Expert Discusses Widespread Opposition to Hyperscale Data Centers Across US
Dennis Talluto discusses the widespread opposition to hyperscale data centers across the United States, attributing it to their massive size, acreage, and significant power demands for AI workloads. He explains how large language models drive the need for these facilities, despite acknowledging media overstatement of energy and water challenges. Talluto also touches on concerns regarding eminent domain, state-local government relations, and the societal implications of data collection through AI.
Dennis Talluto, founder and principal of LVL Strategies, discusses on the Overton Window Podcast the growing opposition to hyperscale data center development across the United States. He attributes the "paranoia" and "noise level" primarily to the sheer size and scope of modern data centers, which have escalated from 50,000-100,000 square feet drawing four megawatts decades ago to half a million square feet demanding 500 megawatts to a gigawatt for a single site.
Talluto, who has 45 years in tech and has consulted for data center developers, explains that the shift to artificial intelligence and large language models is a key driver for these massive facilities. Unlike transactional computing, AI requires immense processing power and storage for conversational applications, leading to data centers that often need their own on-site power generation and specialized cooling, demanding substantial acreage.
While Talluto acknowledges media exaggeration of energy and water usage challenges, he highlights the distinct features of hyperscale centers compared to smaller, less visible data centers, such as those prevalent in the Detroit metropolitan area. He also raises concerns about eminent domain, the interplay between state and local government, and the societal implications of AI's ability to collect and mine personal data, transforming users into valuable "products" for data center owners and their clients.