INFLECTION.
The Weekly Magazine of Innovation
Issue 04  ·  Friday, 13 November 2026  ·  Deep Dive — Materials & Magnetism
FERROMAGNETANTIFERROMAGNETALTERMAGNET
Cover Story · Materials

Physics Just Found a Magnet
With Nothing to Show for It

For nearly 90 years, magnetism came in two flavors. This week a team at SLAC confirmed a third: altermagnetism, a state that looks nonmagnetic from outside but is not. The surprise is where the value hides.

→  Why a magnet that attracts nothing may be the useful one
→  A squeeze that rewires a crystal · A $21.5B drone bet · A 23× membrane
→  The Contrarian: the discovery is a map, not a device
Contents
Dispatch · Editor's note02
Feature · The Magnet That Hides03
Feature · Breakthroughs04
Feature · So What? Field Notes & Glossary05
Against the Grain · The Contrarian06
Signals07
By the Numbers08
The Long View · Sources09
Dispatch

Zero is not the same
as nothing.

Engineers learn early that a quantity can cancel and still matter. A tug-of-war at a dead standstill is not an absence of force; it is two forces held in balance. Magnetism has had a version of this story for almost a century, and this week it got a new chapter.

In 1933 antiferromagnets were identified: materials whose tiny atomic magnets point in alternating directions, so the outside world sees no net magnet at all. Together with ordinary ferromagnets, that made two categories. The term "altermagnet" appeared only in 2022. On 4 November, ScienceDaily relayed SLAC's confirmation, published in Science, that a third category is real, in manganese telluride.

Our lens for this issue: when a thing cancels from the outside, ask what it still does on the inside. Read what follows with one question in your pocket: in your own systems, which "net zero" is quietly hiding structure you could use?

Inflection · Issue 0402
Feature · Materials

The Magnet That Hides

Altermagnets cancel their magnetism on the outside and keep it on the inside. Confirming one took a synchrotron, a twist of X-ray light, and two years of repeats.

By the Inflection desk · Researched with Claude · 7 Nov 2026

Picture a ballroom where every dancer spins, half clockwise and half counterclockwise. Seen from the balcony the room looks calm: no net rotation. A ferromagnet is the room where everyone spins the same way, a fridge magnet. An antiferromagnet is the calm room, dancers in perfect alternation. For nearly 90 years those were the only two rooms physicists believed in. Altermagnets are a third: the room looks calm from the balcony, yet the dancers on the clockwise side are arranged differently from the counterclockwise side, and anyone who walks the floor can feel the difference.

The setup: two boxes, one stubborn crystal

Magnets matter to computing because information can ride on electron spin, the property that makes each electron a tiny magnet. This field is called spintronics. The trouble is that the two known categories each give up something. Ferromagnets are easy to read but leak stray fields, so neighbours interfere. Antiferromagnets are quiet and dense but hard to read, because their net signal is zero.

The altermagnet was proposed as an escape from that trade: SLAC's report describes materials that "combine the best of both." The catch was proof. Manganese telluride (MnTe) had long been filed under ordinary antiferromagnets. To reclassify it, experimenters needed a fingerprint that only a third class could leave.

The fingerprint: handedness

The fingerprint they chose was chirality, handedness. Lead author Daniel Jost and colleagues fired circularly polarized X-rays, light whose electric field corkscrews left or right, at MnTe on the ID32 beamline of the European Synchrotron (ESRF) in France, and compared the scattering from the two polarizations. A difference between left and right is the signature of a "chiral altermagnon," the title of their Science paper.

The next page shows how that evidence stacked up, and how a second team in Tennessee found a way to steer the same family of materials with a squeeze.

Inflection · Feature03
Feature · Breakthroughs

1. The proof took two experiments, not one

The SLAC-led team ran its X-ray experiment in 2024 and repeated it in early 2025 with reproduced results. In a field where a single beamtime can mislead, a repeat a year apart is the difference between a hint and a verdict. Jost: "This chiral signature gave us the solid experimental ground to say altermagnetism stands in a class of its own."

2. The class is about 4 years old

The word "altermagnets" was introduced in 2022, while antiferromagnets date to 1933. A theoretical category went from naming to experimental confirmation in roughly four years. ESRF postdoc Rachel Nickel: "Altermagnetism was only proposed a few years ago, and the rapid progress in confirming these models is incredible."

When a category goes from proposal to proof in four years, the scarce thing is no longer the idea. It is the list of materials that fit it.Inflection thesis

3. A squeeze turns the knobs

At Oak Ridge National Laboratory, researchers placed tiny crystals of MnTe and iron sulfide (H-FeS) between two titanium plates, gently compressed them, and watched with neutron diffraction at the Spallation Neutron Source. The strain changed the magnetic domains and the anomalous Hall effect, an electrical signal that reveals how the magnetism steers electrons.

