Pull to refresh
Logo
Daily Brief
Following Why Ranks Sign Up
The third magnetic state: racing to build AI-era memory

The third magnetic state: racing to build AI-era memory

New Capabilities

Japanese researchers prove ruthenium dioxide films can store data using altermagnetism—a newly confirmed form of magnetism that could break the memory bottleneck strangling AI

December 26th, 2025: AI-Era Memory Breakthrough Announced

Overview

A team in Japan just proved that ultra-thin films of ruthenium dioxide exhibit altermagnetism—a third fundamental class of magnetism that remained hidden in plain sight until 2019. The breakthrough combines the stability of antiferromagnets with the fast electrical readout of ferromagnets, addressing the core limitation that keeps magnetic memory slow and bulky.

The stakes are massive. AI systems are drowning in a memory crisis—processors can compute 60,000 times faster than they could 20 years ago, but memory bandwidth improved only 100-fold. Moving data from memory burns 500 times more energy than actual processing. Altermagnetic memory could theoretically switch bits 1,000 times faster than today's technology while packing denser without interference, cutting through the bottleneck that's forcing AI data centers into a supply crunch expected to last until 2027.

Play on this story Voices Debate Predict

Key Indicators

3rd
Fundamental magnetic class
First new category discovered since antiferromagnetism in the 1930s
1000x
Potential speed increase
Altermagnets could switch in picoseconds vs. nanoseconds for ferromagnets
$100B
Projected market by 2040
Global altermagnetic materials and applications market forecast
500x
Energy waste ratio
Data movement consumes 500x more energy than computation in AI systems

Interactive

Exploring all sides of a story is often best achieved with Play.

Ever wondered what historical figures would say about today's headlines?

Sign up to generate historical perspectives on this story.

Log in Sign Up

People Involved

Zhenchao Wen
Zhenchao Wen
Lead author on Nature Communications paper
 Takeshi Seki
Takeshi Seki
Collaboration lead from Tohoku University
 Tomáš Jungwirth
Tomáš Jungwirth
First predicted altermagnetism in 2019

Organizations Involved

National Institute for Materials Science
National Institute for Materials Science
National Research Institute
Lead institution on ruthenium dioxide research

Japan's only national research institute dedicated exclusively to materials science.
Tohoku University
Tohoku University
Research University
Collaboration partner on spintronics research

Leading Japanese university with world-class spintronics research center.
Czech Academy of Sciences
Czech Academy of Sciences
Research Institution
Theoretical prediction and experimental confirmation of altermagnetism

Led theoretical prediction of altermagnetism and experimental proof at Swiss Light Source.

Timeline

  1. AI-Era Memory Breakthrough Announced

    Announcement

    NIMS publicly announces RuO₂ altermagnetic breakthrough as solution for AI memory bottleneck. Team plans to develop advanced memory devices exploiting natural speed and density advantages.

  2. Tohoku Hosts Major Spintronics Symposium

    Conference

    MSSp2025 symposium in Sendai brings together researchers working on materials science and spintronics for sustainable futures, including altermagnetic memory applications.

  3. Ruthenium Dioxide Demonstration Published

    Research

    NIMS-led team publishes Nature Communications paper proving altermagnetism in ultra-thin RuO₂ films. First demonstration in practical material suitable for manufacturing at scale.

  4. Altermagnetism Named Top Physics Breakthrough

    Recognition

    Scientific community recognizes altermagnetism as major 2024 physics breakthrough. Researchers worldwide begin searching for altermagnetic materials suitable for manufacturing.

  5. First Experimental Proof: Manganese Telluride

    Research

    International team at Swiss Light Source proves altermagnetism exists in manganese telluride using photoemission spectroscopy. Discovery validates 2019 theory and opens path to applications.

  6. Altermagnetism Theoretically Predicted

    Research

    Tomáš Jungwirth and colleagues at Czech Academy of Sciences and University of Mainz identify magnetic materials with spin structures that don't fit ferromagnetic or antiferromagnetic classifications. Experts debate whether a third category could have remained unnoticed.

Scenarios

Predict which scenario wins. Contrarian picks score more — points lock in when the scenario resolves.

Log in to predict. Track your picks, climb the leaderboard. Log in Sign Up
1

Altermagnetic Memory Reaches Commercial Production by 2028

NIMS and partners successfully scale RuO₂ film production and demonstrate working memory devices. Major semiconductor manufacturers license the technology, integrating altermagnetic cache into AI accelerators by 2028. The global altermagnetic materials market grows toward the projected $100 billion valuation by 2040 as the technology replaces MRAM in high-performance applications. This scenario depends on solving manufacturing challenges—growing high-quality films at scale remains complex and costly.

Discussed by: Market Research Reports Inc., industry analysts tracking advanced materials and spintronics
Consensus —
2

Competing Memory Technologies Win the Race

While altermagnetic research advances, established memory technologies iterate faster. Samsung's 14nm embedded MRAM, Stanford's hybrid gain cells, or processing-in-memory architectures capture the market before altermagnetic devices reach production readiness. Semiconductor manufacturers hesitate to invest in altermagnetism without broader validation, choosing incremental improvements to proven technologies over architectural leaps. Altermagnetism remains an academic curiosity or niche application.

Discussed by: Industry observers noting parallel development of STT-MRAM, hybrid gain cells, and processing-in-memory
Consensus —
3

Fundamental Integration Barriers Slow Adoption

RuO₂ altermagnetic films prove difficult to integrate with existing CMOS fabrication processes. Issues with thermal budgets, interface quality, or reliability under real-world conditions push commercialization timelines beyond 2030. The technology works in laboratories but incorporating these materials into actual semiconductor designs requires engineering advances that take longer than market forecasts anticipated. Development continues but on a slower trajectory, with commercial deployment delayed 5-10 years beyond optimistic projections.

Discussed by: Materials science researchers highlighting challenges in semiconductor integration
Consensus —

Historical Context

The Von Neumann Bottleneck (1945-Present)

1945-Present

What Happened

John von Neumann's 1945 computer architecture separated processing from memory, connecting them via a bus. As processors accelerated dramatically—60,000x faster over 20 years—memory bandwidth improved only 100x. By 1977, John Backus described the problem in his Turing Award lecture, calling it the von Neumann bottleneck. Today, AI systems spend 500 times more energy moving data than computing with it.

Outcome

Short Term

Cache systems and separate memory paths provided temporary relief through the 1990s-2000s.

Long Term

The bottleneck became critical as AI emerged, forcing development of processing-in-memory, compute-in-memory, and novel memory technologies to break the architectural limitation.

Why It's Relevant Today

Altermagnetism attacks the same problem from a different angle—faster, denser memory that reduces the gap between processing speed and memory bandwidth that's strangling AI.

Antiferromagnetism Discovery (1930s)

1930s

What Happened

Louis Néel predicted and researchers subsequently confirmed antiferromagnetism—materials where magnetic moments align antiparallel, producing zero net magnetization. The discovery took decades to move from physics curiosity to practical application. Only in the past five years have antiferromagnets been seriously investigated for memory applications after European researchers demonstrated electrical control of antiferromagnetic spins.

Outcome

Short Term

Remained primarily a physics phenomenon studied for fundamental understanding through the 20th century.

Long Term

Recent breakthroughs in electrical control opened applications in ultra-fast, high-density memory immune to external magnetic interference.

Why It's Relevant Today

Altermagnetism follows a similar path—fundamental physics discovery that could take years or decades to commercialize, but combining advantages of both ferromagnets and antiferromagnets for memory.

MRAM Development (1990s-Present)

1990s-2025

What Happened

Magnetic RAM promised non-volatile storage with SRAM speed since the 1990s, but commercialization proved difficult. Spin-transfer torque MRAM emerged in the 2000s. By 2025, Samsung demonstrated 14nm embedded MRAM with the smallest cell size yet, while companies like Everspin partnered with GlobalFoundries to commercialize STT-MRAM. The technology took 25+ years to reach commercial viability, with widespread adoption still limited.

Outcome

Short Term

Initial MRAM products faced scaling, cost, and integration challenges that limited adoption.

Long Term

STT-MRAM now targets automotive, edge AI, and embedded applications, offering SRAM-class performance with 100x lower standby power and 2.5x higher density.

Why It's Relevant Today

Shows the timeline for radical memory technologies: altermagnetic memory may follow a similar 20-30 year path from lab demonstration to widespread commercial deployment.

Sources

(9)
+3
ScienceDaily Tohoku University Phys.org Nature Communications SciTechDaily IBM Research IEEE Spectrum ScienceDaily Phys.org