Basura and Big Data Launch Screaming Bezzy Server Toy

Creative studio Basura and electronic music project Big Data have collaborated on a satirical campaign called Adopt a Data Center, featuring a limited-edition plush toy named Bezzy. Designed to look like an anthropomorphic server rack with googly eyes, the handmade plushie emits a piercing, high-pitched screech when squeezed. The sound is an actual recording captured by a resident living near a massive data center in Virginia, mimicking what neighbors describe as a non-stop car alarm.

Only 20 of the Bezzy dolls have been produced, retailing for $99 each on the official campaign website and the Big Data storefront. The launch coincides with a new single from Big Data, the musical project of producer Alan Wilkis, who is well-known for the 2013 hit song Dangerous featuring Joywave. Along with the physical toy, the campaign includes satirical stickers with slogans like "Open your heart. And your power grid." Clicking the adopt now button on the project's website redirects users to resources and research detailing the societal impacts of data infrastructure.

The project highlights the growing bipartisan backlash against the rapid expansion of physical AI infrastructure across thousands of acres in the United States. As tech companies build out massive facilities to power computationally expensive models, local communities are increasingly protesting the resulting noise pollution, heavy energy consumption, and reliance on gas power. According to Basura founder Rajeev Basu, the campaign is based on dozens of interviews with residents living near these facilities.

For AI practitioners and developers, the campaign serves as a stark reminder of the physical and social footprint of digital products. While software optimization often focuses on algorithmic efficiency, the physical reality of training and running large models has tangible consequences for local communities. As public scrutiny intensifies, developers may face pressure to consider not just the computational performance of their models, but also the geographic and environmental impact of the data centers hosting them.