What is a cloud GPU platform?

A cloud GPU platform is a service offered by cloud GPU providers that allows users to access and utilize GPU technology remotely. Instead of installing physical GPUs on local machines, users can leverage the power of cloud GPUs hosted on efficient cloud GPU platforms. These platforms, such as Google Cloud GPUs and NVIDIA GPU instances, harness the high-performance capabilities of GPUs, including the NVIDIA Tesla series, making them accessible to users through the cloud.

Why do you need cloud GPU services?

Cloud GPU services are essential for individuals and businesses that require immense computational power without the capital expense of buying and maintaining physical GPUs. As demand for high-performance computing increases across areas such as artificial intelligence, deep learning, and graphics rendering, an efficient cloud GPU platform can offer scalable, cost-effective solutions. Moreover, with the emergence of the best cloud GPU platforms, users can now rent GPU power on demand, making it suitable for short-term, intensive tasks or projects. This way, users can leverage the cutting-edge capabilities of services like Google Cloud GPUs or NVIDIA GPU instances without committing to a significant hardware investment.

How secure are cloud GPU services?

Security is a top priority for any cloud GPU provider. The best cloud GPU platforms implement stringent security measures to ensure users' data and applications remain protected. This includes data encryption during transit and at rest, secure access controls, regular security audits, and more. Providers of services like NVIDIA GPU instances and Google Cloud GPUs invest heavily in maintaining the integrity and confidentiality of user data. As with any cloud service, while the provider takes measures to secure the infrastructure, users should also follow best practices in data management and access control to ensure optimal security.

Cloud providers allocate specific quotas for GPU instances, which can vary by type and region. To request a quota increase, developers must specify the instance type (e.g., P3XL) and the region (e.g., Oregon). Providers often evaluate the developer's intended usage and current consumption patterns before approving a quota adjustment, ensuring resources are allocated efficiently. The procedure and processing time for quota increases vary by provider.

What is the GPU supply shortage?

The GPU supply shortage occurs when demand for graphics processing units far exceeds supply, driven by growth in gaming, cryptocurrency mining, and AI workloads. Manufacturing bottlenecks, supply chain disruptions, and scalping have further limited availability, causing higher prices and delays.

How does the GPU shortage affect cloud GPU providers?

Cloud GPU providers rely on physical GPUs to offer virtual GPU services. The shortage limits their capacity, leading to availability constraints and higher costs for users. However, cloud GPUs provide an alternative to owning hardware, helping users access GPU power despite the shortage.To mitigate shortages, users adopt multi-cloud strategies and automation tools to optimize GPU access across providers. Meanwhile, manufacturers and cloud platforms innovate to improve GPU efficiency and expand supply, aiming to ease the shortage over time.

What is NVIDIA Confidential Computing, and why is it important for cloud GPU providers?

NVIDIA Confidential Computing is a hardware-based security technology that protects AI models and data while they are actively processed on NVIDIA Hopper and Blackwell GPUs. It creates a trusted execution environment that encrypts data in use, preventing unauthorized access, even from cloud operators or administrators. For cloud GPU providers, this means they can offer customers strong guarantees of data privacy and model confidentiality on shared infrastructure, meet regulatory compliance, and attract security-conscious users without sacrificing performance. This technology enables secure AI workloads at scale across public, hybrid, and edge clouds, making it a critical differentiator for cloud providers relying on NVIDIA GPUs

AI AI Hardware

Top 30 Cloud GPU Providers & Their GPUs in 2026

Sedat Dogan

updated on Jan 22, 2026

See our ethical norms

We benchmarked the 10 most common GPUs in typical scenarios (e.g., finetuning an LLM like Llama 3.2). Based on these learnings, if you:

know which GPU model (e.g., H100) you need, check out the models each GPU cloud provider offers.
don’t know which model to use; see the most cost-effective GPUs
want to see major cloud GPU providers:

Cloud	Brands*	Models**	Combinations***
RunPod	AMD chips like MI300X	33	53
AWS	AWS chips like Trainium	7	19
Azure	Working on own chips	6	14
GCP	Google Cloud tensor processing units (TPUs)	8	30
OCI		6	17
Alibaba Cloud	Alibaba chips like Hanguang 800	5	6
Nvidia DGX		23	23
Vast.ai		25	50
CoreWeave		13	13
AceCloud		9	17

Customers have links and are placed at the top in lists without numerical criteria.

Ranking: Sponsors are linked and highlighted at the top. After that, hyperscalers are listed by US market share. Then, providers are sorted by the number of models that they offer.

