Italian AI startup iGenius is set to finalize a significant data center project in southern Italy this summer, utilizing Nvidia technology and necessitating a $1 billion investment over five years, according to CEO Uljan Sharka.
Sharka revealed that the project's expense led the startup to extend a funding round from an undisclosed source.
He highlighted that the new supercomputer involved in the project would be capable of processing "115 billion of billions" of calculations per second, a tremendous leap compared to Europe's prior supercomputing capabilities.
Sharka expressed ambitions to surpass previous benchmarks due to the enhanced capabilities of Nvidia's "Blackwell" chips, boasting a 35-fold increase in computing power while consuming 25 times less energy than their predecessors.
The data center is slated to accommodate approximately 80 of Nvidia's top-tier servers, designated as GB200 NVL72 machines, each equipped with 72 Blackwell chips.
With a timeline in mind, Sharka stated, "Keeping the current pace, we aim to complete the project this summer," attributing the selection of southern Italy as the project site to the region's surplus renewable energy capacity.
Established in 2016, iGenius has risen to prominence as one of Europe's few AI startups valued at over $1 billion, competing with industry players like France's Mistral and Germany's DeepL.
iGenius recently launched Colosseum 355B, a robust language model tailored for clients in heavily regulated industries with stringent data protection requirements, developed using cutting-edge Nvidia technology.
Sharka emphasized iGenius's discrepancy from competitors such as OpenAI by specializing in open-source AI software models for chatbots, providing companies the autonomy to execute models on their premises for enhanced data security.
Regarding the LLM model unveiled, Sharka disclosed ongoing discussions with potential users from financial services, heavy industry sectors, and governmental entities.