GPU servers rental for deep learning

Advantages of working with us

Affordable Prices:

Competitive GPU rent pricing: starting from € 0.02 / minute

High-End Servers:

Enterprise servers with GPU cards: H100 80Gb, A100 80Gb, RTX 3090, RTX 4090, RTX A6000, 6000 Ada, etc

Reliable Data Rooms:

A Tier 3 compliant data rooms supplier, stringently upheld to ISO 27001 data center standards.

Customizable Solutions:

Private networks, system enhancing by adding of RAM, SSDs, HDDs and other special requests.

Secure In-House Infrastructure:

All our servers are under our complete control in a secure environment that excludes third-party access.

Stable Internet Connection, Power Supply and Redundancy:

Four redundant internet connections and NL-IX peering.
All equipment is connected to two UPS feeds.
Fully redundant network infrastructure.

Included Traffic:

The price includes 10TB of traffic per month.

No Vendor Lock-in:

Users can easily migrate data in and out, Transparent Pricing.

How to use LeaderGPU® service

Why do clients trust us?

Trust Logos from independed companies, which confirm that you can Trust us.

We are proud to be partnered with ThuisWinkel. This guarantee that LeaderTelecom B.V. is a reliable supplier which applies highest professional standards and provide high quality services.

Sectigo Trust Logo confirms trust for this site and submitting any confidential information is safe here. It confirms the site belongs to LeaderTelecom BV and passed extended verification.

This seal confirms that our company exists and our financial statements were verified. That is why this site is secure and safe for you.

What our clients say about us

Instructions and Benchmarks

Getting started

Benchmarks

LeaderGPU® for TensorFlow is 300% faster and provide cost-effectiveness by at least 3,2 times comparing to its competitors.
LeaderGPU® for MXNet is 10,7 time faster comparing to its competitors. Refer to benchmark test results.
LeaderGPU® for Machine learning on Tesla V100. Refer to Tesla benchmark results.

Latest News

11-12-23

Wishing you success and luck for 2024

Dear LeaderTelecom B.V. customers, our entire team is happy to wish you a Merry Christmas and a Happy New Year.

30-10-23

LeaderTelecom has received the prestigious FD Gazellen 2023 award - for the fifth time in a row!

As a leading GPU cloud server provider and SSL retailer, LeaderTelecom is excited to announce its victory in the annual FD Gazellen 2023 competition, marking the fifth consecutive win!

More news

Choose the best solution!

Customer may use the Services for blockchain processing

Amount of available servers is temporary limited

We would like to convey our sincerest regret for the inconvenience it may have caused however due to fast growing demand for our GPU capacities we are almost run out of available servers. We would like to advise you on current waiting queue. Our first priority is currently focused on clients which are using servers right now. We are working hard to increase amount of new servers, however all suppliers in The Netherlands are currently experiencing shortage of GPU cards in stock.

Please refer to availability info in the table below.

Dedicated Server Features

Servers

Specialized enterprise-grade GPU-servers (bare-metal servers only)
PCIe 3.0 x16 slot

Billing

10 euro minimum order
Unlimited suspend/resume on minute-based servers
Unlimited quantity for 1 user, among the available servers
No setup fee
No hidden fees

Guaranteed bandwidth

Internal network (LeaderGPU® to LeaderGPU®):

Included internal traffic: UNLIMITED
Burst: 10 Gbps

Internet traffic (LeaderGPU® to Internet):

Burst internet: 1 Gbps
Included internet traffic (monthly based payments): 10 Tb/month
Included internet traffic (weekly based payments): 2.5 Tb/week
Included internet traffic (daily based payments): 300 Gb/day
Included internet traffic (minute/hourly based payments): 0 Gb
Additional 1Gb (not included): € 0,09/Gb

Traffic is billed by either the outbound or inbound traffic, whichever is higher.

Customer service

Email: info@leadergpu.com
Phone: +31 20 7640722
Billing: Monday — Friday: 9:00 - 18:00 CET time
Technical Support: Monday — Friday: 9:00 - 18:00 CET time
Incident reporting: 24/7

Operating systems

Ubuntu 18.04 / 20.04 Server / 22.04 Server: Access via SSH / WebSSH / RDP. Free of charge.
Windows Server Standard 2016 / 2019: Free of charge.

IPs included

1 IPv4/IPv6 address included
Assign additional public IPv4 addresses to your servers — € 10/month per IPv4 address

Additional storage

On the basis of a monthly charge:

256Gb: € 29,69 without VAT / month (€ 0,116/1Gb)
512Gb: € 58,88 without VAT / month (€ 0,115/1Gb)
1Tb: € 116,74 without VAT / month (€ 0,114/1Gb)
2Tb: € 225,28 without VAT / month (€ 0,11/1Gb)
3Tb: € 276,48 without VAT / month (€ 0,09/1Gb)
4Tb: € 307,2 without VAT / month (€ 0,075/1Gb)

Graphical Interface

LeaderGPU® customers can now use graphical interface via RDP out of the box in the following configurations:

Ubuntu 18
Ubuntu 20 Server
Ubuntu 22 Server
Windows® Server 2016
Windows® Server 2019

Data Rooms Features

Power

Meets the standards for a TIER-IV certification
Each server is connected to dual-power feeds A (UPS) - a unique feature by LeaderGPU®
Generators: N+1

Cooling

Stable 24°C
Backup units (N+1)

Security

ISO 27001 and NEN 7510 certification
24/7 on-site security guards, fingerprint scanners for access to different zones,
fenced perimeter, CCTV cameras, intrusion detection

Fire safety

Very early smoke detection (VESDA-system)
Gas fire extinguishing systems

What our partners think about us

CUDA® technology

CUDA® — parallel computing architecture designed by NVIDIA®. The implementation of this architecture allowed to significantly increasing performance of GPU computing.

Advantages of CUDA:

Using the common programming language C
Applying of technologies such as Parallel Data Cache and Thread Execution Manager
Default libraries for FFT and BLAS
Unification of hardware and software parts for GPU computing
Multi-GPU Support
Hardware Debugging
Excellent scalability to different projects
Special CUDA® driver

NVIDIA® H100 GPU

The NVIDIA® H100, powered by the new Hopper architecture, is a flagship GPU offering powerful AI acceleration, big data processing, and high-performance computing (HPC). The Hopper architecture introduces fourth-generation tensor cores that are up to nine times faster than their predecessors, providing a performance boost on a wide range of machine learning and deep learning tasks. With 80 GB of HBM2e high-speed memory, this GPU can handle tasks related to language models or AI with ease.

The NVIDIA® H100 is excellent for:

Inference
Deep learning
High-performance computing (HPC)

GPU Architecture: Hopper
Memory: 80GB HBM2e
NVIDIA CUDA Cores: 14,592 (in PCIe version)
Memory Bandwidth: 2TB/s

GPU NVIDIA® GeForce™ RTX™ 6000 Ada

GeForce™ RTX™ 6000 Ada is the new flagship graphics card. Based on the updated Ada Lovelace architecture, featuring an astounding 18,176 CUDA® cores and 142 third-gen RT Cores, this graphics card is great for a wide range of complex applications, while also being versatile. The amount of GDDR6 video memory with error correction is 48 GB.

NVIDIA® RTX™ 6000 video cards are great for:

3D modelling and rendering
Deep learning
High-performance computing (HPC)
AI/data science tasks

GPU: Ada Lovelace architecture
Memory: 48 GB GDDR6 with ECC
NVIDIA CUDA cores: 18176
RT cores: 142
Memory Bandwidth: 960 GB/s

GPU NVIDIA® GeForce™ RTX™ 4090

The GeForce™ RTX™ 4090 graphics card is the flagship solution with 16384 CUDA cores, updated Ada Lovelace architecture and fourth-generation tensor cores. The Ada Lovelace architecture delivers significant performance and efficiency gains. The efficiency of the new graphics cards is backed up by DLSS 3 technology.

RTX ™ 4090 video cards are great for:

Blockchain processing
Universal GPU card

GPU: Ada Lovelace architecture
Memory: 24 GB GDDR6X
NVIDIA CUDA cores: 16384
Memory Bandwidth: 1018 GB/s

NVIDIA® A10 GPU

NVIDIA® A10 is the Ampere-generation GPU, which offers 72 RT Cores and 24 GB GDDR6 for high-performance computing, rendering, deep learning, etc. As NVIDIA® benchmarks show, A10 delivers up to 2.5X more inference performance compared to T4.

A10 video cards are great for:

3D modelling and rendering
Deep learning
Inferencing

GPU: Ampere architecture
Memory: 24 GB GDDR6
NVIDIA CUDA cores: 9216
Memory Bandwidth: 600 Gb/s

GPU NVIDIA GeForce™ RTX™ 3090 Ti

The RTX™ 3090 Ti is another cutting-edge graphics card from the Ampere generation that is ideal for ray tracing and machine learning. The memory in the RTX™ 3090 Ti is 24 GB of GDDR6X. Memory speed has increased: 21 Gbps compared to 19.5 Gbps memory on the RTX™ 3090. Also the number of CUDA Cores has increased – now there are 10,752 compared to 10,496 in the RTX™ 3090 version. There are 2 more RT cores and 8 more Tensor cores, which fans of 3D rendering and AR/VR development will appreciate.

RTX ™ 3090 Ti video cards are great for:

Blockchain processing
Universal GPU card

GPU: Second-generation RTX™ architecture
Memory: 24 GB GDDR6X
NVIDIA CUDA cores: 10752
Memory Bandwidth: 1.1 TB/s

NVIDIA® A40 GPU

NVIDIA® A40 is the Ampere-generation GPU, that offers 10,752 CUDA cores, 48 GB of GDDR6-memory, 336 Tensor Cores and 84 RT Cores. This video card is ideal for a variety of calculations in the fields of data science, AI, deep learning, rendering, inferencing, etc. Peak memory bandwidth is 696 GB/s. ECC memory is also available in this GPU.