4. Two journals, one message

The work appeared in Advanced Materials (on H-FeS, called an "altermagnet candidate") and Physical Review X (on MnTe). Pengcheng Dai: "By controlling the magnetic domains, we can directly influence the transport behavior."

5. Honest limits, stated plainly

The coverage gives no pressure value, no operating temperature and no device. SLAC's own account says "significant development is still needed," and that the next step is learning to control altermagnetic states with light and electricity. The strain study treats applications as conditional: "if scientists can identify and develop suitable materials."

Inflection · Feature04
Feature · So What?

From lab to market: what it means

The claimed prizes, per the sources, are spintronics, computer memory, quantum devices and more energy-efficient data centers. None is demonstrated. So where might value land first? Three readings, flagged as ours rather than the papers'.

First, the toolchain before the chip. Verifying an altermagnet needed a synchrotron beamline and neutron scattering. Whoever makes those measurements faster and cheaper owns a toll road, whatever the devices do.

Second, strain as a dial. A crystal whose magnetic behavior responds to a squeeze is a sensor and a switch in one. That is a design principle: materials that tune mechanically can be tuned on a chip by stress layers.

Third, search beats serendipity. A confirmed class with clear symmetry rules is a search specification. Computational screening can propose candidates; the lab then confirms with the same two-step approach as this week's work.

The practical question for a technology leader: if memory and sensing eventually shift to materials like these, which of your roadmap assumptions rely on stray-field tolerance that no longer needs to exist? This is our inference, not a finding of the sources.

Field Notes
How you prove a hidden magnet

1 · Light. Send X-rays that corkscrew left, then right, into a crystal of manganese telluride.

2 · Compare. Measure what scatters in each case. An ordinary antiferromagnet treats both the same.

3 · Read the difference. A left/right mismatch is a handedness signature: the altermagnet fingerprint. Repeat a year later to be sure.

Analogy: two identical-looking gloves. Only a hand tells you which is which.
Glossary
Spintronics
Electronics that use the spin of electrons, not just their charge, to carry information.
Chirality
Handedness: a property that differs from its mirror image, like left and right gloves.
Anomalous Hall effect
A sideways voltage produced by a material's own magnetism, used here to read magnetic behavior.
Uniaxial strain
Squeezing or stretching a crystal along a single direction.
Inflection · Feature05
The Contrarian
Against the Grain

The headline is a device.
The story is a map.

The consensus reading of this week: a new magnet means faster, cooler memory, so watch the chip makers. Our Move 37 is less comfortable. Nobody has shown a device, and the paper never claimed one. What was proven is a classification, and classifications are where search industries begin.

Think of the periodic table before anyone mined it: a table of empty boxes told prospectors where to dig. A confirmed third class of magnet tells computational chemists what symmetry to screen for, and tells instrument makers what signature to look for. The first money is likely in finding materials and verifying them, not in selling a memory cell. That is an inference, and it could be wrong.

Why the consensus disagrees

One material, one technique. The confirmation is in MnTe by resonant X-ray scattering. The strain work calls H-FeS only a "candidate." Breadth of confirmation is still thin.

No numbers where engineers need them. The coverage we read gives no operating temperature, no strain magnitude, no switching speed or energy. A memory cell is built from those, not from a symmetry class.

Control is the unsolved part. SLAC itself says the next step is learning to control altermagnetic states with light and electricity, and that "significant development is still needed."

Applications are promises. Memory, quantum devices and greener data centers appear as hopes in press coverage. Histories of "next-generation memory" are full of materials that were beautiful in the lab.

Bet on whoever can find and verify the next altermagnet faster, not on whoever promises the first chip.The Contrarian · verdict, held loosely
Inflection · The Contrarian06
Signals
$21.5BRobotics

Zipline nearly triples its valuation in under a year

Zipline closed a $1.5 billion round at a $21.5 billion post-money valuation on 5 November, led by Paradigm, WCM Investment Management and Layer Global. The prior mark was $7.6 billion after a $600 million raise in January 2026. Walmart adds 300 delivery sites; Dallas-Fort Worth saw more than 1 million deliveries in 2026 with a 19-minute median delivery time in October. The article notes backers will watch whether sites become recurring volume rather than pilots.