* All providers offer Nvidia GPUs. In addition, some cloud providers provide hardware from other AI chip makers, as indicated in this column.

** Distinct Nvidia GPU models offered. For example, “A100 40 GB” and “A100 80 GB” are counted as separate models. Note that different interconnect types (SXM, PCIe) for the same GPU model are grouped together.

*** Distinct multi-GPU combinations offered. For example, “1 x A100 40 GB” and “2 x A100 40 GB” are counted as separate multi-GPU combinations. While cloud providers may offer various GPU counts, our analysis focuses on power-of-2 configurations (2, 4, 8, 16, 32 GPUs) for standardized efficiency comparison.

GPUs can be delivered serverless, as virtual GPUs, or as bare-metal GPUs. While serverless offers the easiest way to manage workloads, bare metal offers the highest level of control over the hardware. If you are specifically looking for these, see the relevant sections:

Serverless GPUs
Virtual GPUs (or vGPUs)

While listing pros and cons for each provider, we relied on our GPU benchmark and online reviews.

What are the major Virtual GPU providers?

Virtual GPUs (vGPUs) are virtual machines that allow multiple users share GPUs over the cloud. They are the most commonly offered form of cloud GPUs. Leading providers include:

Hyperscalers (AWS, Azure, GCP)

Hyperscalers have some common aspects:

Pros

Pre-loaded drivers & apps: Configuring an instance with the right drivers is time-consuming due to dependencies among the GPU chip, its drivers, the operating system, and applications. For example, if Ubuntu 25.0 doesn’t support the NVIDIA Tesla K80 driver, you will need to choose an older version of Ubuntu to work with it.

All top 3 hyperscalers allow users to manage machine images, which facilitates this process. However, users still need to identify the right machine image for the hardware that they select. Names of these services are:

Amazon Machine Images (AMI)
Azure Extensions
GCP Custom Images

Cons

Quota approval is necessary for almost all GPUs. Don’t expect to start a cloud account and start using GPUs immediately.
Latest cards like H100 are frequently unavailable on demand.
It is hard to identify GPU capacity. During our benchmark, we could check the GPU cards that we can launch by region. For example, the AWS price calculator provides this functionality. However, we couldn’t find capacity data for any region. Therefore, we needed to try initiating instances across many combinations of regions and instance types to find a configuration that used GPUs.

Amazon Web Services (AWS)

AWS is the largest cloud platform provider and a leading cloud GPU provider.¹Amazon EC2 (Elastic Compute Cloud) offers GPU-powered virtual machine instances, facilitating accelerated computations for deep learning tasks.

Recent Updates

Savings Plans for EC2 P6-B200 Instances: AWS introduced savings plans for its P6-B200 instances, likely tailored for high-performance AI workloads.

Pros

Offers seamless integration with other popular AWS solutions like:

Straightforward quota process: We applied for each GPU instance separately and received the quota for all GPU types on AWS within about a day of our application, without further discussion.
SageMaker is used for creating, training, deploying, and scaling ML models. SageMaker Studio Lab comes with 15GB of free persistent storage and compute credits.
Redshift, OpenSearch, Amazon S3 (Simple Storage Service), Amazon RDS (Relational Database Services), or other AWS services, which can serve as storage solutions for training data

Cons

Shutting down GPUs took hours during our benchmark. Other providers complete this within minutes.
Fewer GPU options than some GPU-focused providers like Coreweave.
Steep learning curve: As the first and largest cloud, it offers comprehensive capabilities that can make the UI feel cluttered.

Pricing

Spot Instances can offer significant discounts, sometimes up to 90% off the on-demand prices.

Microsoft Azure

Microsoft Azure, the second-largest cloud provider, offers a cloud-based GPU service, Azure N-Series Virtual Machines, that leverages NVIDIA GPUs, as do other providers, to deliver high-performance computing capabilities. This service is particularly suited to demanding applications such as deep learning, simulations, rendering, and AI model training.

Microsoft is also rumored to have started producing its own chips.²

Pros

Straightforward quota process: The process was similar to AWS, but the request form was more time-consuming.
A less steep UI learning curve than with providers like AWS.

Cons

Some users find that certain advanced features within Azure require a high level of technical expertise to configure and manage effectively³

Pricing

See all Azure GPU prices & compare with other providers.

Google Cloud Platform (GCP)

Google Cloud Platform (GCP) is the third biggest cloud platform.

