Daniel's been asking about graphene. Why one layer of carbon atoms gives you something stronger than steel, more conductive than silver, and nearly invisible all at the same time. And then the part I actually want to get into: twenty years of "about to change everything," so where is it real, where is it still a slide in a pitch deck, and what kept it in the lab for so long.
And it's a good moment to ask, because something has changed in the last eighteen months. Not the physics. The physics has been nailed down since 2004. What's changed is the boring part. Dispersion, standards, price per kilogram. That's where the story actually is.
So let's start with the structure, because almost everything else follows from one fact about how those carbon atoms bond.
Right. Start with what graphene actually is, because people say "single layer of carbon" and nod, and then move on, and that's the whole thing. It's a hexagonal honeycomb lattice, one atom thick. And the cleanest way to think about it is that graphite, the stuff in your pencil, is just stacks and stacks of graphene. You're peeling one sheet off the stack. That's it. The material isn't new. Theorists had been describing it for decades.
And the surprise in 2004 wasn't that it existed in theory. It was that a single atomic layer could be stable enough to exist free-standing at all. Everyone assumed it would curl up or fall apart.
Geim and Novoselov at Manchester pulled it off with what is famously just sticky tape. Scotch tape, repeatedly applied to graphite, thinning the stack down. Nobel Prize in 2010. And Corn, here's the thing that makes this episode worth doing: the properties are real and measured. This was never hype in the sense that the numbers were made up. Tensile strength, conductivity, transparency, all measured, all replicated. The problem was never whether graphene is remarkable.
The problem was making it.
The problem was always making it at quality, at volume, at a cost that makes sense. Lab to fab. That's the whole arc.
So the properties are real and measured. The question is why one atomic arrangement produces all of them at once. Not one at a time. All at once.
So here's the one structural fact. Each carbon atom in graphene bonds to three neighbours. Not four, which is what carbon does in diamond. Three. And those bonds, the sp2 sigma bonds, are at a bond length of about zero point one four two nanometres, and they form that flat honeycomb.
Which leaves the fourth electron.
It doesn't sit in the plane. It sits in a p orbital perpendicular to the plane, sticking straight up and straight down, and those p orbitals on every atom overlap and hybridise into what are called delocalised pi and pi-star bands. Bands, plural, because one is the conduction band and one is the valence band. And Corn, that single arrangement, the sigma framework plus the free-floating pi electrons above and below it, that is the source of almost everything. Strength comes from the sigma bonds. Conduction comes from the pi electrons. Thermal comes from both. Transparency comes from how thin it is and how the light interacts with that pi system.
So it's not four separate wonders. It's one arrangement doing four jobs.
One arrangement doing four jobs. That's the correct way to hold it.
Let me give you the strength numbers, because they're the ones that get quoted and I want to make sure we get them right.
Go.
Young's modulus of roughly one terapascal. Tensile strength of a hundred and thirty gigapascals. And the one I find more intuitive, breaking strength of about forty-two newtons per metre, which is roughly a hundred times stronger than the strongest steel. And then the specific strength number, strength per unit weight, forty-eight thousand kilonewton metres per kilogram, versus a hundred and fifty-four for steel. That's over three hundred times higher on a per-weight basis.
And the reason is twofold. First, carbon-carbon sigma bonds are among the strongest bonds in nature, full stop. And second, because it's a single layer, there's no bulk defect. In a block of steel, failure starts at a grain boundary, a void, an inclusion, some microscopic flaw. There's nowhere in a perfect monolayer for that flaw to live. You're testing the bond, not the material's imperfections.
Which is also the problem, of course. The moment you stack layers, or the moment you've got a grain boundary in your CVD film, you've reintroduced exactly the flaw you were praising the material for not having.
And that's the manufacturing story in one sentence, and we'll get there. But hold it.
Hold it. Electrical next, because this is the one that surprises people.
So in a normal semiconductor, silicon, there's a gap between the valence band, where electrons are stuck, and the conduction band, where they can move. You have to push electrons across that gap. That's your bandgap, one point one electron volts in silicon, and it's why silicon can be switched on and off. It's a feature.
And graphene doesn't have that.
Graphene doesn't have that. The conduction and valence bands meet at what are called the Dirac points. So it's not a semiconductor with a small gap. It's a zero-gap semiconductor. Or a semimetal, depending on who you ask. And the consequence is strange. The charge carriers in graphene behave as if they have no mass. They're described as massless Dirac fermions, with linear energy-momentum dispersion. And the Fermi velocity, the speed at which these carriers move, is about ten to the sixth metres per second. A thousand kilometres a second. A fraction of a percent of the speed of light.
