100x less electricity: The breakthrough that could solve AI’s massive energy crisis

vm_admin

2 months ago

100x less electricity: The breakthrough that could solve AI’s massive energy crisis

By Colin Collins
Publication Date: 2026-03-29 01:17:00

AI systems are consuming energy at an increasingly unsustainable rate, prompting researchers to explore fundamentally different computational approaches. A new neurosymbolic method combines statistical learning with rule-based reasoning, significantly improving both efficiency and accuracy in robotic tasks. Photo credit: Shutterstock

A new hybrid AI approach could dramatically reduce energy consumption while improving reliability.

Artificial intelligence doesn’t just change software. This also leads to a sharp increase in electricity consumption. In the United States alone, AI systems and data centers consumed approximately 415 terawatt-hours of electricity in 2024, according to the International Energy Agency. This is more than 10% of the country’s total energy production and is expected to double by 2030.

This trend raises a difficult question for the future of AI: Can these systems become more powerful without dramatically increasing the cost of power?

Researchers at Tufts…