Nvidia's Advanced Chips Enable German Startup DeepL to Translate Internet in 18 Days
The German AI startup DeepL announced on Wednesday that it has deployed one of Nvidia's latest supercomputing systems, a move that will empower the company to translate the entire internet in just 18 days. This represents a massive leap in processing power, drastically reducing the time from a previous estimate of 194 days.
A Quantum Leap in Translation Speed
This significant acceleration highlights the immense computational power of Nvidia's new hardware. By reducing the time required for such a massive task by over 90%, DeepL can now train and refine its advanced AI translation models at an unprecedented pace, further solidifying its position as a major competitor to services like Google Translate.
Powered by Nvidia's Latest Technology
The Cologne-based startup is deploying Nvidia's DGX SuperPOD system. Each server rack in this system is equipped with 36 of Nvidia's B200 Grace Blackwell Superchips, one of the company's most advanced products on the market. These chips are specifically designed for the intensive workloads of training and running the large AI models that are the core of DeepL's proprietary technology.
Enhancing AI Models and Products
According to Stefan Meeskin, Chief Scientist at DeepL, the primary goal is to provide their research scientists with significantly more computational power to build more advanced models. He explained that this enhanced infrastructure will directly improve current products, such as "Clarify," a tool launched by the company this year. Clarify is an innovative feature that asks users clarifying questions to ensure the correct context is integrated into translations—a capability that Meeskin noted was not technically feasible until recently.
A Strategic Move for Both Companies
This collaboration serves the strategic goals of both companies. For DeepL, it provides the power needed to innovate and maintain a competitive edge in the AI translation space. For Nvidia, it's a key opportunity to showcase the capabilities of its cutting-edge chips and expand its customer base beyond tech giants like Microsoft and Amazon, demonstrating how startups are leveraging their hardware to build next-generation AI applications.
About DeepL
DeepL is a German artificial intelligence company based in Cologne. It has developed its own proprietary deep learning models to provide highly accurate and nuanced machine translation services, positioning itself as a leading alternative to other major translation platforms.
Looking Ahead: The Future of AI Applications
DeepL's adoption of Nvidia's supercomputing technology is indicative of a broader trend where startups are gaining access to powerful tools previously reserved for only the largest corporations. This will likely accelerate the development of more sophisticated and powerful AI applications across various industries, pushing the boundaries of what is possible.
Source: Al Arabiya Business