Powerful GPU instances for AI/ML and HPC workloads

Support AI/ML and HPC with
6 GPU types

Designed for security
0 gap

Scalability powered by
6 AZs

Start from
$1.55

GPU instances
designed for demanding AI workloads

Powerful capabilities designed to accelerate AI/ML
and high-performance workloads at any scale.

Flexible compute foundation

Whatever your workload, GreenNode gives you the right compute, the right price, and the flexibility your team needs.

Best for cost-efficient AI inference, fine-tuning, and pro visualization (Omniverse/VFX). A high-performance GPU with 96GB memory and FP4 precision, delivering strong throughput at about 32% lower cost than H100.

Best for: Large-scale LLM inference, long-context processing, and memory-intensive AI workloads where the H100's 80GB VRAM is the bottleneck. The go-to upgrade for serving 70B+ parameter models at full precision.

Architecture: Hopper VRAM: 141GB HBM3e Transformer Engine + FP8 precision NVLink 4.0

Best for: Large-scale AI training, LLM pre-training, and HPC workloads demanding extreme compute. The go-to choice when training foundation models from scratch at scale.

Architecture: Hopper VRAM: 80GB HBM3 Transformer Engine + FP8 precision NVLink 4.0

Best for: GenAI training and inference, model fine-tuning, visual AI, 3D workloads, and advanced data analytics. Sits between the A40 and H100 in price-to-performance for real-world AI workloads.

Architecture: Ada Lovelace VRAM: 48GB GDDR6 ECC Designed for 24/7 continuous operation

Best for: Engineering simulations, AI inference, visual computing, VDI, and data science workloads requiring rock-solid reliability. Ideal for enterprise environments where uptime is non-negotiable.

Architecture: Ampere VRAM: 48GB GDDR6 ECC Designed for 24/7 continuous operation

Best for: AI agents, LLM inference, RAG pipelines, vision-language models, embedding models, and computer vision. The most powerful consumer GPU available today for developers running mid-to-large scale models.

Architecture: Blackwell VRAM: 32GB GDDR7 1.3–2× faster than RTX 4090

Best for: Medium-scale AI inference, small ML application development, and fine-tuning smaller models. The sweet spot between AI performance and cost for individual developers or small teams.

Architecture: Ada Lovelace VRAM: 24GB GDDR6X 16,384 CUDA cores

Interconnect built on NVIDIA
architecture principles

GreenNode GPU instances follow NVIDIA reference architecture,
combining NVLink/PCIe, InfiniBand (RDMA), and RAIL topology
for maximum AI/HPC performance and scalability.

NVLink / PCIe Switch

Ultra-fast GPU-to-GPU communication across and within nodes, eliminating training bottlenecks.

InfiniBand (RDMA)

Lossless, low-latency networking for efficient distributed training and minimal server overhead.

Rail Topology

Parallel network paths for higher bandwidth, redundancy, and stable performance at scale.

Auto-scale GPUs and maximize utilization

From a single GPU to thousands, GreenNode provisions capacity
before you need it and extracts full performance from every instance.

Kubernetes-native GPU auto-scaling

Dynamically scale GPU resources from small deployments to thousands of GPUs — with predictive scaling that provisions capacity before you need it.

Secure every connection
without compromise

SSH, TCP, and HTTP access with built-in protections — keeping sensitive business data safe and every connection fully controlled.

Enhanced GPU utilization
with MIG

Securely partition a single GPU into multiple isolated instances — running parallel AI workloads with predictable performance, higher utilization, and lower infrastructure costs.

GPU instances
Flexible pricing, no idle cost

From large-scale training to real-time inference, GPU Cloud is built to
power the full spectrum of AI and high-performance workloads.

NVIDIA H200

Starting at

$3,076 /30 days

Ideal for ultra-large LLM training, long-context inference & advanced HPC workloads.

NVIDIA H100

Starting at

$2,803 /30 days

Ideal for large-scale AI training, LLMs, and HPC workloads

NVIDIA A40

Starting at

$860 /30 days

Ideal for AI inference, visual computing, and data science workloads

NVIDIA L40S

Starting at

$1,248 /30 days

Ideal for fine-tuning & mid-scale training, and advanced data analytics

RTX 4090

Starting at

$610 /30 days

Idea for medium-scale inference and small ML application development.

RTX 5090

Starting at

$769 /30 days

Ideal for next-gen AI development & advanced exprimentation.

Ready to scale your AI workload?

From training to inference, pick the right GPU for your needs. Monthly billing, no hidden fees, no idle cost.

Get started

Do more with
GreenNode

Go beyond compute — manage, scale, and grow everything in one simple cloud ecosystem.

AgentBase

Securely deploys and operates capable AI agent at scale, providing enterprise-grade infrastructure for production, security, and observability.

Explore AgentBase

AI platform

Unified platform to train, fine-tune, and deploy AI models at scale.

Explore AI Platform

Vector Database - vDB OpenSearch

Enables fast search, real-time analytics, and large-scale log, event data exploration, and vector database for RAG

Explore vDB OpenSearch

Kubernetes

Fully managed Kubernetes for container orchestration

Explore VKS

Availability zones

20+

Regional CDN nodes

1000+

Customers

24/7

Customer support

Powering 1000+ businesses
on their digital transformation journey

From fast-growing startups to leading enterprises, companies
choose GreenNode for secure, high-performance AI cloud
solutions that help them innovate and scale

Process millions of complex documents with 90% faster turnaround

Optimize operations and meet compliance with local cloud infrastructure

FE CREDIT leverages GreenNode to run mission-critical systems and applications, ensure regulatory compliance, and streamline IT operations across its nationwide network. With the high-performance and scalable cloud infrastructure, FE CREDIT has strengthened system stability, improved customer experience, and reduced the operational burden on its IT team.

14M+

Customers with seamless experiences

21,000

Reliable service points across Vietnam

Process millions of complex documents with 90% faster turnaround

Using GreenNode’s IDP OCR solution, a leading Vietnamese bank has fully automated the extraction and classification of thousands of complex banking forms. The system intelligently processes long, unstructured documents and eliminates manual review, reducing document-handling time by more than 90% while ensuring high accuracy and compliance.

150M+

Documents processed automatically

90%

Reduction in document processing time

Run nearly 300 retail stores efficiently on GreenNode infrastructure

ACFC & VFBS operate close to 300 fashion and F&B stores nationwide on GreenNode's stable, scalable cloud platform. By migrating core systems—including POS, ERP, HRM, and operational reporting—to the cloud, the team has reduced IT workload, strengthened security, and ensured uninterrupted operations across all brands and locations.

350+

Stable POS systems

25

Brands & F&B chains supported by unified IT infrastructure

PangoCDP migrates to GreenNode domestic cloud for speed, savings, and compliance

PangoCDP, Vietnam's leading Customer Data Platform, moved from international infrastructure to GreenNode, eliminating high latency, cutting data transfer costs, and achieving full Vietnamese data compliance.

44x

Faster deployment

30%

Reduction in operating costs

Sotabox.ai goes live and run smoothly in just 1 month on GreenNode

Within 30 days, SotaTek successfully deployed the entire Sotabox.ai platform on GreenNode, reducing DevOps costs by 30% and accelerating time-to-market for their global AI product. With flexible infrastructure and PaaS services such as vDB, Kafka, and VKS, the engineering team can build, scale, and optimize faster while ensuring performance and stability for end users.

1 month

To deploy and operate

30%

Reduction in DevOps costs

Frequently asked questions

What NVIDIA GPU types does GreenNode offer?

GreenNode offers 5 NVIDIA GPU types to support a wide range of AI and HPC workloads: NVIDIA H100 (ideal for large-scale AI training and LLMs), NVIDIA A40 (visual computing and data science), NVIDIA L40S (fine-tuning and mid-scale training), RTX 4090 (medium-scale inference), and RTX 5090 (next-gen AI development). All instances are available on-demand with no long-term commitment.

Which AI and machine learning workloads are GreenNode GPU instances optimized for?

GreenNode GPU instances are built to support the full spectrum of AI/ML and HPC workloads — including large-scale model training, LLM development, real-time inference, fine-tuning, visual computing, data science, and advanced analytics. With 5 GPU types available, you can match the right hardware to your specific workload requirements and budget.

How does MIG (Multi-Instance GPU) partitioning work on GreenNode?

MIG (Multi-Instance GPU) is a feature that allows a single physical GPU to be securely partitioned into multiple isolated instances. On GreenNode, MIG enables you to run parallel AI workloads on the same GPU while maintaining predictable performance and full isolation between instances. This significantly increases GPU utilization and reduces infrastructure costs — especially for inference and fine-tuning workloads that don't require a full GPU.

Where are GreenNode GPU instances available — which regions and availability zones?

GreenNode GPU instances are available across 6 availability zones (AZs) in Southeast Asia, spanning Ho Chi Minh City (HCM-1A, HCM-1B, HCM-1C), Hanoi (HAN-1A, HAN-1B), and Bangkok (BKK-1A). This regional coverage allows businesses in Vietnam and Thailand to deploy GPU workloads closer to their end users — reducing latency and meeting local data compliance requirements.

How quickly can I deploy a GPU instance on GreenNode?

You can go from signup to a running GPU instance in under 5 minutes. GreenNode's self-service portal lets you select your GPU type, configure your environment, and launch instantly — no manual provisioning or lengthy onboarding required. For enterprise deployments or custom configurations, our team is available to assist with dedicated setup and support.

Powerful GPU instances for AI/ML and HPC workloads

GPU instances designed for demanding AI workloads

GPU instances Flexible pricing, no idle cost

Ready to scale your AI workload?

14M+

21,000

150M+

90%

350+

25

44x

30%

1 month

30%

Frequently asked questions

Specific use cases in mind?We're here to support your business needs.

Powerful GPU instances
for AI/ML and HPC workloads

GPU instances
designed for demanding AI workloads

GPU instances
Flexible pricing, no idle cost

Specific use cases in mind?
We're here to support your business needs.