Magnons: The Future of Energy-Efficient Chips

Magnon Computing: A New Path to Low-Energy Chips

As our demand for more computing power grows, the methods used until now are slowly becoming insufficient. Most notably, all silicon-based chips, be it CPUs, GPUs, or others, tend to consume a lot of energy.

This consumed energy is converted to heat, which must be dissipated from data center servers, or else they can damage electronic components.

This results in energy supply and cooling becoming the major limitations for new AI data centers, as much if not more than the development and supply of advanced chips.

One way to reduce computing energy consumption is to adopt spintronics, which uses electron spins to perform computation. This is a technology already used for hard drives and data storage, but is quickly getting closer to commercial application in computation as well.

A new step toward this goal has been taken by researchers at the University of Delaware and the University of Maryland. They found how spin waves can be turned into electric current, revealing a path toward both a deeper understanding of magnetic materials.

They published their findings in PNAS¹, under the title “Magnon-induced electric polarization and magnon Nernst effects”.

Spintronics Advantages and Potential Applications

Electronic components, such as transistors, are traditionally built from silicon and rely on semiconductors. The 0 and 1 signals in binary indicate the passing or blocking of an electric current.

An alternative way to perform computation is through spintronics devices, which run on the spin of electrons (a fundamental quantum characteristic) rather than the electric current (the flow of electrons).

Source: Insight IAS

Data can be encoded in both the spin angular momentum, which can be imagined as a built-in “up” or “down” orientation of the electron, and orbital angular momentum, which describes how electrons move around atomic nuclei.

Because this contains more information than just 0 & 1, spin can contain more data per atom than traditional electronics.

Spintronics has a few other advantages over classical electronic systems, notably:

• Faster data, as spin can be changed much quickly.
• Less energy consumption, as spin can be changed with less power than it takes to maintain a flux of electrons to create a current.
• Simple metals can be used instead of complex semiconductor materials.
• Spin is less volatile than the semiconductor status, making the data storage more stable.

Swipe to scroll →

Spintronics is already used for hard drives and has allowed data storage capacity to grow over the last decade.

“Spin is a quantum mechanical property of electrons, which is like a tiny magnet carried by the electrons, pointing up or down.

We can leverage the spin of electrons to transfer and process information in so-called spintronics devices.”

Talieh Ghiasi – Postdoc Researcher at Delft University of Technology

Controlling Magnons for Next-Gen Spintronics

Magnons Explained

The researchers focused on spin currents, also called magnons, which act like a magnetic wave, moving the electrons’ spins along their way.

Source: Hubpage

Magnons are the key part of potential spintronics devices. This is because, as the electrons themselves remain stationary as magnons pass through them, there is no heat to be dissipated, the major limiting factor in silicon chips.

In this study, the researchers found that the transport of magnons can induce measurable electric polarization. They used a material structured like a honeycomb made of nickel-phosphorus-selenium (NiPSe3).

They found that the induced net electric polarization is about 1,000x greater than that of previously used materials, like manganese-phosphorus-sulfur (MnPS3).

Beyond Spin Currents

Previously, other research teams found that magnons can be used to convert spin loss into energy, greatly increasing the efficiency of spintronics systems. Other progresses in chiral spintronics and scalable spin-wave networks also brought promising additions to potential future spintronic-based computers.

This study goes even further by proving that magnons, despite their charge-neutrality, can induce electric polarization through their spin and orbital moments.

The way it works is through the “Nerst Effect“, the creation of an electric field when a material is submitted to both a temperature gradient and a magnetic field.

More importantly, they also found that the net electric polarization can be controlled by external magnetic fields (a method called magnon hybridization).

These findings reveal that electric fields could be used to both detect and manipulate magnons under certain conditions by leveraging their spin and orbital angular moment.

Toward a Unified Magnetic Computing Model

This discovery was built on previous work by the same team that looked at creating a unified understanding of spin, magnons, and electrons’ orbits altogether.

“These results integrate advances in magnonics, spinorbitronics, and orbitronics to create a unified framework that can be used to understand and control the manipulation of magnetic states.”

This latest discovery can turn magnons from interesting phenomena underlying spintronics to fully controllable effects, using magnetic fields.

It also creates a theoretical framework that will later be useful to discover or engineer materials with substantial magnon orbital moments, making their magnon properties easier to manipulate at will.

So while spintronics chips are not yet ready to be routinely integrated into AI data centers and computers, they are very quickly progressing into a unified understanding, both at the theoretical and at the practical engineering levels.

In turn, this could have important implications for how we build the next generation of both classical and quantum computers.

In the meantime, spintronics is already used by companies to build memory electronics and sensors, and they might be instrumental in creating more advanced spintronics chips in the future.

Investing in Spintronics

1. Everspin Technologies

Everspin is a branch of Freescale (now known as NXP

(NXPI -3.61%)) dedicated to developing MRAM memory systems. It got spun out of Freescale and IPOed in 2016.

Everspin is considered the leader of MRAM technology (Magnetoresistive Random-Access Memory), inheriting Freescale’s experience of being the first to commercialize an MRAM chip in 2006.

Because MRAM is a memory that persists even in the absence of a current, it is increasingly used in sensitive use cases where critical data is too important to risk loss.

Driven by pervasive applications such as data analytics, cloud computing, both terrestrial and extraterrestrial, artificial intelligence (AI), and Edge AI, including Industrial IoT, the market for persistent memory is projected to grow at a CAGR of 27.5% between 2020 and 2030

Everspin

Source: Everspin

The company estimates the market will reach a $7.4B size by 2027. The company has had no debt and positive free cash flow since 2021.

Everspin MRAM products are currently occupying a small but growing niche, serving markets where reliability is crucial, like aerospace, satellites, data recorders, patient monitoring devices, etc.

Source: Everspin

The growth of chipsets, AI, and synaptic systems might also be a long-term boost for the company.

2. NVE Corporation

Another leader of spintronics, NVE has been working on this technology since its first patent in MRAM technology in 1995. It produces spintronic sensors and isolators, mostly used in measurement and sensor systems for cars, gears, medical devices, power supplies, and other industrial devices.

Source: NVE

This puts NVE in a somewhat different category than Everspin, with NVE more of an industrial company with a strong position in a niche market (magnetometer using spintronics), while Everspin is more of a memory/computing company working with and in competition with the likes of Intel, Qualcomm, Toshiba, and Samsung, who are also developing their own MRAM product.

It can make the stock more (or less) attractive depending on investors’ profiles, with NVE’s stock more likely to appeal to more conservative investors seeking a dividend yield and safety.

Study Referenced

^{1. D.Quang To, Federico Garcia-Gaitan, Yafei Ren, et al. Magnon-induced electric polarization and magnon Nernst effects. PNAS.  October 23, 2025. https://doi.org/10.1073/pnas.2507255122}