SUPPORT
Frequently asked questions
Answers on GPU cloud vs bare metal, AI and machine learning workloads, pricing, compliance, and data centre locations — everything you need to evaluate Aolani Cloud for production infrastructure.
CATEGORIES
Introduction to Aolani
What is a neocloud?
A neocloud is a specialised cloud provider built specifically to power large-scale AI workloads.
Compared to traditional hyperscalers, which primarily offer generalised cloud and web services that could also include AI workloads, neoclouds focus solely on AI workloads.
Neoclouds deliver raw AI compute, giving customers capital-efficient access to large GPU clusters so teams can train and run AI and machine learning models at scale.
Neoclouds were more commonly known as “GPU cloud” or “GPU-as-a-Service (GPUaaS) providers” before the industry gained traction in 2023, which is also when Aolani was established.
Who is Aolani?
Aolani is a Singapore-founded neocloud building the infrastructure behind Asia’s AI future.
Established in 2023, Aolani provides hyperscalers and large-scale AI companies with high-performance GPU capacity and dedicated bare-metal infrastructure to train, deploy and scale AI systems.
Our bare-metal GPUaaS model delivers the performance, flexibility and cost efficiency organisations need to run production AI workloads without the capital and operational complexity of owning physical hardware.
Aolani operates on a robust governance framework built for the secure management and operation of critical AI compute resources, meeting the most stringent regulatory requirements of every market and industry it serves.
As an NVIDIA Cloud Partner, Aolani has priority access to the latest GPUs — including GB200 and GB300, with Vera Rubin to come — ensuring availability for customers even during global supply constraints.
Is Aolani a Singapore company?
Aolani was founded in Singapore, and its global headquarters office is located at 391B Orchard Road, Ngee Ann City Tower B, Level 22, Singapore 238874.
Aolani also maintains an operating presence in Malaysia.
Who is Aolani built for?
Aolani is built for companies scaling their AI capabilities. Aolani serves organisations that require high-performance GPU infrastructure and compliant network frameworks to build, train, and deploy complex AI models.
Aolani focuses on a targeted set of AI infrastructure capabilities, including dedicated bare-metal GPU servers and multi-node distributed AI training.
Aolani’s customer base extends beyond hyperscalers and large-scale AI companies to include sovereign AI programmes, regional enterprises, and public sector agencies that require dedicated access to high-performance GPU infrastructure.
About GPU Cloud and Bare Metal
What is a GPU cloud?
A GPU cloud is a cloud computing service that provides on-demand access to Graphics Processing Units (GPUs) over the internet, eliminating the need for companies to purchase and maintain physical hardware.
A GPU cloud uses specialised chips designed to process thousands of data tasks at the same time. This processing capability makes a GPU cloud the foundation infrastructure required to build, train, and run modern AI, deep learning, and large language models (LLMs).
What is the difference between a standard GPU cloud and a bare-metal GPU cloud?
A standard GPU cloud uses virtualisation software to divide physical server hardware into separate virtual machines. A bare-metal GPU cloud bypasses the virtualisation software to give users access to the entire, unshared physical server.
Should I use virtualised GPUs or bare metal?
Virtualised GPUs offer the best flexibility for testing and variable workloads. Companies that choose this prioritise agility over raw, uncompromised power. Typical applications include rapid prototyping, development and testing, fractional GPU needs, or unpredictable inference workloads with sudden traffic spikes.
Bare metal suits consistent, intensive workloads that are highly sensitive to performance variability. Bare-metal infrastructure excels at running large-scale distributed training across multiple GPUs and high-volume inference. With workloads running directly on dedicated hardware, this configuration grants firms greater control over their environment. Typical use cases include training LLMs from scratch, or supporting high-volume, sub-millisecond production inference.
In summary, a virtualised GPU cloud fits organisations that value agility and scalable access to GPU resources, whereas bare metal benefits companies looking to run private clusters or sustain workloads where dedicated infrastructure and predictable performance are critical.
Can I run a private AI cluster on Aolani?
Yes, companies can run a private AI cluster on Aolani.
Get in touch with us and let us help you identify the right infrastructure path for your needs.
What infrastructure do I need for large scale AI model training?
Training at scale requires fast networking to keep GPUs perfectly synchronised, storage that can handle frequent checkpointing without slowing, and dedicated hardware so performance stays consistent.
Aolani runs on bare metal with InfiniBand and RoCEv2 networking tuned specifically for distributed training, so companies get the full performance of the hardware.
Cloud for AI and machine learning
What cloud infrastructure works best for AI inference and deployment?
Inference runs on every user request, and it prioritises low latency and consistent uptime over raw training throughput.
Aolani delivers 99.9%+ uptime backed by 24/7 monitoring. The provider hosts this on the same dedicated bare-metal infrastructure it uses for training, ensuring performance remains predictable as demand scales.