Recent Updates

New Compute Environments: GCP introduced A3 Ultra instances with NVIDIA H200 GPUs, enhancing AI performance.⁴

Pros

Provides the most flexibility (among the top 3 hyperscalers) in CPU, GPU, and storage combinations: We can select a CPU and memory size, then attach one or more GPUs to the instance. This provides more flexibility than choosing specific instance types, as is the case with other hyperscalers.
Easier-to-use UI compared to AWS
Offers some free GPU options for Kaggle and Colab users
Customers can use 20+ products for free, up to monthly usage limits

Cons

Configuring the right CPU, GPU, and storage combination is more complex since almost any combination is possible. Users also need to add the prices of different components (e.g., GPU, storage) to calculate the total price of the instance.
The quota process required filling out complex forms and took us days.

Pricing

See all GCP GPU prices in all regions

NVIDIA DGX Cloud

NVIDIA is the leader in GPU hardware. NVIDIA launched its GPU cloud offering, DGX Cloud, by leasing space in leading cloud providers’ data centers (e.g., OCI, Azure, and GCP).

DGX Cloud offers NVIDIA Base Command™, NVIDIA AI Enterprise, and NVIDIA networking platforms. DGX Cloud instances launched with 8 NVIDIA H100 or A100 80GB Tensor Core GPUs.

An initial customer, Amgen’s research team claims 3x faster training of protein LLMs with BioNeMo and up to 100x faster post-training analysis with NVIDIA RAPIDS.⁵

The offering is enterprise-focused, with the list price of DGX Cloud instances starting at $36,999 per instance per month at launch.

Pros

Support from NVIDIA engineers
Multi-node scaling that can support training across up to 256 GPUs, enabling faster large-scale model training
Pre-configured with NVIDIA AI software for quick deployment, reducing setup time

Cons

Offering is not suitable for firms with limited GPU needs
The service is provided on top of cloud providers’ physical infrastructure. Therefore buyer needs to pay for the margins of both the cloud provider and NVIDIA.

IBM Cloud

The GPU offered by IBM Cloud supports a flexible server selection process and integrates seamlessly with the architecture, applications, and APIs of IBM Cloud. This is accomplished through a globally distributed network of interconnected data centers.

Pros

Powerful integration with IBM Cloud architecture and applications
Worldwide distributed data centers increase data protection

Cons

Limited adoption compared to the top 3 providers.⁶

Oracle Cloud Infrastructure (OCI)

Oracle ramped up its GPU offering after formalizing its partnership with NVIDIA.⁷

Oracle provides GPU instances in both bare-metal and virtual machine formats for quick, cost-effective, and high-efficiency computing. Oracle’s Bare-Metal instances enable customers to run tasks in non-virtualized environments. These instances are accessible in regions such as the United States, Germany, and the United Kingdom, with availability under both on-demand and interruptible pricing models.

Customers

Oracle serves some of the leading LLM providers like Cohere, a company that Oracle also invested in.⁸

Pros

Wide range of cloud products and services. Among the tech giants’ cloud services, only OCI offers bare metal GPUs.⁹ For GPU cluster users, only OCI offers RoCE v2 for its cluster technology among the tech giants’ cloud services.
Cost-effective compared to other major cloud providers
Offers provision for a free trial period and some free-forever products

Cons

User interface perceived as clunky and slow by users¹⁰

Some users find the documentation difficult to understand¹¹

The process of starting to use Oracle Cloud compute services was viewed by some users as bureaucratic, complicated, and time-consuming.

RunPod

RunPod is a cloud computing platform specializing in GPU-accelerated services tailored for artificial intelligence (AI) and machine learning (ML) workloads. Designed to streamline AI model development, training, and deployment, RunPod offers a range of features to enhance computational efficiency and flexibility.

Pros

RunPod users mention quick setup times, enabling users to initiate GPU instances within seconds.
The platform offers a diverse selection of GPU configurations, including high-performance options such as NVIDIA H100 PCIe and A100 PCIe.
RunPod users found the interface and CLI intuitive and easy to use for deploying and managing AI workloads.
Users mention over 50 pre-configured templates, including popular frameworks like PyTorch and TensorFlow.

Cons

While the platform is user-friendly, some advanced features may require a learning period to fully utilize their capabilities.

Pricing

GPU instances are billed by the minute, with no extra charges for data ingress or egress. GPU price starts at:

$2.39 per hour for NVIDIA H100 PCIe GPUs
$1.64 per hour for A100 PCIe GPUs.
Network storage is available at $0.05 per GB per month.

