---
title: "Breakthrough in memory physics: how optimal control theory promises to reduce chip energy consumption by tens of times"
description: "Physicists from the University of Edinburgh have found a way to reduce memory energy consumption by tens of times. Instead of \"brute force,\" they proposed using optimal control theory to switch bits. This could be a lifeline for data centers in the era of generative AI. 🧠⚡💾"
date: 2026-08-02T14:21:29.000Z
lang: en
url: https://xab.info/en/posts/breakthrough-in-memory-physics-optimal-control-theory-reduces-energy-consumption
tags: [university-of-edinburgh, memory-storage, artificial-intelligence, physics, energy-efficiency]
publisher: "XAB.info"
---

# Breakthrough in memory physics: how optimal control theory promises to reduce chip energy consumption by tens of times

![Futuristic memory chip with neon data streams, symbolizing a breakthrough in memory physics and reduced power consumption through optimal control theory](https://xab.info/media/2026/08/02/fiziki-edinburgskogo-universiteta-snizili-potreblenie-pamyati/fiziki-edinburgskogo-universiteta-snizili-potreblenie-pamyati-1.webp)

The development of generative artificial intelligence has posed a severe challenge to the global technology community: modern systems require a colossal amount of energy to store and process data. The problem of memory energy consumption is becoming critical, and scientists around the world are looking for ways to mitigate the consequences of this trend. New research by physicists from the University of Edinburgh offers a solution that could radically change the approach to writing data to electronic memory.

### Abandoning the "brute force" method

Researchers have concluded that current methods for changing a bit in a magnetic memory cell are far from ideal. Today, the data writing process often resembles a "brute force" method. To change the state of a cell, a sufficiently powerful pulse directed against the initial magnetization is used, which requires significant energy expenditure.

Instead, the team proposed using a magnetic field with a specially calculated varying shape. Such a pulse switches the magnetic element from the "0" state to the "1" state along the most energetically favorable trajectory, minimizing system resistance.

### Mathematics at the service of physics

To calculate ideal pulses, scientists applied optimal control theory — a powerful mathematical tool that allows finding the least costly way to achieve a specified system state. Computer modeling was conducted using single-layer van der Waals magnets, such as Fe3GaTe2, Fe3GeTe2, and CrSBr.

The modeling results showed impressive efficiency of the new approach. In these materials, spins were successfully rotated coherently within 1–10 picoseconds. At the same time, the amplitude of the optimized magnetic field turned out to be more than ten times lower than when using traditional pulses.

### Figures and prospects

The energy gain from implementing the new methodology turned out to be substantial. In the models considered, the energy of a single switch was approximately 0.94–9.7 nJ (nanojoules). For comparison, when using ordinary magnetic field pulses, this figure ranged from 42.8 to 91.2 nJ.

However, the potential of the technology does not end there. Additional optimization of device size, magnetic anisotropy, and damping coefficient theoretically allows reducing energy costs to femtojoules. According to the authors' calculations, with certain parameters, this method could be more economical than even modern STT-MRAM and SOT-MRAM technologies, where the magnetic state is changed by currents creating spin-transfer torque or spin-orbit torque.

It is important to note that the proposed methodology is not limited to magnetic memory. It can be applied to control currents and lasers, allowing us to talk about optimizing data handling in a broad sense. This opens the way to creating more energy-efficient computing systems of the future.