Get in touch with our team to explore the most suitable infrastructure for your goals.
What cloud services are best for machine learning workloads?
Bare-metal GPU clouds, like Aolani, offer the best environment for machine learning workloads.
Machine learning requires consistent, dedicated GPU performance at every stage, and this is achieved by eliminating the shared virtualisation infrastructure that slows systems down.
Aolani provides dedicated bare-metal infrastructure across the entire machine learning lifecycle, so there is no need to switch providers as a workload moves from experimentation to production.
What GPUs does Aolani support?
Aolani supports current-generation, high-performance NVIDIA GPUs engineered for intense AI workloads.
NVIDIA GB200: The GB200 Grace Blackwell platform features enhanced memory capacity and bandwidth built for training and real-time inference on trillion-parameter large language models.
NVIDIA GB300: The Blackwell Ultra platform, delivering a major step up in memory capacity and inference throughput for frontier-scale reasoning and long-context workloads.
NVIDIA Vera Rubin: The next NVIDIA platform generation, coming to Aolani as it becomes available.
Aolani deploys GPU configurations within high-density, bare-metal environments equipped with the exact power, specialised liquid cooling, and low-latency inter-node networking required to run clusters.
What is the difference between multi-GPU and multi-node workloads?
Multi-GPU means using multiple GPUs inside a single server, while multi-node means connecting multiple servers together across a network.
While both are used to scale AI workloads, they solve different challenges. In multi-GPU, the AI model scales across several GPUs housed inside the same physical machine, making communications between the chips exceptionally fast and ideal for medium-sized models or single-node inference.
In multi-node, the AI model needs to be distributed across multiple physical servers as operations grow. With separate machines needing to communicate, a high-performance bare-metal GPU cloud uses low-latency fabrics like InfiniBand to sync the nodes in real time and prevent data bottlenecks.
Can Aolani support multi-GPU and multi-node workloads?
Yes, Aolani fully supports both multi-GPU and multi-node workloads.
The infrastructure is built specifically for distributed AI and demanding machine learning tasks. Aolani provides deployments that scale from individual GPU instances to distributed GPU clusters, making it suitable for workloads that require multiple GPUs within a single server or across multiple servers.
Pricing and onboarding
How does Aolani pricing work?
Aolani’s pricing is scoped around your needs — including workload requirements, GPU type, capacity, deployment model, reservation period, and support requirements, with cost efficiency at its centre.
Do you offer reserved or minimum committed capacity?
Aolani partners with our clients to develop a solution that best fits their needs. Speak to our team today to explore the best approach for your business.
Compliance and data sovereignty
How does Aolani handle hardware audits?
Aolani maintains a rigorous hardware audit programme for all GPU assets as part of our steadfast commitment to compliance and governance.
We take proactive steps to engage independent, reputable third-party audit firms to conduct physical, on-site inspections of our GPUs in the data centres. These audits are conducted at least once every six months, covering:
Quantity and inventory check
Serial number verification
Physical configuration verification
Maintenance records and operation records verification
Aolani also retains a panel of global and independent legal counsel to ensure our compliance remains rigorous, current, and up to date with any regulatory developments.
Is Aolani suitable for regulated industries?
Aolani is fundamentally built for compliance across the most highly regulated industries.
Aolani also retains a panel of global and independent legal counsel to ensure our compliance remains rigorous, current, and up to date with any regulatory developments.
What certifications do Aolani’s data centre partners hold?
Aolani partners exclusively with data centres that are ISO/IEC 27001:2022 compliant. Additionally, the company conducts sanctions screening against OFAC and Trade.gov Consolidated Screening List for all customers, partners and vendors, before entering any engagement.
How does Aolani ensure its partner data centres meet compliance and regulatory standards?
Aolani takes proactive steps to ensure the data centres we work with actively meet compliance and regulatory standards. The company aligns intentionally with active compliance programmes for US BIS/EAR export-control requirements, import regulations, and all applicable local digital, communications, and data-protection licensing obligations.
Aolani has retained a panel of external regulatory counsel that regularly reviews this programme. We have also engaged independent third-party audit firms to ensure that we remain compliant with evolving regulatory requirements.
Data centres and locations
Where are Aolani’s data centres located?
Aolani partners with data centres located in Johor and Kuala Lumpur, Malaysia.
Is Aolani planning to expand its data centre presence into Vietnam, Thailand and other markets?
Southeast Asia forms a core centre of the world’s AI infrastructure. The region offers strategic neutrality, competitive operating costs, and faster deployment timelines.
With its growing regional presence and its ability to operate across multiple jurisdictions, Aolani targets strategic growth directly within Southeast Asia’s expanding market.
