Is this magnet headline for real?

Short versionNo curve, no claim. The demagnetization curve is the one document that turns a performance statement into something a stranger can check, and it is cheap to produce if you actually have a magnet.

A permanent magnet's entire performance is one curve. The numbers on a datasheet — remanence, the two coercivities, energy product — are not four separate facts, they are four points on it, and everything an engineer needs to know about whether the magnet survives its application is in the curve's shape rather than in any single number. If that sentence is new, the long version is here.

Measuring it is not exotic. It is a standardized test with a standardized instrument, defined internationally in IEC 60404-5,¹ and any company that makes magnets has run it thousands of times, because it is also how they decide what to ship and what to scrap. So when a company does not publish a curve, there are only two explanations. Either they do not have a magnet yet — which is a perfectly respectable place to be, and they should say so — or they have one and do not like how the curve looks.

When a curve IS published, read it for four things:

Both curves, not one. A proper plot shows the normal curve (B against H) and the intrinsic curve (J against H) together. The coercivity that decides whether a magnet survives a motor lives on the intrinsic curve. A plot showing only B is not wrong, but it is the flattering half.

Temperature. A single room-temperature curve is the beginning of a datasheet, not a datasheet. Real producers publish a family of curves at 20, 60, 100, 150 °C and upward, because that is where the knee moves and the knee is what kills parts in service. Room temperature is where every material looks its best, which is presumably why it is often the only temperature offered.

What was measured. A hysteresigraph running the standard test on a real magnet, or a vibrating-sample magnetometer running a few milligrams of aligned powder in epoxy? Both produce a curve that looks like a curve. Only one of them describes something you could bolt into a machine. Powder measurements skip the three things that hurt: alignment loss in a real pressed part, density, and the demagnetizing field of a real shape. If the method and the sample mass are not stated, that omission IS the finding.

Whether the numbers reconcile with each other. This one you can do on your phone.

In practiceI came at coercivity backwards. My first real work in magnetics was figuring out how to erase things — hard drives and floppy disks — which is the same property viewed from the wrong end, and it taught me early that taking magnetization away is far harder than putting it there. That asymmetry is the whole commercial value of a permanent magnet, and it is precisely the property that new materials are short of. When somebody leads with how much field their material produces, I am already thinking about how easily it gives it back.

Short versionCompanies that make things need customers and cash flow. Companies that are still trying to make it happen need investors. Both are legitimate. Which one a piece of communication is aimed at tells you where the company actually is, regardless of what it says about itself.

This is the fastest read on the list, and it costs nothing: open the website and ask who the sentences are written for.

• A grade table with tolerances and coating options
• A total-addressable-market slide
• Lead times, minimum order quantities, and a way to request a quote
• Memoranda of understanding and letters of intent described as "agreements"
• Quality certifications with numbers on them — ISO 9001, IATF 16949 for automotive, AS9100 for aerospace
• Capacity stated as "will be capable of," with no date for "is"
• Datasheets with revision numbers and dates
• Groundbreaking photography as the main visual asset
• A request for YOUR drawing
• An executive team weighted toward finance, communications and government affairs

In practiceBe especially skeptical of "in talks with." A friend in the investment world once asked me what I thought of a company that was telling prospective investors it was in talks with a large manufacturer. I happened to be at that manufacturer at the time, and I was the magnetics subject-matter expert in the room for those very talks. The entire substance of them was, roughly: nice to meet you, but you are in fantasyland. They were not lying about the meeting. They simply understood that a meeting and a relationship look identical from the outside, and priced accordingly. My friend got hours of diligence back for the cost of one email — which is the argument for asking someone in the industry before you ask the company.

The careers page is the underrated version of the same test. A company that is making product is hiring process engineers, furnace and vacuum technicians, metrology and quality people, and maintenance. A company that is raising is hiring communications, business development and policy. Neither is a character flaw. But job postings are the one corporate document nobody edits for narrative, which makes them unusually honest.

Then there is the single most efficient question in this entire article, and it takes one email: can I buy one? Price per kilogram, lead time, minimum quantity, and a certificate of conformance. A producer answers with numbers or with a polite "not in that size." A company that is not yet producing will either tell you so — which is a good answer, and tells you where they are — or will spend three paragraphs not telling you.

Short version"The only." "The first." "Number one." Parse one of these carefully enough and it is usually technically true — and usually a distinction without a difference. Companies with real technical achievement tend to describe it in language so modest that it is easy to scroll past.

Superlatives in this industry are almost always load-bearing on their qualifiers. Take the claim apart and remove them one at a time. "The only US producer of sintered neodymium magnets" has at least four qualifiers doing work: only, US, producer, and sintered. Strip "sintered" and the claim may be false. Strip "US" and it is certainly false. And "producer" is doing the heaviest lifting of all, because producing one magnet and producing a thousand tons of them are separated by roughly a decade and several hundred million dollars, and the word does not distinguish between them.

