- Type
- AI cloud and data centre
- Operator
- YTL Power International Berhad
- Location
- Kulai, Johor, Malaysia
- Hardware
- Nvidia GB200 Grace Blackwell GPUs
- Status
- Operational (late 2025)
- Related
- Sovereign AI, ILMU
- Type
- AI cloud and data centre
- Operator
- YTL Power International Berhad
- Location
- Kulai, Johor, Malaysia
- Hardware
- Nvidia GB200 Grace Blackwell GPUs
- Status
- Operational (late 2025)
- Related
- Sovereign AI, ILMU
YTL AI Cloud is an artificial intelligence cloud service operated by YTL Power International Berhad, a subsidiary of the Malaysian conglomerate YTL Corporation. Built in partnership with Nvidia at the YTL Green Data Center Park in Kulai, Johor, it provides high-performance computing capacity for training and running large-scale AI models. The facility was commissioned in late 2025 and represents one of the most significant pieces of AI infrastructure in Southeast Asia, positioning Malaysia among the first countries in the region to offer domestic sovereign AI capability.
Facility and infrastructure
The YTL Green Data Center Park spans roughly 1,640 acres in Kulai, in the southern state of Johor, close to Singapore. The park is designed as a large campus with planned capacity in the region of 600 megawatts. A distinctive feature is its power strategy: the site draws on a large on-site solar plant complemented by backup supply from the national grid, reflecting an emphasis on lower-carbon operation for energy-intensive AI workloads.
The first AI data centre building on the campus is powered by liquid-cooled Nvidia GB200 Grace Blackwell systems, arranged in the NVL72 rack configuration that connects many GPUs into a single tightly coupled unit. Liquid cooling is used because the density and power draw of Grace Blackwell hardware exceed what conventional air cooling can dissipate efficiently. This hardware places the facility among the most capable AI compute sites in the region.
YTL AI Cloud services
On top of this hardware, YTL offers the YTL AI Cloud, a service through which enterprises, government agencies and global hyperscalers can access GPU capacity for AI training and inference. Reported early customers include global hyperscale operators, indicating that the facility serves both domestic and international demand. The cloud is intended to host a range of workloads, from foundation model training to production inference for digital applications.
A flagship workload is ILMU, a Malaysian-developed and Malaysian-owned large language model. Running ILMU on domestic infrastructure is central to the sovereign AI narrative, because it means a nationally significant model can be trained and served without depending on foreign data centres. This supports data residency requirements and gives Malaysia greater control over a strategic technology.
Strategic significance
The completion of the Kulai facility is part of a broader wave of data centre investment in Johor, which has become one of the fastest-growing data centre markets in Asia thanks to available land, power and proximity to Singapore. For Malaysia, the YTL AI Cloud advances several national objectives at once: attracting foreign technology investment, building local AI talent, enabling sovereign control over critical AI systems, and supporting the ambitions set out in national digital plans.
| Attribute | Detail | | --- | --- | | Campus size | About 1,640 acres | | Planned capacity | Around 600 MW | | Cooling | Liquid cooling for GPU racks | | Signature model | ILMU sovereign LLM | | Region role | Early Southeast Asian sovereign AI site |
The YTL AI Cloud is a cornerstone of Malaysia's sovereign AI ambitions. By training and hosting ILMU on domestic Nvidia Grace Blackwell hardware, Malaysia reduces its dependence on foreign cloud providers for strategically important AI systems. This aligns with the goals of the National AI Office and the MyDigital Blueprint, which envision Malaysia as a regional digital economy hub, and it supports compliance with the Personal Data Protection Act by keeping sensitive workloads within Malaysian jurisdiction.
The facility sits within the wider Johor data centre boom, part of the Johor-Singapore Special Economic Zone that has drawn major investment from global operators. Malaysia's combination of available land, competitive power costs and proximity to Singapore has made states such as Johor attractive to hyperscalers, while agencies including MDEC and MIDA actively court such investment. The YTL project shows a Malaysian company moving up the value chain from power generation into AI infrastructure.
There are national considerations to manage. Data centres are heavily energy and water intensive, prompting scrutiny from the Energy Commission and state authorities over grid capacity and sustainability, which is one reason the Kulai site pairs its operation with a large solar plant. For Malaysian talent, HRD Corp training programmes and university partnerships are important to ensure local engineers can operate and build on advanced AI infrastructure rather than relying solely on foreign expertise. Enterprises such as Maybank, CIMB, Petronas and Telekom Malaysia stand to benefit from domestic high-performance AI capacity for their own model deployments.