Helsinki, Finland — 6. November 2024 — DataCrunch, a leading cloud computing platform dedicated to AI training and machine learning inference, proudly announces the successful deployment of NVIDIA H200 Tensor Core GPUs in its data center in Finland. This strategic move positions DataCrunch as one of the first European providers to offer this next-generation GPU technology, enabling unprecedented levels of AI performance and energy efficiency.
With 141GB of HBMe memory and 4.8TB memory bandwidth the NVIDIA H200 is the most powerful GPU available today, offering breakthrough capabilities designed for complex AI training and inference tasks. With improvements in data throughput, faster multi-GPU scaling, and enhanced memory bandwidth over its predecessor, the H100, the H200 empowers AI developers and enterprises to accelerate projects from research to production.
Key Features of the DataCrunch NVIDIA H200 Deployment:
Best GPU for AI training: According to benchmarks by NVIDIA, the H200 delivers up to
47% performance boost compared to the H100 on single-node GNN training.
Cutting-Edge AI Inference: H200 can deliver up to 2x the inference performance on the 70 billion parameter Llama2 model compared to the H100.
Sustainable Computing: True to DataCrunch’s commitment to green computing, the data center operates on 100% renewable energy with zero water usage. The waste heat generated by the facility contributes to local district heating, demonstrating a balance between high-powered AI infrastructure and environmental responsibility.
Enhanced Scalability: Users benefit from improved H200 HGX parallel processing and low-latency networking to seamlessly scale AI workloads across clusters.
"Deploying the NVIDIA H200 GPUs marks a pivotal moment for us and our users," said Ruben Bryon, Founder and CEO of DataCrunch. “With this addition, we empower data scientists and AI developers across Europe to push the boundaries of innovation while upholding our commitment to sustainability and cost efficiency.”
Supporting Europe’s AI Sovereignty
In addition to delivering cutting-edge technology, DataCrunch’s strategic base in Finland supports European AI sovereignty by offering local, high-performance computing resources, ensuring data residency, and fostering regional innovation.
Availability
DataCrunch Cloud Platform users can access NVIDIA H200 GPUs instances starting today, 5 November, 2024. Dedicated GPU clusters of up to 128x H200 GPUs can also be reserved. To learn more about availability and service options, visit https://datacrunch.io
About DataCrunch
DataCrunch provides high-performance, cost-effective cloud GPU solutions for AI training and interference. Leveraging cutting-edge technology and market-responsive pricing models, DataCrunch empowers customers to achieve their AI goals efficiently and affordably without gatekeeping. Founded in 2020 by Ruben Bryon (CEO), DataCrunch has raised $18M to date.