The University of Albany (UAlbany) in New York has unveiled a new AI supercomputer.
UAlbany said the system is one of the most advanced AI supercomputers to be hosted by a US university and will be used to support research into the treatment of diseases.
Costing $16.5 million and housed at the university’s data center, the machine contains 24 Nvidia DGX systems for a total of 192 tensor core GPUs, and has three miles of fiber optic cabling. No information about whether the system contains A100s or H100s or its total compute power has been detailed.
It was paid for by the $75m awarded to the university by Governor Kathy Hochul in her April 2022 budget, with half allocated for the completion of the university’s engineering building and the other half set aside to expand UAlbany’s AI supercomputing capacity.
“This new AI supercomputer is a major step forward for the University at Albany — for our faculty, for our staff, for our researchers, and especially for our students,” said UAlbany president Havidán Rodríguez. “UAlbany has been recognized as a national leader for our thoughtful and holistic approach to integrating AI into our teaching and research, and this new AI supercomputer is the next critical evolution of that strategy.”
In February 2024, IBM installed a cluster of IBM Artificial Intelligence Unit (AIU) prototype chips at UAlbany’s Center for Emerging Artificial Intelligence Systems. The $20m joint venture made the university one of the biggest AI test beds in the US and was the first AIU cluster to be installed on a university campus globally.
Debuted in late 2022, IBM’s AIU is a complete system-on-a-chip (SOC) that has been designed to run and train deep learning models faster and more efficiently than general-purpose CPUs. It consists of 32 individual AI processing cores, contains 23 billion transistors, and was built using 5 nm node process technology.