CoreWeave

CoreWeave is a specialized GPU cloud provider. NVIDIA is one of CoreWeave’s investors. CoreWeave claims to have 45,000 GPUs and to have been selected as the first Elite-level cloud services provider by NVIDIA.¹²

Jarvis Labs

Jarvis Labs, established in 2019 and based in India, specializes in enabling swift, straightforward training of deep learning models on GPU compute instances. With its data centers located in India, Jarvis Labs is recognized for its user-friendly setup that enables users to start operations promptly.

Jarvis Labs claims to serve 10,000+ AI practitioners.¹³

Pros

No credit card required to register
A simple interface for beginners

Cons

Although Jarvis Labs is gaining momentum, its suitability for your business’s enterprise-level tasks would need to be validated. It seems to be catering to small workloads, as it does not offer multi-GPU instances.

Lambda Labs

Originally, Lambda Labs was a hardware company offering GPU desktop assembly and server hardware solutions. Since 2018, Lambda Labs has offered Lambda Cloud as a GPU platform. Their virtual machines are pre-equipped with predominant deep learning frameworks, CUDA drivers, and a dedicated Jupyter notebook. Users can connect to these instances through the web terminal in the cloud dashboard or directly using the given SSH keys.

Lambda Labs claims to be used by 10,000+ research teams and has a purely GPU-focused offering.

Paperspace CORE by DigitalOcean

Paperspace is a cloud computing platform that offers GPU-accelerated virtual machines to develop, train, and deploy machine learning models.

Paperspace claims to have served 650,000 users.¹⁴

Pros

Offers a wide range of GPUs compared to other providers
Users find the prices fair for the computing power provided
Users find the customer service to be friendly and responsive

Cons

Some users complain about machine availability, both in terms of the free virtual machines and specific machine types not being available in all regions¹⁵

The integrated Jupyter interface is criticized and lacks some keyboard shortcuts, although a native Jupyter Notebook interface is offered

Longer loading or creation times for machines
Monthly subscription fee on top of machine costs can be a downside, and multi-GPU training can be expensive

What is a serverless GPU?

Serverless GPU computing enables users to access powerful GPU resources without managing servers, with providers handling provisioning, scaling, and maintenance. This approach supports pay-as-you-go pricing, often with scale-to-zero functionality that eliminates idle costs, making it ideal for sporadic or unpredictable workloads.

Serverless GPUs are widely used for AI tasks, including training deep learning models, running generative AI applications, and performing batch inferencing, offering significant advantages in simplicity and cost savings over traditional cloud setups.

Explore the serverless GPU providers on Serverless GPUs.

What are bare-metal GPU providers?

Bare-metal GPU providers deliver dedicated physical GPU servers without virtualization, offering direct hardware access for maximum performance and minimal latency.

These solutions are ideal for compute-intensive workloads such as artificial intelligence (AI), machine learning (ML), deep learning, graphics rendering, scientific simulations, and high-performance computing (HPC).

By eliminating the virtualization layer, bare-metal GPUs ensure consistent performance, reduced latency, and full utilization of GPU resources, making them a preferred choice for enterprises and startups with demanding computational needs.

What cloud GPU cloud providers are based in Europe?

European businesses may prefer to keep their data in Europe for

GDPR compliance and data security
Offering faster AI inference services to European users

This is possible with some of the global cloud providers, but there are also European-based cloud GPU providers.

Seeweb

Seeweb is a public cloud provider headquartered in Italy that runs 100% on renewable energy. Seeweb supports IaC via Terraform and offers 5 different GPU models.

Datacrunch.io

Datacrunch provides Nvidia’s A100, H100 RTX6000, and V100 models in groups of 1, 2, 4, or 8. The company is based in Helsinki, Finland, and relies on 100% renewable energy.

OVHcloud

OVHcloud is a public cloud provider headquartered in France. It started offering Nvidia GPUs in 2023 and plans to expand its offering.¹⁶

Scaleway

Scaleway offers H100 instances, operates in 3 European regions (Paris, Amsterdam, Warsaw), and runs 100% on renewable energy. For large-scale users, the Nabu 2023 supercomputer, with its 1,016 Nvidia H100 Tensor Core GPUs, is available.

What are the upcoming GPU cloud providers?

These providers have limited reach or scope, or recently launched their offerings. Therefore, they were not included in the top 10:

Alibaba Cloud

Alibaba’s offering may be attractive for businesses operating in China. It is also available across 20 regions, including those in Australia, Dubai, Germany, India, Japan, Singapore, the USA, and the UK.