This is not the same as lying. The claims are generally accurate, which is why legal cleared them. The question to ask is not "is this true" but "if it is true, what does it get me" — and often the answer is nothing, because the category was drawn narrowly enough to contain exactly one company.

The inversion is the useful part. Genuine technical achievement in magnetics tends to arrive sounding boring. A grade that holds the same remanence with a couple more kilooersteds of intrinsic coercivity, at the same heavy-rare-earth content, is a serious piece of process metallurgy that took years — and it gets announced as a new grade in a table. Nobody writes a press release about it, because the people who need it already understand what it means. While loud claims are not proof of weakness, in my experience the correlation runs the wrong way often enough to be worth a flag.

What a confident technical claim actually looks like: numbers with conditions attached to them. A named standard. An explicit statement of what the material cannot yet do. Ranges rather than single values. And a willingness to say "at 20 °C" out loud rather than leaving you to assume it.

Short versionSquare feet, headcount, disciplines, IP position, feedstock, and which process steps actually happen inside the building. Nearly all of this is public, and the arithmetic between the numbers is where announcements come apart.

Space. Ten thousand square feet in a multi-tenant building is a laboratory, a machine shop, or a finishing operation. It is not a magnet plant, and it cannot become one without moving. For calibration, the newest fully integrated Western plant — MP Materials' Independence facility in Fort Worth — is a 250,000 square foot building² producing on the order of 1,000 metric tons of finished magnets a year with roughly 160 to 200 people.³ That is about 250 square feet of building per annual ton, and a couple of tons per person per year.⁴

So when an announcement pairs a 10,000 square foot leased unit with a claim of several hundred tons a year, it is claiming an order of magnitude more output per square foot than the best new plant in the West. That is not impossible — Chinese plants are tighter, and a finishing-only operation is denser still — but it is a question, and it should have an answer in one sentence.

People, and which disciplines. A scrappy team of five can invent a material. They cannot run presses, sintering furnaces, machining, coating, magnetizing and inspection across shifts. So look for whether the org chart contains the jobs the process requires: a metallurgist, somebody who owns vacuum and furnaces, a metrology and quality function, and maintenance. Maintenance is the one nobody puts in the pitch deck and the one that decides whether the plant runs in year two.

Education and track record. Who there has education in what they are doing, and has anyone done this at scale before, anywhere? Magnet making is powder metallurgy, and powder metallurgy is a craft with a long apprenticeship. Adjacent expertise — thin films, catalysis, battery materials, software-driven materials discovery — is genuinely valuable and is not the same thing. One hallmark of the overly ambitious academic entrepreneur is the urge to reinvent every wheel, including the ones that were rolling perfectly well.

IP, in two separate questions. What do you own, and how are you avoiding what other people own? Those are different, and companies routinely answer the first when asked the second. The foundational neodymium-iron-boron composition patents from the early 1980s expired long ago, so the live intellectual property is process — alloy recipes, grain-boundary diffusion, forming and machining methods — and a great deal of it is unpatented know-how that never appears in any portfolio. Which cuts both ways: a thin patent list is not fatal, and an impressive-looking pile of provisional applications is not a moat.

Feedstock. Where does the material come from, step by step backward? Magnet, alloy, metal, oxide, concentrate, mine. Every one of those is a separate industry in a separate place, and the honest answer usually involves China at one or more of them. Buying Chinese metal is not a scandal. Claiming a non-Chinese supply chain while buying Chinese metal is. Our supply chain atlas maps who does what at each step if you want to check the claim against the industry.

Which processes actually happen in the building. A surprising number of "magnet manufacturers" buy blocks, then machine, coat and magnetize them. That is a real business, it is useful, customers need it, and it is not making magnets. Ask directly: do you melt and strip cast? Mill? Press and align? Sinter and heat treat? Machine, coat and magnetize? Or do you buy at some point in that sequence — and if so, from whom? The thirteen steps are worth having in front of you for exactly this conversation.

In practiceOn a plant visit, the equipment is the least informative thing in the room. Anyone can buy a press. What tells you whether a place is running is the evidence of use: scrap bins with real scrap in them, maintenance logs with handwriting, tooling with wear on it, a metrology lab whose instruments have calibration stickers that have been renewed more than once, and operators who argue with each other about a process detail in front of a visitor. A spotless facility with pristine equipment and a very polished tour is a facility that has not made much yet. That is fine at the right stage — as long as everybody says so.

Short versionEvery question on this page should be answerable in a few crisp sentences by somebody who knows their own business. A hedged, jargon-heavy or redirected answer to a simple question is the most reliable warning sign I know, and it needs no technical background at all to notice.

Here are the questions, with what a good answer sounds like. Note how many of the good answers are "no" or "not yet" — early is fine, evasive is not.

