YTL AI Cloud Among Asia’s First Cloud Providers to Achieve NVIDIA Exemplar Cloud Status

Rigorous architectural validation ensures deterministic performance, reproducibility, and maximum utilization across massive-scale AI training, inference, and Sovereign AI workloads.

YTL AI Cloud has officially achieved NVIDIA Exemplar Cloud status, establishing YTL AI Cloud as one of the first cloud infrastructure providers in Asia to earn this validation, placing it within an elite tier of global hyperscalers and AI-specialised clouds.

YTL AI Cloud joins a select group of global NVIDIA Exemplar Cloud providers and is among the first NVIDIA Cloud Partners in the Asia-Pacific region to complete validation.

The achievement reinforces YTL AI Cloud’s position as a leading provider of AI infrastructure and expands access to advanced AI computing capabilities for businesses, researchers, developers and public sector organisations across Malaysia and the wider region.

The NVIDIA Exemplar Cloud initiative is a highly rigorous operational and architectural blueprint designed to reduce the variance and performance degradation often found in public cloud environments. It validates that a cloud provider’s infrastructure meets or exceeds NVIDIA’s stringent performance, efficiency, and scalability targets.

This validation ensures enterprise customers achieve deterministic, predictable performance against NVIDIA benchmarking recipes, baselines, and reference workload expectations, maximizing return on infrastructure spend.

To secure Exemplar Cloud achievement, YTL AI Cloud’s infrastructure was subjected to NVIDIA Exemplar’s rigorous training benchmarking. The benchmark evaluated a 512-GPU environment across leading AI model workloads, including DeepSeek-V3, Llama 3.1, Qwen3, Grok-1, Nemotron 3, Nemotron-H, GPT(OSS), and related large-model scenarios. The validation focused on core infrastructure performance metrics, including mean step time, TFLOPS per GPU, and run-to-run consistency across repeated executions.

“Being one of the first NVIDIA Exemplar Clouds in Asia shows that we have the talent and infrastructure to lead in building a sovereign, world-class AI infrastructure that performs at the same level as the leading hyperscalers globally,” said Dato’ Seri Yeoh Seok Hong, Managing Director of YTL Power International.

“To push the boundaries of AI, our customers demand cloud infrastructure and operations that perform consistently under the highest standards,” said Philip Lin, Chief Executive Officer at YTL AI Cloud.

“Achieving NVIDIA Exemplar Cloud status validates our organization’s proven ability to handle the most intensive, real world AI workloads, reducing operational and infrastructure risk so that our customers can focus on their core business objectives.”

“YTL AI Cloud is bringing NVIDIA accelerated computing infrastructure to customers across Asia,” said Marc Hamilton, Vice President, Solutions Architecture and Engineering, NVIDIA.

“Built to NVIDIA NCP Reference Architecture standards, the platform achieves NVIDIA Exemplar Cloud status, and will help advance sovereign AI adoption, frontier model development, scientific research and innovation across the region.”

Architected for Deterministic Performance and Bare-Metal Efficiency

For senior cloud executives and enterprise architects, the primary challenge of running massive AI models is infrastructure jitter and sub-optimal hardware utilization caused by virtualization layers and network bottlenecks. YTL AI Cloud addresses this challenge with a unified AI cloud architecture that tightly integrates GPU compute, high-speed networking, storage, and orchestration, enabling large-scale GPU clusters to perform as a coherent, high-efficiency AI compute fabric.

This is achieved via NVIDIA NVLink™ technology, which provides high-bandwidth, low-latency chip-to-chip interconnectivity, coupled with NVIDIA’s Spectrum-X Ethernet Networking Platform to ensure linerate fabric throughput and anchored by NVIDIA Blackwell GPUs.

Managed and optimized by YTL AI Cloud’s proprietary Cloud Orchestration Platform (COP), the infrastructure abstracts operational complexity while enforcing strict hardware-level execution.

The proprietary COP dynamically manages compute, network topology, and high-performance storage to maintain high utilization rates and guarantee complete workload reproducibility across largescale clusters.

Proven Infrastructure Consistency Across Large-Model Workloads

The results demonstrate YTL AI Cloud’s ability to deliver stable, scalable, and performance-dense AI infrastructure for frontier-scale model workloads. Across the representative workloads across repeated runs, the platform met or exceeded NVIDIA’s performance targets across all test phases indicating strong platform consistency and operational maturity.

Demonstrating alignment with NVIDIA’s highest benchmarking profile, YTL AI Cloud recorded exceptional internal consistency and hardware efficiency across a variety of massive-scale training and operational profiles.

The results confirm that YTL AI Cloud’s software and hardware layers are tuned perfectly to prevent thermal throttling or interconnection delays, allowing code to run exactly as intended by model developers.