And it's not that they're actually massless. It's that the band structure makes them behave that way.
It's an effective mass of zero. The lattice does it. And that's why the numbers are so absurd. Room-temperature electron mobility above fifteen thousand centimetres squared per volt-second. Resistivity of ten to the minus eight ohm-metres, which is lower than silver. Silver, Corn. The best conductor most people can name.
Thermal, then.
Thermal conductivity of five thousand three hundred watts per metre-kelvin. Highest of any known material. And it's the same story. The pi electrons carry heat, and the lattice itself is so stiff that phonons, the vibrational packets that carry heat in a solid, travel through it with almost no scattering. Two channels, both efficient.
And transparency, which is my favourite, because it's the one where a fundamental physics constant shows up in a materials property.
A monolayer absorbs about two point three percent of visible light. Two point three. So roughly ninety-seven point seven percent passes through. And that two point three percent is not a measured accident. It's pi times the fine-structure constant. Alpha, the number that governs how strongly light couples to charged particles, appears directly in how much light a single layer of carbon eats. That is a beautiful piece of physics.
And the same sheet that lets ninety-seven point seven percent of light through is impermeable to every gas. Including helium.
Including helium, which is the smallest atom there is. The honeycomb is tight enough that nothing gets through. So you have a material you can see through that nothing can pass through. Those two facts should not belong to the same object.
Right, so pull it together. Silicon conducts but is brittle. Steel is strong but heavy. Carbon fibre is strong but insulating. Indium tin oxide is transparent and conductive but fragile and expensive, and it depends on indium, which is not a fun supply chain. Graphene does all of it at once.
And that's the honest reason it keeps coming up. Not that any one property is unprecedented. It's that no prior material combines them. A material that is simultaneously strong, light, conductive, thermally conductive, transparent, and flexible is a material that solves four or five engineering problems at the same time. That's why people keep trying to make it work.
Which brings me to the thing I want to bust before we move on, because it comes up constantly and it matters for the electronics conversation later.
Graphene is not a superconductor.
And it's not a drop-in replacement for silicon in logic chips either.
No, and this is the important one. Because it's a zero-gap semimetal, there's no natural bandgap. And a transistor needs a bandgap. A switch needs an off state. In silicon, the bandgap is what lets you stop the current and get a clean zero. Graphene, in its pristine form, doesn't stop. You can't switch it off the way you switch off a silicon transistor. You have to engineer a gap in, by patterning it into nanoribbons, by stacking it in bilayers, by straining it, and every one of those interventions costs you some of the properties that made you want graphene in the first place.
So it's brilliant at being a sensor, or a high-frequency device, or a photodetector, and it is not the thing that replaces the chip in your laptop.
Correct. And I'd argue the electronics story later in the episode only makes sense if you hold that. The graphene electronics that are actually shipping are not logic. They're sensors, and they're very, very good sensors.
So the science was never the problem. The problem was making it at scale. And that's where the story gets interesting, because it's a story about paste.
It's a story about paste, and about price per kilogram, and about standards documents, and Corn, I know how that sounds, and I promise it's the actual story.
Then let me set up the bottleneck as cleanly as I can. There are two broad manufacturing routes and they trade off against each other. Chemical vapour deposition, CVD, grows graphene on a substrate, typically copper or silicon, and gives you electronics-grade film. High purity, controlled layer count, wafer-scale. At high cost and low volume. Then liquid-phase exfoliation, which is closer to the sticky-tape idea industrialised, blasts graphite apart in a solvent and gives you graphene flakes in bulk. Cheap by comparison, high volume, but the flake quality varies, the layer count varies, and you get defects.
And that's the fork. One route gives you the purity, the other gives you the tonnes. And for a long time, you had to pick. Ben Jensen, who runs 2D Photonics, put it about as well as anyone has. He said the material, when it came out of academia, was hyped to death, but the challenge is going from lab to fab. That's the sentence. The material was never the problem.
So let's take the boring parts one at a time, because they're the answer to Daniel's question about what held it back. Start with dispersion.