A40 video cards are great for:

3D modelling and rendering
Deep learning
High-performance computing (HPC)
Inferencing

GPU: Ampere architecture
Memory: 48 GB GDDR6
NVIDIA CUDA cores: 10752
Memory Bandwidth: 696 Gb/s

NVIDIA® RTX™ A6000 GPU

The NVIDIA® RTX ™ A6000 GPU is a modern, powerful, next-generation GPU based on Ampere architecture. The second generation of RT cores can significantly speed up rendering. The third generation of Tensor cores makes it possible to accelerate the training of neural networks by up to 5 times. 48 GB of GDDR6 memory will enable you to tackle a variety of data science challenges.

RTX™ A6000 video cards are great for:

3D modelling and rendering
Deep learning
High-performance computing (HPC)
AI and data science challenges

GPU: Second-generation RTX™ architecture
Memory: 48 GB GDDR6
NVIDIA CUDA cores: 10752
Memory Bandwidth: 768 GB/s

NVIDIA® A100 GPU

The NVIDIA® A100 delivers powerful AI acceleration, big-data processing and high-performance computing (HPC). The A100 delivers up to 20 times the performance of the previous generation’s NVIDIA® Volta ™. NVIDIA® A100 Tensor Core technology supports a wide range of maths calculations. 40 GB of high-speed HBM2e memory allows you to solve any task, even the most complex. The memory bandwidth of the A100 is 1.7 times faster than the previous generation.

The NVIDIA® A100 is great for:

Inferencing
Deep learning
High-performance computing

GPU: Ampere architecture
Memory: 40 GB HBM2e
NVIDIA CUDA cores: 6912
Memory Bandwidth: 600 Gb/s

GPU NVIDIA GeForce™ RTX™ 3090

GeForce ™ RTX ™ 3090 is a state-of-the-art graphics card based on the Ampere architecture. It is a second-generation RTX™ architecture that doubles the performance of ray tracing and AI tasks. Improvements affect Ray Tracing (RT) Cores and Tensor Cores. There are also new streaming multi-processors. The new graphics card includes 10496 CUDA™ Cores and 24 GB of GDDR6X memory, which opens up totally new possibilities in machine learning and rendering.

RTX ™ 3090 video cards are great for:

Blockchain processing
Universal GPU card

GPU: Second-generation RTX™ architecture
Memory: 24 GB GDDR6X
NVIDIA CUDA cores: 10496
Memory Bandwidth: 936 GB/s

GPU NVIDIA® GeForce™ RTX™ 4090

RTX ™ 4090 video cards are great for:

Blockchain processing
Universal GPU card

GPU: Ada Lovelace architecture
Memory: 24 GB GDDR6X
NVIDIA CUDA cores: 16384
Memory Bandwidth: 1018 GB/s

GPU NVIDIA GeForce™ RTX™ 2080Ti

GeForce RTX graphics cards are powered by Turing GPU architecture and the all-new RTX graphics platform. These give you performance up to six times that of the previous generation graphics cards and let you game with the power of real-time ray tracing and AI.

RTX ™ 2080Ti video cards are great for:

Blockchain processing
Universal GPU card

GPU: First-generation RTX™ architecture
Memory: 11 GB GDDR6
NVIDIA CUDA cores: 4352
Memory Bandwidth: 616 GB/s

GPU NVIDIA NVLink™ Tesla® P100

NVIDIA® NVLink™ Tesla® P100 - the most advanced graphics accelerators ever created. They are based on a new architecture GPU NVIDIA® Pascal™ and is the world's fastest computer servers with a capacity exceeding hundreds of classic server-based CPU. If you need high performance and accuracy of calculations - Tesla® P 100 is the best choice.

Tesla P 100 GPU's are great for:

3D modeling
Deep learning
High-performance computing

GPU: P100 Pascal architecture
Memory: 16 GB HBM2
NVIDIA CUDA cores: 3584
Memory Bandwidth: 720 GB/s

GPU NVIDIA Tesla® V100

GPU NVIDIA® Tesla® V100 - the most efficient GPU, based on the architecture of NVIDIA® Volta. NVIDIA® Tesla® V100 accelerators, connected by NVLink™ technology, provide a capacity of 160 Gb/s, which allows a whole host of problems to be solved, from rendering and HPC to training of AI algorithms.

Tesla® V100 GPU's can be used for any purpose.

Tesla V 100 GPU's are great for:

Deep Learning
Rendering
High-performance computing (HPC)

GPU: Volta GV100 architecture
Memory: 16 GB HBM2
NVIDIA CUDA cores: 5120
Memory Bandwidth: 900GB GB/s

GPU NVIDIA® Tesla T4

Tesla® T4 - a modern powerful GPU demonstrating good results in the field of machine learning inferencing and video processing. The Tesla® T4 includes RT Cores for real-time ray tracing and delivers up to 40X times better throughput (compared to conventional CPUs).

Tesla® T4 GPU's can be used for any purpose.

Tesla T4 GPU's are great for:

Inferencing
Deep Learning
Rendering

GPU: NVIDIA® Turing™ architecture
Memory: 16 GB GDDR6
NVIDIA CUDA cores: 2560
Memory Bandwidth: 320+ Gbps

Payment Options

Rent GPU servers with per-minute payment from LeaderGPU® right now!

Go to Servers