Published in News

US IRS to buy Nvidia SuperPod AI supercomputer

by on17 February 2025


Advanced Machine Learning

The Internal Revenue Service (IRS) is preparing to purchase a state-of-the-art Nvidia SuperPod AI supercomputer to bolster its machine learning capabilities, with applications ranging from fraud detection to taxpayer behavior analysis.

This acquisition is part of a broader governmental effort to integrate artificial intelligence into federal operations, replacing traditional bureaucracy with automated systems.

According to a recently uncovered 5 February acquisition document, the IRS computing center in Martinsburg, West Virginia, will house the advanced AI cluster.

The setup will consist of 31 Nvidia servers, each powered by eight of the company's flagship Blackwell processors. While the hardware has not yet been purchased or installed, SuperPod systems typically start at $7 million. The contract materials also mention a significant memory upgrade from Nvidia.

Despite being smaller than the AI-training data centers operated by companies such as OpenAI and Meta, the SuperPod remains a formidable computing resource. Nvidia markets the system as a comprehensive AI training and inference solution, making it a powerful tool for machine learning applications. Similar government AI investments include MITRE Corporation’s $20 million SuperPod setup, aimed at developing AI models for federal agencies.

The exact nature of the IRS’s intended use for the SuperPod remains unclear as the taxman is being really quiet about it. It appears to have been fleshed out under the previous President as a 2024 Treasury Inspector General for Tax Administration report identified 68 AI-related projects underway at the IRS. The purchase materials state that the IRS requires a “robust and scalable infrastructure” to handle complex machine learning workloads, highlighting the SuperPod’s role in this effort.

The IRS Research, Applied Analytics, and Statistics (RAAS) division will oversee the SuperPod. RAAS has previously employed machine learning for automated fraud detection, identity theft prevention, and deeper insights into taxpayer behavior. The acquisition suggests that the IRS is advancing its AI capabilities significantly, raising questions about how artificial intelligence will shape future tax enforcement and oversight.

Last modified on 17 February 2025
Rate this item
(1 Vote)

Read more about: