Domain models for creative, operations, and engineering teams
These domain-focused experiences demonstrate how shared inference capacity and APIs can support specialized workflows across visual production, infrastructure operations, and R&D.
Southeast Asia-Focused AI Inference Infrastructure
JUDONGLI ENERGY SDN BHD
Token Factory
From our Malaysia base, we plan to deploy and operate leading open-source and open-weight models, turning GPU capacity into reliable tokens, APIs, private inference, and enterprise AI services across Southeast Asia.
Token Factory Service Layer
From GPU capacity to production-ready AI services
A connected portfolio spanning model services, compute cloud, infrastructure solutions, and applied AI experiences for Southeast Asia.
Expert Model Services
Specialized model introductions for art, operations, and R&D workflows.
Compute Cloud Services
GPU and HPC capacity for multi-model inference, API delivery, and scalable customer workloads.
HPC & AI Solutions
End-to-end planning across compute infrastructure, AI deployment, and operations.
SaaS Applications
Intelligent application experiences led by itv for creative and media workflows.
Who We Are
Building Southeast Asia's AI inference infrastructure from Malaysia
Judongli Energy Sdn Bhd (formerly Judongli Electronic Sdn Bhd) is building a Southeast Asia-focused AI inference infrastructure and Model-as-a-Service platform from its home market and infrastructure base in Malaysia.
We are not simply renting GPUs, and we are not building around a single model. We plan to deploy, optimize, and operate high-demand open-source and open-weight models, converting high-density GPU capacity into reliable tokens, APIs, private inference, and production AI services.
In the Token Factory, GPU infrastructure is the production equipment, models are the operating software, and tokens delivered through APIs, private inference, and dedicated capacity are the output. The platform combines model selection, API aggregation, workload scheduling, reliability, and operational control.
Our Vision
From model training to large-scale AI consumption
The AI market is moving from model development toward large-scale application and inference. As capable open models mature, more customers can select the right model and consume it continuously without training from scratch. Delivering that experience at low cost and high reliability requires more than one model.
- Operate a model portfolio across model sizes, languages, latency targets, and industry workloads.
- Design for redundancy, peak demand, workload continuity, and predictable service quality.
- Support private inference, SLA requirements, data isolation, and dedicated customer capacity.
- Plan an initial deployment of approximately 1,000 B300/GB300-class GPUs with a roadmap beyond 2,000 GPUs.
Malaysia is the starting point. The regional roadmap extends into Indonesia and Thailand, followed by the broader ASEAN market as customer pipelines and utilization develop.
AI Service Portfolio
Expert model services for specialized workflows
Running above the Token Factory infrastructure layer, Judongli Energy's expert model portfolio introduces practical AI capabilities for creative, operations, and research development teams.
Art Model
Supports creative concept exploration, visual style reference, asset direction, and production planning for design and content teams.
Operations Model
Introduces AI-assisted monitoring, fault analysis, incident summarization, and operational knowledge support for infrastructure and service teams.
R&D Model
Helps software and engineering teams with technical research, requirement analysis, code reasoning, documentation review, and development knowledge workflows.
Token Factory Infrastructure
Turning GPU capacity into production-ready tokens and APIs
The Token Factory connects high-density compute, a curated open-weight model portfolio, inference operations, and API delivery so developers, enterprises, and sovereign customers can consume AI capacity with the reliability their workloads require.
GPU Capacity
Open-Weight Models
Inference
Tokens & APIs
Customer Workloads
Planned Inference Capacity
A scalable B300/GB300 deployment roadmap
The current planning case targets an initial deployment of approximately 1,000 B300/GB300-class GPUs, with a clear path to expand beyond 2,000 GPUs as market development, customer pipelines, and measured utilization support each phase.
- Multi-model orchestration across model sizes, languages, and latency requirements.
- Workload scheduling with redundancy, peak-capacity planning, and service continuity.
- Low-latency APIs, private inference environments, and dedicated capacity options.
Capacity figures describe a planned deployment roadmap and target-market opportunity. Customer pipeline, committed demand, deployment timing, and utilization are tracked as separate commercial measures.