SOURCE · The Robotics Media, "Zipline Raises $1.5 Billion at $21.5 Billion Valuation," 5 Nov 2026 (citing DRONELIFE)
23×Materials · Energy

A "clogged" filter could change oil refining

KAIST and Georgia Tech report in Nature an uncoated polyacrylonitrile membrane with channels under 2 nanometers that separates crude oil at room temperature, with permeation about 23 times higher than prior leading membranes, stable for 28 consecutive days. Distillation above 350 °C is replaced as a first step. The 31.6% energy and 37.6% CO2 cuts are projections from process simulations, not refinery results.

SOURCE · Choi et al., Nature 2026; 654:955, doi.org/10.1038/s41586-026-10677-3 (via ScienceDaily, 5 Nov 2026)
5,400 mAutonomy

Driverless electric trucks go to the thin air

EACON says 60 battery-electric autonomous haul trucks reached full site-wide operation at Tibet Julong Copper's Zhibula mine at roughly 5,400 meters, which it calls a world first above 5,000 meters. Claims include an 80% cut in driver staffing needs and about 12,000 tonnes of carbon avoided a year. These are EACON's own figures and need a year of continuous operation to confirm.

SOURCE · The Robotics Media, "EACON's 60 Autonomous Electric Haul Trucks…," 5 Nov 2026 (citing PR Newswire, 4 Nov)
Inflection · Signals07
By the Numbers
~90 yrsThe time magnetism was thought to have two categories
2022Year the term "altermagnets" was introduced
23×Crude-oil membrane permeation vs. prior leading membranes
31.6%Projected refining energy cut (simulation, not yet demonstrated)
$21.5BZipline post-money valuation after its $1.5B round
5,400 mAltitude of EACON's 60-truck autonomous electric fleet
1,100 TWhYearly energy of global distillation, about 130 one-gigawatt nuclear plants
Inflection · By the Numbers08
The Long View

Structure hides in the quiet places

Three stories, one pattern. A magnet that attracts nothing turned out to be a new class. A filter that "clogs" in a way that helps separates oil without boiling it. A drone company's valuation tripled because its boring metric, deliveries that repeat, finally moved. In each, the interesting thing was where an apparent zero, a blockage, or a pilot hid a structure.

The discipline to take away is unglamorous: when a result looks like nothing, change the instrument. Circularly polarized light made the invisible visible. A squeeze made it adjustable. Ask of your own dashboards which cancelling terms are hiding signal, and what measurement would separate them.

Sources & Further Reading

1. ScienceDaily, "SLAC scientists find conclusive proof of a mysterious third type of magnetism," 4 Nov 2026 — https://www.sciencedaily.com/releases/2026/11/261104012256.htm

2. Jost et al., "Chiral altermagnon in MnTe," Science 2026; 394(6823):541 — https://doi.org/10.1126/science.adv3196

3. SLAC release, 29 Oct 2026 — https://www6.slac.stanford.edu/news/2026-10-29-slac-researchers-find-proof-altermagnetism-new-magnetic-phase

4. ScienceDaily, "A tiny squeeze dramatically changes the magnetism of exotic materials," 4 Nov 2026 — https://www.sciencedaily.com/releases/2026/11/261101010541.htm

5. Yao et al., Advanced Materials 2026; 38(56) (H-FeS); Liu et al., Phys. Rev. X 2026; 16(3), doi 10.1103/589s-s1yy (MnTe); ORNL release — https://www.ornl.gov/news/simple-squeeze-changes-magnetic-behavior-in-emerging-materials/

6. ScienceDaily, "Scientists discover that a clogged filter could revolutionize oil refining," 5 Nov 2026 — https://www.sciencedaily.com/releases/2026/11/261103003207.htm ; Nature paper — https://doi.org/10.1038/s41586-026-10677-3

7. The Robotics Media, Zipline $1.5B round, 5 Nov 2026 — https://theroboticsmedia.com/article/zipline-raises-1-5-billion-21-5-billion-valuation-drone-delivery

8. The Robotics Media, EACON 60 autonomous electric haul trucks, 5 Nov 2026 — https://theroboticsmedia.com/article/eacon-60-autonomous-electric-haul-trucks-zhibula-copper-mine

Note: the "So What?", the Contrarian verdict and the Long View are Inflection's own inference, not findings of the sources. Figures marked as projections or company claims are unverified beyond the cited pages; the coverage was read via secondary news pages, not the full papers.

Inflection · The Long View09
INFLECTION.
The Weekly Magazine of Innovation
Next Issue · Friday, 20 November
The Lens Question

Which of your net-zero numbers is quietly hiding structure you could use?

Researched, written & designed with Claude. Typeset in Poppins & Lora on the Anthropic palette.