27 Jun 2024
Winners Announced for the Large AI Grand Challenge
The European Commission announced the winners of the Large AI Grand Challenge: for innovative AI start-ups from Europe will share the prize of €1 Million and 8 million computational hours.
The winners are:
- Lingua Custodia (France) - A fintech company specialising in AI and Natural Language Processing (NLP) for the finance sector, aiming at enhancing the fintech sector operations with solutions designed to achieve speeds five times faster than current systems.
- Unbabel (Portugal) - A language technologies company headquartered in Lisbon, combining AI and human translation for multilingual support, encompassing all 24 official EU languages.
- Tilde (Latvia) - Experts in language technologies, offering machine translation and AI-powered chatbots, targeting Balto-Slavic languages spoken by 155 million individuals within the EU and candidate countries.
- Textgain (Belgium) - An AI start-up enabling companies and governments to gain insights from unstructured data through predictive text analytic and focusing on the analysis of hate speech, an area of significant concern that has historically received limited attention.
These four start-ups will share a total prize of €1 million and an allocation of 8 million GPU hours on two of the world-leading EuroHPC JU supercomputers, LUMI and LEONARDO. The awarded supercomputing time will be essential for developing their large-scale AI models over the next 12 months and will enable them to reduce training times from years to weeks. Following this period, the winners are expected to release their developed models under an open-source license for non-commercial use or publish their research findings.
The Large AI Challenge received a total of 94 proposals, showcasing the competitive nature of Europe's AI landscape.
On top of this competition organised by the Commission, and given the huge success of the initiative, EuroHPC JU allocated additional compute time to the supercomputer MareNostrum 5 hosted by Barcelona Supercomputer Centre. Therefore, 800,000 computational hours will be allocated to the 5th ranked proposal from Multiverse Computing, a quantum computing start-up focusing in enhancing the energy efficiency and speed of large language models.
For more information: