BEIJING, Nov 3 (Market Insider) — In a stunning technological leap, researchers at Peking University have unveiled a revolutionary analog computing chip that they claim operates up to 1,000 times faster than the world’s most powerful digital processors — including Nvidia’s H100 — while consuming just one percent of the energy.
Published on October 13 in Nature Electronics, the study details how this breakthrough could reshape the global semiconductor landscape, with profound implications for artificial intelligence (AI) and 6G communications. The Chinese research team says the chip overcomes two long-standing barriers that have constrained digital processors for decades: energy efficiency and data bandwidth limitations.
From Ancient Concept to Modern Revolution
Unlike digital chips, which process information in binary sequences of 1s and 0s, the new analog processor uses continuous electrical currents that flow through a lattice of Resistive Random-Access Memory (RRAM) cells. Each memory cell can both store and process data, eliminating the need to constantly transfer information between memory and the central processor — a major bottleneck in modern computing.
When tested on complex wireless communication algorithms — such as matrix inversion in MIMO systems — the chip achieved digital-level accuracy at a fraction of the energy cost. According to the researchers, once fine-tuned, it outperformed Nvidia’s H100 and AMD’s Vega 20 by up to 1,000 times in processing speed.
“Traditional digital architectures are hitting a performance ceiling as data volumes explode,” the paper notes. “Analog computation offers 1,000x higher speed and 100x better energy efficiency than today’s most advanced digital chips.”
“Old Tech” Reimagined for the AI Age
Analog computing is far from new — it dates back over 2,000 years, to ancient mechanical systems like the Antikythera mechanism of Greece, an intricate gear-based device that calculated celestial orbits. Yet the concept was largely abandoned in the modern digital era because of precision challenges.
Recent advances in memory hardware, however, have revived the analog dream. The Peking University team designed a hybrid chip that merges analog speed with digital accuracy by using two complementary circuits, one for fast, approximate calculations, and one for incremental refinement, achieving precision close to digital systems.
This dual-circuit model enables massively parallel, real-time computation, ideal for next-generation applications such as AI model training, autonomous robotics, and 6G signal processing, where millions of simultaneous data streams must be analyzed instantly.
Built With Existing Tech — Ready for Scale
Perhaps most remarkably, the researchers fabricated the chip using commercially available manufacturing processes, meaning it could be scaled up for mass production in the near future — a significant step as China pushes to reduce dependence on Western semiconductor technology.
The team’s next goal is to expand chip size and processing capacity, paving the way for large-scale analog processors that could power future AI systems and data centers — potentially reshaping the global semiconductor hierarchy.
If successful, this technology could mark China’s most significant computing breakthrough in decades — a fusion of ancient analog principles with modern precision engineering that might redefine what’s possible in the AI era.