Dispersion is the one nobody outside the industry has heard of, and it might be the single biggest practical obstacle. Graphene is hydrophobic. It does not want to sit in water. In a water-based formulation, whether that's a concrete admixture or a coating or a battery slurry, graphene flakes clump together. They agglomerate. And when they agglomerate, you don't get a uniform distribution of nanoparticles through your material, you get clumps and gaps. And HydroGraph, who make this stuff, describe exactly what that produces. Localized points of failure.
So you've got a wonder additive, and it fails because you couldn't stir it in properly.
Which sounds trivial and is not, because the entire value proposition of adding a fraction of a percent of graphene to something is that it distributes uniformly. A clump of graphene is just a piece of graphite with extra steps. And here's the tell that tells you how real this problem was: in the last six months, pre-dispersed graphene pastes have become a product category. That's not a technology story. That's a supply chain admitting the customer can't do the last step.
HydroGraph launched one in July. Fractal Graphene Paste, a twenty percent aqueous dispersion, sold specifically as the thing you add to your mix so you don't have to solve dispersion yourself.
And Levidian just launched one this month. NanoFlow A1-30, a thirty weight percent graphene aqueous dispersion, aimed at concrete and water-based formulations. Thirty percent solids is a high loading, which is the point. You ship as little water as possible and you ship the graphene already separated.
So the naive version of the graphene story is "the science is done, we're just waiting for a factory." The real version is "we spent a decade figuring out how to keep the flakes apart in a bucket."
And that's not a small engineering problem, Corn. That's a colloid science problem. That's surfactants and surface functionalisation and steric stabilisation. It's chemistry at the interface. It just doesn't photograph well.
Next bottleneck. Cost.
Cost is improving fast. Industrial-grade graphene nanoplatelets are now trading at fifty to seventy-five dollars a kilogram. And that number only means something in comparison, so here it is: it's approaching parity with carbon black, which is the commodity filler that goes into tyres and plastics and inks at enormous volume. Carbon black is cheap. If graphene is at fifty to seventy-five a kilogram, you're no longer in the world of exotic materials. You're in the world of additives that a procurement department is willing to consider.
But the honest caveat.
The honest caveat is that per unit of performance, graphene is still three to five times more expensive than graphite in most bulk applications. So you're not replacing graphite on price. You're replacing it on the argument that a tenth of the loading does more work, which means you have to be able to prove that, which means you have to be able to specify what you bought.
Which is the third bottleneck, and it's the one I find most damning. Standards.
Buyers still get incompatible layer count and surface area claims from different suppliers. Iso and ASTM specifications exist. Adoption is inconsistent. And the consequence is a trust problem dressed up as a technical one. If I'm a formulator, and I can't be confident that the graphene I buy in March behaves like the graphene I bought in January, I can't qualify it into a product. Full stop. Qualification is the gate. Nothing goes into a commercial product without it, and you can't qualify a moving target.
So a supplier can be making perfectly good material and still fail commercially, because the buyer can't write a specification around it.
That's the least glamorous failure mode in the whole story, and it might have cost more years than any single technical problem. You can solve dispersion with chemistry. You can solve cost with scale. You can't solve "everyone's measuring differently" without the industry agreeing on something, and industries are bad at agreeing on things.
Here's the commercial insight, and I think it's the real answer to what Daniel's asking. Graphene succeeds where it's a low-dose additive.
Say more, because I think that's the reframe and I want to get it right.
The original pitch was replacement. Graphene replaces silicon, graphene replaces indium tin oxide, graphene replaces lithium-ion, graphene replaces steel. And every one of those pitches failed, because in every one of those cases, graphene has to beat an incumbent that is cheaper, better understood, and already qualified, on the incumbent's own terms.
The additive pitch is different. You're not replacing anything. You're adding a fraction of a percent of something to a material that already works, to nudge one property past its baseline. The material still does its job. The graphene is the seasoning.
That's the version that's actually shipping. Not because it's more exciting. Because it doesn't require anyone to rip out a supply chain.
There's a phrase I keep coming back to from the commercialization tracking work. Graphene succeeds commercially where it acts as a low-dose performance additive, not where it has to do the heavy lifting alone. That's the thesis. And once you see it, you can't unsee it in the shipping data.
Let's do the shipping data. Where is it actually real? And I'll say up front, the answer surprised me.
Go on.
Concrete. Cement.
Cement, yes. And it is the largest near-term volume market, and the reason is only partly about graphene.
Explain that, because that's the interesting part.
