a mix of HGX- and MGX-based computing clusters during the first phase.

To address increasing power-density requirements as AI models continue to scale, 3 E Network has established an engineering strategy of “forward-looking infrastructure readiness combined with pragmatic computing cluster deployment.” This blueprint addresses a critical industry pain point: the misalignment between facility lifespans (typically 10-15 years) and AI chip iteration cycles (1-2 years). By separating facility infrastructure from compute hardware, the facility aims to help manage long-term total cost of ownership and preserve the value of hardware investments by reducing the cost and disruption of future upgrades.

Reconstructing the Foundation: “Vera Rubin-Ready” to Address High Physical and Thermal Demands

As AI chips evolve toward higher performance, data centers are facing significant challenges in power, cooling, and structural load-bearing. To ensure the Mikkeli facility remains capable through future generational transitions, the project sets rigorous metrics during the civil and Mechanical, Electrical, and Plumbing (“MEP”) stages, aligned with the Vera Rubin architecture:

  • Full-Stack Liquid Cooling and Resilient Piping Networks: Anticipating the application of the Vera CPU and Rubin GPU Superchip, alongside the new HBM4 high-bandwidth memory, single-rack thermal loads are expected to reach levels of tens to over a hundred kilowatts. Consequently, the facility is designed with a direct-to-chip liquid cooling and Coolant Distribution Unit circulation system compatible with blind-mate technology. Leveraging Finland’s natural cooling resources, this architecture is designed to provide adequate cooling redundancy and continuously optimize Power Usage Effectiveness under high loads.
  • 1.6T Scale-out and High-Density Interconnect Infrastructure: Addressing the massive physical volume of intra-rack high-density cabling required by the 6th-generation NVLink, the facility reserves ample rack-level routing space; concurrently, to support the capabilities of the ConnectX-9 architecture, the data center’s backbone fiber cable trays and routing pathways are fully aligned with 1.6T scale-out non-blocking network topologies. The facility reserves wiring space for next-generation Spectrum-X Ethernet and advanced InfiniBand architectures, providing the physical infrastructure required for rack-scale interconnects.
  • High-Capacity Structural Reinforcement & Hybrid Power Architecture: Anticipating the near two-ton weight of liquid-cooled computing racks, the facility implements reinforced flooring designed for high load capacities during the civil engineering phase. In terms of power routing, the facility adopts a flexible dual-track design: it supports the deployment of high-voltage smart PDUs to accommodate standard MGX and HGX nodes, while concurrently providing full forward-compatibility for 48V centralized DC power shelves and rack-level Battery Backup Unit specifications. This design aims to handle the high power density and transient power spikes anticipated with the Vera Rubin architecture, enhancing grid-level stability.
  • Deep Decoupling of MEP Facilities and IT Compute Modules: The design incorporates out-of-band environmental monitoring and micro-leak detection, together with standardized interfaces between facility infrastructure and IT equipment. This approach is intended to reduce the need for modifications to the facility during future compute hardware upgrades and minimize operational disruption.

Dual-Track Commercialization: HGX-Based Training and MGX-Based Inference

While building a high-standard physical foundation, 3 E Network is focused on steadily and efficiently commercializing its computing capacity and developing recurring revenue streams. The planned initial introduction of the HGX and MGX mixed computing architecture is intended to create an elastic compute pool. This dual-track approach corresponds to the two core commercial demands of the AI market, aiming to optimize hardware asset utilization and Return on Investment:

  1. HGX Core Zone (Addressing Large Model Pre-training): This zone is designed to accommodate a high-density deployment of high-performance HGX baseboards. Backed by centralized power and liquid cooling support, this zone is dedicated to providing excellent single-node interconnect bandwidth and rack-scale cluster scalability. The HGX cluster is designed to handle complex, trillion-parameter pre-training tasks of AI enterprises and foundational model developers. Its lossless interconnect significantly mitigates the “data wall” bottleneck, providing efficient compute throughput for training clusters.
  2. MGX Elastic Zone (Addressing Commercial Inference, Vertical Fine-tuning, and Agile Deployment): This zone is designed to accommodate modular clusters based on MGX specifications. By supporting flexible, heterogeneous combinations of CPUs, GPUs, and Data Processing Units, this zone is tailored for a broad spectrum of enterprise needs. As AI enters the application deployment phase, demand for concurrent compute power in Retrieval-Augmented Generation, industry-specific model fine-tuning, and edge computing has grown rapidly. The MGX cluster supports the efficient launch of these tasks with a more agile cost structure, accelerating AI adoption across industry-specific applications while helping 3 E Network capture a wider B2B customer base.

