Here comes Colosseum, Italian supercomputer for Ai among the largest in the world. And Nvidia is involved
This was announced by iGenius, a Milan-based company led by Uljan Sharka. Fundamental to the partnership with the Californian giant
3' min read
Key points
3' min read
The new Italian supercomputer for artificial intelligence is called Colosseum. It was announced by iGenius, a Milan-based company led by Uljan Sharka, already known for releasing AI software that gave it the label of Italian OpenAI. Today, the Milan-based company - which is also based in the United States - announced the launch of one of the world's largest Nvidia Dgx SuperPODs, equipped with the now famous Nvidia Grace Blackwell superchips, to support the deployment of advanced AI models dedicated to customers in the financial, healthcare and public sector sectors globally.
The Colosseum supercomputer,' iGenius explained, 'is equipped with Nvidia Dgx GB200 systems containing thousands of Nvidia Grace Blackwell superchips designed for real-time training and inference for LLM models with over a trillion parameters.
iGenius will use the Colosseum supercomputer to realise advanced AI applications, including open-source generative AI training and large language models with over a trillion parameters. Colosseum will be located in a region of southern Italy (the company has not disclosed the precise location for security reasons) and will be equipped with a state-of-the-art liquid cooling system and enable a performance of 115 exaflops (115,000,000,000,000 operations per second), all powered by renewable energy from Italy.
Colosseum - they wrote in a note from iGenius - 'represents the basis for the next phase of the collaboration between iGenius and Nvidia to develop AI models for sectors requiring the highest levels of security, reliability and data accuracy, such as financial consulting, healthcare and public administration'.
The iGenius AI models 'will be developed using the Nvidia AI Enterprise software platform, the Nvidia Nemotron architecture, and the Nvidia NeMo framework, and will be offered as Nvidia Nim microservices. iGenius plans to offer customers the ability to use the new models and microservices as fully managed applications, including cloud options or on-premise solutions with NVIDIA technology to meet industry and government regulations for AI.
