Introducing UpCloud’s On-Demand NVIDIA L40S GPUS

We’re stoked to announce the general availability of our new NVIDIA L40S GPU instances, bringing a new era of computational power and flexibility to your fingertips. Starting from just €1.11 per hour. This isn’t just about faster computing; it’s about unlocking new possibilities for your most demanding applications, from cutting-edge AI to high-fidelity media rendering. 

With UpCloud’s open source, European ethos, our GPUs have no vendor lock-in, no data exposure, and no third-party dependencies. You have full control over your data, models, and infrastructure, billed hourly, without rigid contracts or long-term commitments.

Powering next generation workloads

Combining powerful AI compute with best-in-class graphics and media acceleration, the L40S GPU is built to power the next generation of data center workloads, from generative AI and large language model (LLM) inference and training to 3D graphics, rendering, and video.

Diverse Use Cases 

The applications for GPUs extend far beyond their origins in gaming graphics. Our new GPU offerings are designed to empower a wide range of industries and use cases:

Artificial Intelligence & Machine Learning (AI/ML):
This is where GPUs truly shine. Powering neural networks, predictive models, classification problems (like identifying objects in images), and recommendation engines (e.g., “You liked X, so you might like Y”). Whether you’re fine-tuning large language models (LLMs) or building sophisticated AI applications, GPUs provide the necessary horsepower.

Video Processing & Media Rendering:
Our end users asked for accelerated video effects, transcoding, and rendering in the cloud – and we delivered! UpCloud’s L40S GPUs feature 3 dedicated NVENC/NVDEC engines, enabling highly efficient, hardware-accelerated video encoding and decoding for simultaneous stream processing and faster media workflows.

Natural Language Processing (NLP):
From contextual advertising and news scraping (especially financial data) to search and recommendation engines, GPUs can significantly speed up the processing and indexing of vast amounts of text data.

High-Performance Computing (HPC):
Applications like Monte Carlo simulations, complex scientific modeling, and large-scale data analysis all benefit immensely from the parallel processing capabilities of GPUs.

Gaming & Entertainment:
For game publishers and live sports/entertainment streaming, GPUs are crucial for rendering high-quality assets, delivering immersive experiences, and handling real-time video streams.

Enterprise:
In private cloud environments, GPUs can revolutionize tasks like document summarization for healthcare records, efficient document retrieval, and prompting for common themes within large datasets, significantly accelerating processes compared to manual methods.

Three reasons to choose UpCloud GPU Servers.

We understand that when it comes to high-performance computing, you need more than just raw power. That’s why UpCloud GPUs are designed with your needs in mind, offering a unique blend of performance, control, and cost-efficiency.

#1 – Build AI Without Giving Up Control.

Unlike some providers, we believe you should maintain full control over your data, models, and infrastructure. With UpCloud GPU, there are no lock-ins, no data exposure, and no third-party dependencies.

• Launch your first GPU instance in minutes: No contracts, no hassle. See what on-demand GPUs should feel like.
  
• Privacy-first by design: Our GPU infrastructure is hosted in GDPR-compliant data centers in Finland, ensuring strong jurisdictional protections for your AI workloads.
  
• Open-source: Whether you’re working with Ollama, PyTorch, or anything from Hugging Face, our infrastructure supports your open-source choices without imposing restrictions or hidden service layers.
  
• Sovereign infrastructure: You dictate what runs, when, and how. Your architecture remains yours.
  
• No platform dependencies: We don’t force you into proprietary ML platforms or opaque orchestration layers, giving you the freedom to build your way.

#2 – Cost-Efficient GPU Power for Real Workloads.

Stop paying for GPU time you don’t use. UpCloud’s hourly billing model aligns perfectly with agile development and iterative workflows.

• Only pay for active time: Whether you’re training models in short sprints or running inference jobs periodically, you’ll never pay for idle infrastructure again.
  
• Avoid overprovisioning: Our usage-based model allows you to scale compute precisely to your workload, eliminating the waste associated with traditional monthly GPU billing.
  
• Power & cost control: Ideal for startups and project teams requiring high-end GPUs without long-term commitments or significant capital expenditure.
  
• Transparent pricing: No hidden egress charges or resource bundling. Just clear, usage-based pricing.

#3 – Flexible GPU Access, ZERO Lock-In.

Spin up GPU resources exactly when you need them, without rigid contracts or long-term commitments.

• Instant start, zero commitment: Go live today for exploration or production without needing to talk to sales or sign a 12-month agreement.
  
• Freedom from fixed billing: Many providers still mandate fixed-month billing or pre-allocated GPU blocks, limiting your flexibility. UpCloud is built for agile development and on-demand resource access.
  
  Ready to unleash the power of GPUs for your next project? 

NVIDIA L40S are now available for Public Cloud customers, hosting in our Helsinki “HEL2” data center. Expansion to new data center locations will begin shortly.  

Find out more about our GPU Cloud servers and Get Started today!