The Daily Byte
Breaking
Welcome to The Daily Byte — real stories from across Pakistan.
Saturday, 21 November 2026 · Pakistan

The Daily Byte

Stories from Pakistan that deserve to be heard
Minor World

Anatrix AX100 Chip Solves Maths in One Step

By The Daily Byte ·16 Nov 2026 ·4 min read
A computer processor chip held above a circuit board

Chinese chip developer Anatrix has unveiled the AX100, a chip it describes as the world's first fully integrated analog computing chip able to solve matrix equations in a single step. The company says the chip completed tape-out in October and solved a matrix equation in under 100 nanoseconds in tests, according to a report in China's Science and Technology Daily on Sunday, 15 November.

Matrix equations sit at the heart of much modern computing. They help artificial intelligence models learn, robots adjust their movements and mobile base stations process signals. Conventional digital chips usually solve them through a series of calculations. Anatrix says its chip takes a different route, representing the equation as an electrical circuit and reading off the answer once the circuit settles into equilibrium.

How the chip works

Instead of crunching numbers step by step, the AX100 encodes the matrix as a circuit and lets physics do the work. When the circuit reaches a stable state, the voltages it holds are the solution. Anatrix chief executive Xiang Rui told the newspaper that each solve consumed about 80 milliwatts of power in the company's tests.

The approach builds on research published by a team at Peking University in the journal Nature Electronics in October 2025. That work showed how resistive memory arrays could solve matrix equations with high precision. Anatrix says it has turned that research into a chip in about a year.

What the tests showed

Anatrix says it tested the chip in three areas. In artificial intelligence experiments involving two models, the method tested with the chip cut the computing cost of updating model parameters by 92 percent and 56 percent respectively. In embodied AI, the tests showed faster motion control and route planning, which could help robots respond more quickly as their surroundings change.

In wireless communications, the company's evaluation found that the chip could process more data with the same amount of energy than the digital chips and graphics processors used for comparison on a specific signal-processing task. The company also says the chip can be manufactured using mature production processes, rather than the most advanced fabrication technology.

From university research to a chip

The AX100 is the latest example of Chinese firms looking beyond conventional digital chips as access to advanced semiconductors remains restricted. Xiang said the chip was fundamentally different from GPU-style digital chips in architecture and underlying logic, and that it gave a Chinese company a first opportunity to take part in defining an important computing paradigm.

Analog computing is not new. Early computers were analog machines, and researchers have long known that analog circuits can solve some problems very quickly. The difficulty has been precision: small errors in analog hardware can corrupt the result. The Peking University research behind the AX100 aimed to solve that precision problem, and Anatrix says its chip puts the method into a complete, manufacturable product.

What is not yet proven

Important limits apply. The description of the AX100 as the world's first of its kind, the test results and the comparisons with digital chips are the company's own claims, reported by Science and Technology Daily. They have not been independently verified, and no independent benchmark results have been published. The chip also solves a specific class of mathematical problem; it is not a general-purpose processor and cannot run the full range of software that a conventional chip can.

Even so, the unveiling adds to a growing field of analog and other non-digital approaches to computing, as the industry looks for ways to run AI workloads faster and with less energy. Whether the AX100 moves from laboratory tests to real products will depend on further testing, manufacturing yields and demand from customers building AI systems, robots and communications equipment.

Sources: Khyber Mail, reporting the Science and Technology Daily; Nature Electronics, Peking University analog matrix computing research.