The right model for each workload
Coordinates open-source and open-weight models across performance, language, cost, and latency profiles.
Capacity designed for continuity
Combines scheduling, redundancy, observability, and peak planning for reliable Token/API delivery.
Infrastructure aligned to customer control
Supports private inference, SLA, data-isolation, and dedicated-capacity requirements for regulated workloads.
Intelligent SaaS Applications
itv infinite canvas and image generation platform
The itv project is an applied workload built on Token/API capabilities: an intelligent creation platform for infinite-canvas planning, AI image generation, and short-drama production workflows that helps creative teams move from concept to production more efficiently.
Try itSupports open-ended visual planning, scene organization, node zooming, and canvas-based creative workflows.
Connects AI image generation with review, iteration, and asset preparation for media production teams.
Introduces story planning, shot organization, and production workflow support for short-form drama projects.
Our Milestones
Explore Judongli Energy's growth journey and key milestones in expanding our AI infrastructure and data center ecosystem across Malaysia:
2019
Company Incorporation
Judongli Electronic Sdn Bhd was officially incorporated on 27 September 2019, laying the foundation for our future development in technology-driven services.
2020
Early Technical Service Expansion
The company began providing equipment consultancy, technical support, and early-stage infrastructure coordination for regional partners, establishing our reputation for reliability and execution.
2021
Strategic Shift Toward Digital Infrastructure
Recognizing the rapid rise of cloud computing and AI workloads, we initiated our transition toward data center-related services, including facility planning, power optimization, and deployment support.
2022
Entry Into Compute Leasing & HPC Services
Judongli expanded into high-performance compute (HPC) leasing, supporting clients in AI model training, GPU resource allocation, and compute-intensive workloads.
2023
Strengthening Data Center Ecosystem Capabilities
We deepened our involvement in the data center ecosystem, covering rack deployment, energy management, and operational coordination with regional DC operators.
2024
Corporate Rebranding to Judongli Energy Sdn Bhd
To reflect our focus on energy-efficient infrastructure and compute-resource services, the company officially rebranded as Judongli Energy Sdn Bhd.
2025
Regional Partnerships & AI Infrastructure Projects
Judongli began collaborating with partners across Malaysia and Southeast Asia to support AI compute clusters, GPU hosting, and data-center expansion projects.
2026
Southeast Asia Inference Platform Roadmap
Judongli established a Token Factory roadmap from Malaysia, combining planned B300/GB300-class capacity, open-weight model operations, and a phased path toward broader ASEAN demand.
Why Judongli
Three markets, one regional inference platform
Developers and AI companies. Aggregated model APIs backed by inference capacity we plan and operate.
Enterprise private AI. Private inference, SLA, data isolation, and dedicated-capacity options for regulated industries.
Government and sovereign AI. Infrastructure control, model-location choice, and data-residency alignment for public-sector workloads.
Malaysia infrastructure base. A home market aligned with national and industry adoption of AI infrastructure.
ASEAN expansion path. Market development into Indonesia and Thailand before broader regional coverage.
Capacity matched to demand. Phased deployment tied to target markets, customer pipelines, and measured utilization.
Infrastructure Partnership
A long-term capacity partner, not a one-time GPU supplier
We are engaging with infrastructure owners that can support a planned initial B300/GB300-class deployment and expand with measured demand over the years ahead. The goal is delivery certainty, operational transparency, and a capacity roadmap that can grow with the Token Factory.
First delivery and current capacity
Clear visibility into available capacity, deployment timing, location, and operating readiness.
Capacity that can expand with utilization
A defined near-term expansion path as customer pipelines convert into measured Token consumption.
A partner for the next stage of scale
Long-term planning that can support a path from approximately 1,000 GPUs toward 2,000 and beyond.
We are looking for the ability to plan, deliver, operate, and scale together.
201901034928 (1344258-M)
Contact us for sales
Talk to us about model APIs, private inference, dedicated capacity, or infrastructure partnerships.