Strategic Perspective and Executive Remarks

Dr. Tingjun Yang, Chief Technology Officer of 3 E Network, stated: “Building modern AI computing infrastructure is a dual test of capital planning and engineering foresight. For the Mikkeli project, we are designing the facility’s physical infrastructure to accommodate the anticipated power, cooling and interconnect requirements of the Vera Rubin architecture and support future hardware upgrades. Our planned initial deployment of HGX- and MGX-based clusters is intended to address demand for model training and inference. Furthermore, relying on Finland’s unique advantages in green energy and natural cooling, we effectively integrate high energy efficiency with computing power. 3 E Network is committed to applying rigorous engineering standards to develop AI data center solutions that balance current commercialization capabilities with future technological scalability for our global partners.”

About 3 E Network Technology Group Limited
3 E Network Technology Group Limited is a business-to-business (“B2B”) information technology (“IT”) business solutions provider committed to becoming a next-generation artificial intelligence (“AI”) infrastructure solutions provider. It upholds the industry consensus of “AI and energy symbiosis” and has a strong vision in the field of energy investment. The Company’s business comprises two main portfolios: the data center operation services portfolio and the software development portfolio. For more information, please visit the Company’s website at https://3emask.com/.

Forward-Looking Statements
Certain statements in this announcement are forward-looking statements. These forward-looking statements involve known and unknown risks and uncertainties and are based on the Company’s current expectations and projections about future events that the Company believes may affect its financial condition, results of operations, business strategy, and financial needs. Investors can identify these forward-looking statements by words or phrases such as “approximates,” “assesses,” “believes,” “hopes,” “expects,” “anticipates,” “estimates,” “projects,” “intends,” “plans,” “will,” “would,” “should,” “could,” “may” or similar expressions. The Company undertakes no obligation to update or revise publicly any forward-looking statements to reflect subsequent events or circumstances, or changes in its expectations, except as may be required by law. Although the Company believes that the expectations expressed in these forward-looking statements are reasonable, it cannot assure you that such expectations will turn out to be correct, and the Company cautions investors that actual results may differ materially from the anticipated results and encourages investors to review other factors that may affect the Company’s future results in the Company’s registration statement and other filings with the U.S. Securities and Exchange Commission.

For more information, please contact:

3 E Network Technology Group Limited
Investor Relations Department
Email: ird@3emask.com
Website: https://3emask.com/


FAQ

What engineering strategies does 3 E Network use to handle future high thermal and power densities?

The Mikkeli blueprint includes direct-to-chip liquid cooling with Coolant Distribution Units and blind-mate compatibility, out-of-band environmental monitoring and micro-leak detection, and reinforced flooring designed for near two-ton liquid-cooled racks. It also adopts a dual-track power design that supports high-voltage smart PDUs for current HGX/MGX nodes and is forward-compatible with 48V centralized DC power shelves and rack-level Battery Backup Units, which is intended to manage high power density and transient power spikes associated with future Vera Rubin-class systems.

How is the network and cabling infrastructure prepared for NVIDIA Vera Rubin architecture and related interconnects?

The facility reserves ample intra-rack routing space for the cabling needs of 6th-generation NVLink and aligns backbone fiber trays and routing pathways with 1.6T scale-out non-blocking network topologies. It also reserves wiring space for next-generation Spectrum-X Ethernet and advanced InfiniBand architectures to support rack-scale interconnects, aiming to provide the physical foundation for lossless, high-bandwidth AI training and inference clusters.

What is the purpose of the dual-track HGX and MGX commercialization approach?

The company plans an initial mixed deployment of HGX and MGX clusters to create an elastic compute pool. The HGX core zone is designed for high-density, liquid-cooled clusters targeting complex, trillion-parameter pre-training for AI enterprises and foundational model developers. The MGX elastic zone supports modular, heterogeneous combinations of CPUs, GPUs and Data Processing Units for commercial inference, vertical fine-tuning, Retrieval-Augmented Generation, and edge or industry-specific workloads, with a more agile cost structure and broader B2B application coverage.

Keep reading