• What are your remanence and intrinsic coercivity at 100 °C?Good: two numbers and how they were measured. Concerning: "we're not publishing that at this time," or "we've made some material that gets to about such-and-such at room temperature" — which is an answer to a question you did not ask, delivered in the vaguest available units.
• Can I buy one? What does it cost and when would it arrive?Good: a price, a lead time, a minimum order — or "we're pre-production, first samples next quarter." Concerning: a paragraph about partnerships.
• Where does your metal come from?Good: a country. Not naming the supplier is fair — those relationships are commercially sensitive and everyone protects them — but nobody who buys metal is unsure which country it came from. "We buy on the open market, which today means China" is a fine answer. Concerning: "our supply chain is secure and diversified," or a sudden heavy lean into recycling. Recycled feedstock is real and growing, but it is also the easiest thing to gesture at when the honest answer to the country question is one you would rather not give.
• Which process steps happen in your building?Good: a list, with the bought-in steps named. Concerning: "we're vertically integrated" with no list.
• What is your yield, and on what?Good: a percentage and the step it applies to, or "we don't know yet, we're at coupon scale." Concerning: a yield number with no denominator.
• How much have you made, in kilograms, ever?Good: a number. Concerning: capacity quoted instead of production.
• Who on the team has done this before, and where?Good: names, companies, roles. Concerning: a list of degrees with no operating history in the sentence.
• What can't you do yet?Good: specifics — a temperature, a size, a tolerance, a coating. Concerning: "nothing," which means either a genius or someone who has not tried.

The reason this test works is that all of these facts are known to the person you are asking. They are not research questions. Somebody who has made the material knows the yield and the kilograms the way you know your own phone number, and the inability to produce one crisply is almost never about confidentiality. Which is worth saying plainly: real confidentiality sounds like "I can't tell you that, it's under a customer agreement." That is a complete answer, delivered in one sentence, and it should reassure you rather than worry you.

Short versionAlmost every new magnet material, and essentially every rare-earth-free one, is short on coercivity. Magnetization is the easy half. Resistance to being demagnetized is the hard half, and it is the half that decides whether you have a magnet or a laboratory curiosity.

A permanent magnet needs two things that come from different places. Magnetization — how much field it can produce — comes mostly from iron and cobalt, which are cheap and abundant. Anisotropy — a strong structural preference for pointing one particular way, which is what makes the magnetization hard to knock over — is what the rare earths supply, and it is why they are in there at all.⁵ Nobody adds neodymium because it is a good magnet on its own. They add it because it makes the iron stubborn.

So a new material press release that leads with magnetization or with a theoretical energy product is advertising the half that was never the problem. Read for the other half, and read for two specific gaps:

The gap between anisotropy and coercivity. Real materials never achieve the coercivity their crystal structure theoretically permits. Defects, grain misalignment and thermal activation take most of it, and realized coercivity is commonly a fifth to a third of the theoretical anisotropy field — the computed value is an upper limit, not a target.⁶ A headline built on an anisotropy calculation is therefore a headline about the best imaginable case, before reality takes two-thirds.

The gap between powder and part. A milligram of powder that measures beautifully is not a magnet. Becoming one means aligning the particles, pressing them, densifying them without ruining the microstructure, and doing it in a shape somebody can use — and every one of those steps costs coercivity. This is the step where most promising materials stop, which is why "we have demonstrated the material" and "we have demonstrated a magnet" are separated by years.

It is worth being concrete about the timeline, because this is where headlines mislead most. ARPA-E's REACT program funded 14 rare-earth-alternative magnet projects in 2011, with the explicit goal of getting rare earths out of vehicle and wind generator motors.⁷ Fifteen years later, the furthest-along of those material families — iron nitride — has a first commercial-scale plant under construction, with production expected in 2027.⁸ That is roughly sixteen years from funded laboratory program to first plant, and credit where it is due: that is the FASTEST anyone has moved, not the slowest. Which makes it the right yardstick. If a headline implies the same journey in two years, it is not off by a bit. It is off by an order of magnitude.

None of which means the idea is hopeless — the physics is real, the motivation is real, and the field has been more productive in the last decade than in the three before it. It means the schedule in the headline is the claim to check, not the chemistry.

In practiceA lot of my career has been spent on rotors that spin fast enough to make people uncomfortable, including Formula 1 hybrid systems and development machines approaching a million rpm. Essentially everything interesting about that work is under NDA, so take this as the shape of the problem rather than an account of anybody's solution: in a real machine, the magnet sits at elevated temperature with the field pushed backward at it, thousands of times a minute, for years. A material with a magnificent remanence and a weak coercivity does not gently underperform in that environment. It demagnetizes — once, permanently, and usually at the worst point in the duty cycle. That is why I read the coercivity line first and the energy product second.

What a real new-material announcement contains: a demagnetization curve, at more than one temperature, measured on a bulk sample whose mass and density are stated, with the measurement method named, and an explicit number for intrinsic coercivity. A paper or a release that discusses energy product at length and never mentions coercivity has told you the answer by leaving it out.