However, a US or EU organization with access to top secret data in domains such as state, defense, or telecom may not prefer to work with a cloud service provider headquartered in China.

Cirrascale

Cirrascale specializes in providing a range of AI hardware to research teams. Though they are one of the smallest teams in this domain, with about 20 employees, they offer AI hardware from 4 different AI hardware producers.¹⁷

Voltage Park

Voltage Park is a non-profit that spent funds, including ~$500 million with NVIDIA, to set up 24,000 cloud H100 GPUs. ¹⁸ It offers low-price GPU rental to AI-focused companies like Character AI.

Identify the most cost-effective cloud GPUs

Hover over each dot to see the most cost-effective cloud GPUs:

We benchmarked all cloud GPUs on AWS with common text and image-related tasks. The performance of the same GPU across all clouds was assumed to be the same.

How to start the correct instance for your cloud GPU needs

Making the right decisions when setting up a cloud GPU instance is essential to streamlining the initial setup. Without careful attention to compatibility between the model, OS, and GPU, this process can take hours, significantly increasing costs since GPU providers charge by the hour. By following these steps, you can avoid unnecessary delays and ensure cost efficiency for your project:

Select the Model: Choose the model you plan to use (e.g., YOLOv9).
Identify its dependencies: The model choice directly influences the framework and libraries (e.g., PyTorch, TensorFlow) you’ll need to build and deploy your solution.
Identify the appropriate CUDA version: CUDA is required to run NVIDIA GPUs efficiently. For example, the PyTorch version you need dictates a specific CUDA version.
Use our benchmark to choose the most cost-effective GPU: Leverage benchmark data to select the GPU that provides the best balance of price and performance for your specific workload.
Check whether the GPU is available in your preferred region: Cloud providers often have varying hardware inventories across regions, and some GPUs might not be available in certain regions. Checking whether the GPU is offered helps avoid deployment delays. However, even if a GPU is offered, it may not be available when you request it since it may be overbooked. You can check the GPUs offered per region on:
1. AWS: Price calculator
2. Azure: Pricing calculator
3. GCP: GPU Availability docs
Choose the right operating system: When selecting your setup on the cloud provider, you will need to choose the operating system (OS) and its version. The OS needs to support the required CUDA version and the GPU drivers.
Deploy the drivers and dependencies or choose a system where they are preloaded: You can either manually install the necessary drivers and dependencies or use pre-configured environments provided by cloud providers, such as Azure’s extensions or AWS’s AMIs, to simplify the setup process.

FAQ

Reference Links

Big Three Dominate the Global Cloud Market, Statista.

Microsoft to Debut AI Chip Next Month That Could Cut Nvidia GPU Costs — The Information

The Information

Azure Virtual Machines Reviews 2026: Details, Pricing, & Features | G2

A3 Ultra with NVIDIA H200 GPUs are GA on AI Hypercomputer | Google Cloud Blog

Google Cloud

“NVIDIA Launches DGX Cloud, Giving Every Enterprise Instant Access to AI Supercomputer From a Browser“. NVIDIA.

Same Statista source as above

Oracle and NVIDIA Partner to Speed AI Adoption for Enterprises

Oracle to Deliver Powerful and Secure Generative AI Services for Business

GPU, Virtual Machines and Bare Metal | Oracle

10.

Oracle Cloud Infrastructure Compute Reviews 2026: Details, Pricing, & Features | G2

11.

Oracle Cloud Infrastructure Compute Reviews 2026: Details, Pricing, & Features | G2

12.

CoreWeave Becomes NVIDIA’s First Elite Cloud Services Provider for Compute

13.

Rent GPUs Online | H100 & A100 GPUs from $0.39/hr | Jarvis Labs

14.

Customers | DigitalOcean

15.

Paperspace Core Reviews 2026: Details, Pricing, & Features | G2

16.

Cloud GPU – Cloud instances for AI | OVHcloud Worldwide

17.

Cirrascale Cloud Services | LinkedIn

Cirrascale Cloud Services

18.

reuters.com

Reuters

CTO

Sedat Dogan

CTO

Follow On

Sedat is a technology and information security leader with experience in software development, web data collection and cybersecurity. Sedat:
- Has ⁠20 years of experience as a white-hat hacker and development guru, with extensive expertise in programming languages and server architectures.
- Is an advisor to C-level executives and board members of corporations with high-traffic and mission-critical technology operations like payment infrastructure.
- ⁠Has extensive business acumen alongside his technical expertise.

View Full Profile