Unbabel Among First AI Startups to Win Millions of GPU Training Hours on EU Supercomputers

Cover Image

The European Union has announced the winners of its "Large AI Grand Challenge" initiative, aimed at accelerating homegrown innovation in large-scale AI model development. Four startups will share €1 million in prize money and receive eight million GPU hours to train their models on two of the EU's high-performance computing (HPC) supercomputers over the next 12 months.

The winning startups are:

French fintech Lingua Custodia, which uses natural language processing (NLP) for financial document processing
Belgian startup Textgain, which analyzes unstructured data, such as social media chatter, using NLP
Latvian startup Tilde, which offers machine translation and AI-powered chatbots in Balto-Slavic languages
Portuguese startup Unbabel, which blends machine translation with human expertise for customer service and productivity applications

Unbabel, a Y Combinator-backed translation business, is one of the most prominent winners. The startup has been around for nearly a decade and has raised close to $100 million in funding.

The EU's AI Challenge received a total of 94 proposals. The winners will release their developed models under an open-source license for non-commercial use or publish their research findings at the end of the training period.

The EU's supercomputing initiative, EuroHPC JU, currently has eight operational supercomputers, with two providing the allocation of eight million GPU hours to the four winners: Finland-based Lumi and Italy-based Leonardo.

A fifth startup, Spain-based Multiverse Computing, just missed out on prize money but will receive 800,000 computational hours on another supercomputer, Spain's MareNostrum 5.

This group of European startups building large-scale AI models won't be the first to tap into HPC hardware. French AI model maker Mistral was a participant in an early pilot phase of the supercomputing provision last summer, using Leonardo to run small experiments.

The EU has recognized the need to reconfigure and retool its HPC infrastructure for the generative AI age. Back in January, the Commission announced a package of "AI innovation" measures, including proposals for upgrading the supercomputers and building out a support layer to improve accessibility for AI startups.