China Revives a 50-Year-Old Technology That Consumes 200 Times Less Energy Than Digital

February 1, 2026 - 04:06

In a bold move to address the soaring energy demands of artificial intelligence, Chinese researchers are reviving a technology from the 1970s: analog computing. This approach, long overshadowed by today's ubiquitous digital processors, is demonstrating a staggering potential to run AI models using 200 times less energy than conventional hardware.

The core innovation lies in using analog circuits to perform the complex matrix multiplications fundamental to AI. Unlike digital chips that process information as binary ones and zeros, analog chips manipulate continuous electrical signals, directly mimicking the neural networks they aim to accelerate. This allows for computations to occur almost instantaneously and with minimal power dissipation at the physical level.

Early prototypes are already showing promise, reportedly matching the performance of advanced digital GPUs on specific AI tasks while consuming only a tiny fraction of the electricity. This breakthrough is critical as the world grapples with the environmental impact of massive data centers required for AI development.

The push signifies a strategic effort to leapfrog current technological constraints. If successfully scaled, this revival of analog computing could redefine the hardware landscape, enabling more powerful and accessible AI systems that are sustainable to operate, potentially shifting the global balance in the race for artificial intelligence supremacy.