The AI revolution is here, and businesses across every sector are racing to integrate machine learning capabilities into their operations. With global AI spending projected to surge from $235B in 2024 to $630B by 2028 (IDC), and 70% of all VC activity now driven by AI companies (EY), the demand for GPU infrastructure has never been higher.
Until recently, advanced AI infrastructure was reserved for tech giants and well-funded enterprises. The cost and complexity of dedicated GPU clusters meant smaller teams often had to rely on less powerful solutions or make do with CPU-only infrastructure. But AI innovation isn’t just happening at the top end of the market, it’s increasingly being driven by startups, research groups, and SMBs that need affordable, flexible access to GPUs.
That’s why we see the launch of our new G6 GPU Instances not just as another product release, but as a step toward making high-performance AI accessible to every team.
To deal with this challenge, we introduced our line of Dedicated GPU servers at the start of 2025. And while our bigger customers love this offering, some of you are just beginning your AI journey and require more flexible options.
Today, we’re solving this problem with the launch of GPU-optimized instances in our Public Cloud. Offering a flexible, on-demand solution that delivers enterprise-grade GPU performance at up to 50% lower costs than hyperscalers, all with full EU data sovereignty.
What Are G6 GPU Instances?
G6 instances represent our entry point into on-demand GPU instances, designed specifically for teams that are looking to create their first AI workflows & projects with enterprise-grade performance without enterprise-level complexity or costs. Each instance is powered by 1-4 NVIDIA L4 GPUs paired with AMD EPYC v3 CPUs.
Technical Specifications at a Glance:
| Instance | GPUs | vCPU | Memory | Storage | Price/Hour | Price/Month |
| LSW.G6.XLARGE | 1 | 4 | 16 GiB | 250 GB | €0.61 | €403.88 |
| LSW.G6.2XLARGE | 1 | 8 | 32 GiB | 450 GB | €0.70 | €459.10 |
| LSW.G6.4XLARGE | 1 | 16 | 64 GiB | 600 GB | €0.82 | €541.45 |
| LSW.G6.24XLARGE | 4 | 96 | 384 GiB | 4200 GB | €3.21 | €2107.75 |
Full pricing table available on our GPU Cloud landing page
About NVIDIA L4 GPUs
The NVIDIA L4 Tensor Core GPUs powering our G6 instances are built on the NVIDIA Ada Lovelace architecture with fourth-generation Tensor Cores, designed for universal acceleration across video, AI, and graphics workloads.
Key Specifications:
- 24GB GPU Memory with 300GB/s bandwidth for handling larger models up to several billion parameters
- Up to 485 teraFLOPs of AI performance with multiple precision support (FP32, FP16, INT8, FP8)
- Third-generation RT Cores for advanced ray tracing and rendering capabilities
What This Means for Your Workloads:
- AI Inference & Training: 2.5X higher generative AI performance compared to previous GPU generations, perfect for models like GPT-6B and Mistral-7B
- Video Processing: Up to 120X better AI video performance than CPU-only solutions, enabling real-time content analysis and streaming
- Graphics & Rendering: Over 4X higher real-time rendering performance for 3D applications, game streaming, and virtual workstations
- Flexible Deployment: Each GPU operates independently, ideal for running multiple models in parallel or handling diverse workloads simultaneously
- Cost-Effective Scaling: Optimized precision formats (FP16, INT8, INT4) let you balance speed and cost-efficiency based on your specific requirements
Who Should Use G6 Instances?
G6 instances are designed as a flexible starting point for teams that need enterprise-grade performance without enterprise-level costs. They’re particularly well-suited for:
AI & ML Engineers
Train and run ML models like ResNet-25M and RNNT-120M, as well as smaller LLMs like GPT-6B and Mistral-7B. Perfect for inference workloads in NLP, recommendation systems, and speech recognition.
Developers & Startups
Build, test, and deploy AI prototypes or inference workloads that can scale efficiently as your applications grow. Ideal for teams experimenting with AI capabilities or handling variable computational demands.
Graphics & 3D Teams
Accelerate rendering workflows, image processing, game streaming, and real-time CGI projects with dedicated GPU performance.
Remote Teams & Virtual Workstations
Run GPU-powered virtual desktops for multiple users without performance bottlenecks, enabling distributed teams to access high-performance computing resources on-demand.
Key Benefits of Choosing Leaseweb G6
- Guaranteed Reliability
All G6 instances are backed by a 99.99% uptime SLA, ensuring your AI workloads run reliably so you can focus on innovation, not infrastructure management.
- Cost Efficiency
Achieve 30-50% cost savings compared to hyperscalers without sacrificing performance or reliability. Our transparent pricing model eliminates surprise costs.
- EU Data Sovereignty
As a European cloud provider based in Amsterdam, Leaseweb ensures full GDPR compliance and data sovereignty. Your data stays within EU borders with separated geographic entities for maximum security.
- Human Support
Skip the chatbots and offshore call centers. Our support specialists are physically present at data centers, providing real human expertise when you need it most.
- AWS-Compatible APIs
Manage and automate G6 instances easily with REST APIs, CLI, or Terraform. Compatible with AWS EC2 workflows for seamless integration with existing infrastructure.
- Flexible Billing & Scaling
Choose hourly or monthly billing to match your needs. Scale instances up and down as your workloads demand, with 1TB of pooled internet traffic included monthly.
Getting Started with G6 Instances
Spinning up a G6 GPU instance takes just minutes:
- Create Your Account: Start with our customer verification process (€1.00 one-time fee)
- Access the Portal: Instant access to our Customer Portal once verified
- Configure & Launch: Select your G6 instance configuration and deploy
- Start Building: Begin working your AI/ML workloads immediately
Note: First-time orders require us to complete our “Know Your Customer” process, a legal requirement in the EU.
What’s Next?
The launch of G6 instances marks the beginning of our expansion into GPU cloud. We’re currently rolling out availability across multiple regions:
- Live Now: Netherlands & Germany for fast, EU-compliant GPU workloads
- Coming Soon: US and Canada regions
We’re also continuously expanding our instance types and GPU options based on customer feedback and market demand.
We know great AI ideas can come from anywhere. That’s why we’re making GPU power easier to access, helping teams of all sizes innovate faster and smarter. G6 instances are just the first step in providing flexible, cost-effective GPU infrastructure that scales with your business needs.
Ready to Get Started?
Whether you’re an AI startup looking to prototype your first models, a growing company scaling your ML infrastructure, or an established business adding AI capabilities, G6 instances provide the perfect balance of performance, cost-efficiency, and flexibility.
Start running your AI workloads today with reliable, EU-compliant infrastructure that won’t break your budget.