DeepL Announces It’s the First to Implement Nvidia’s DGX SuperPOD in Europe. This Computer Platform Should Make DeepL’s AI Platform Faster and More Accurate.
DeepL is the First Company in Europe to Deploy an Nvidia DGX SuperPOD with DGX GB200 Systems. The New Infrastructure is Operational in Sweden at Partner EcoDataCenter. According to DeepL, the Upgrade Provides a Strong Increase in Computing Power, Allowing the Company to Offer Its Translation Services Faster and More Accurately.
SuperPOD
The New SuperPOD Enables DeepL to Drastically Reduce Processing Time for Large Translation Tasks. For Example, the System Can Theoretically Translate the Entire Internet into Another Language in 18.5 Days, Where This Would Previously Take 194 Days. The Translation of Extensive Works Such as the Oxford English Dictionary or Proust’s In Search of Lost Time Now Takes Place in Fractions of Seconds.
read also
DeepL Introduces Clarify for More Accurate Translations
Research and generative functions
The Extra Computing Power Provides DeepL with More Possibilities in the Field of Research and Development. According to the Company, This Will Be the Basis for New Generative AI Functions Within the Platform. These Should Lead to an Improved User Experience, with Faster and More Interactive Translation Sessions. In the Long Term, DeepL Aims to Deploy Multimodal Models That Can Combine Language with Other Forms of Data.
read also
DeepL Adds New LLM Translation Model and DeepL Write to API
The Nvidia DGX SuperPOD Uses a Liquid-Cooled Rack Design and Can Scale Up to Tens of Thousands of GPUs. The System is the Successor to DeepL Mercury, the Company’s Previous Supercomputer Cluster, and is the Third SuperPOD Deployed by DeepL.