Cement production is about eight percent of global CO2 emissions. Eight percent. That's not a rounding error, that's a bigger share than aviation and shipping combined in most accounting. And there is enormous regulatory and commercial pressure to cut it. So you have a giant industry, under pressure, looking for anything that reduces emissions, and willing to pay for it. And graphene-enhanced concrete appears to do that in two ways. It can reduce the amount of cement you need for a given strength, and it firms up the argument for replacing some of the clinker.
Concrete is graphene's unlikely hero. The least glamorous application in the entire list, and it may be the biggest.
Concretene, out of Manchester, claims a twenty to thirty percent cut in embodied carbon. First Graphene ran a six hundred tonne trial with Breedon and Morgan Sindall and FP McCann, targeting roughly sixteen percent CO2 reduction through clinker replacement. And their chief executive, Michael Bell, said the cement and concrete sector has one of the highest volume potentials for graphene integration into product. Which is a very careful sentence from a chief executive, and it's also true.
The lovely detail is that the driver isn't graphene's elegance. It's a regulatory and emissions incentive that happens to be enormous. Graphene is the beneficiary of a problem it didn't create.
Which is how a lot of materials actually get adopted. Because something else made the incumbent expensive.
Second application. Thermal interface materials.
The quiet winner.
I like this one because it's the least sexy and the most obviously real. A thermal interface material is the paste or pad between a chip and its heatsink. It's the thing that fills the microscopic air gaps, because air is a terrible conductor and any gap at all ruins your thermal path. And graphene's thermal conductivity, five thousand three hundred watts per metre-kelvin, is absurdly high. So you load a polymer with graphene and you get a paste that conducts heat far better than a plain polymer.
It competes directly with silver-loaded epoxies. Silver is expensive, and the supply chain is what it is. Graphene is not silver. So you get comparable thermal performance at lower cost, and in data centres and EV power modules, thermal management is the constraint that's actually limiting what you can build.
Which is why I'd call it the quiet winning application. Nobody's writing a breathless article about thermal paste. But it's shipping, it's qualified, and the demand is growing because everything is getting hotter.
Third. Automotive composites.
Black Swan Graphene's sheet moulding compound plant in Montreal. And NanoXplore, which is the largest Western producer by volume, up to four thousand tonnes a year, supplying graphene-loaded thermoplastics. So you're adding graphene to a plastic or a composite to improve stiffness, or reduce weight, or add a bit of conductivity for electrostatic discharge, or all three.
Again, it's the additive model. Nobody's building a car out of graphene. They're building a car part out of a composite that has a fraction of a percent of graphene in it, and the part is better than the one it replaced by enough to justify the switch.
Fourth. Electronics, and here I want to be careful, because this is where hype and reality are most tangled.
Paragraf is the standout, and I think it's fair to say they're the most concrete proof that graphene electronics are real rather than theoretical. They make GFETs, graphene field effect transistors, on six-inch silicon wafers, using CVD. Wafer-scale, published specifications, purchasable. They produced their first six-inch graphene wafer in December, closed a fifty-five million dollar Series C back in August of 2025, introduced the PMF2000 GFET in May, and then the PMF2002 platform and an Enterprise Reader in August. That's a product line, not a paper.
A GFET, to be clear, is not a logic chip.
Correct, and this is where the semimetal thing comes back. You can't switch a pristine graphene transistor off cleanly. What you can do is exploit graphene's extreme sensitivity to what's near it. A GFET is an outstanding sensor. It responds to magnetic fields, to charge, to biological molecules landing on its surface, because the graphene is a single atomic layer and anything touching it perturbs the electronic structure measurably.
The graphene electronics that are shipping are sensors with published specs on wafers you can buy.
Which is a much smaller claim than "graphene replaces silicon." It's also a much more real one.
Now let's do the debunking, because Daniel asked specifically where it's still hype, and there are two that matter.
Batteries is the big one.
No standalone graphene battery has displaced lithium-ion, and the numbers aren't close. GMG's best aluminium-ion cells reached forty-nine watt-hours per kilogram in April, up from twenty-six. Lithium-ion is two hundred and fifty to three hundred watt-hours per kilogram. That's a factor of five to six. And GMG's cells are at technology readiness level four, with small production pushed to 2027.
The honest version of the battery story is: graphene as an electrode additive, improving the conductivity of the electrode, extending cycle life, improving charge rate. That's real. Graphene as a replacement chemistry is not real, and anyone telling you it's imminent is selling something.
The other hype area, and I want to be precise here.
Consumer devices.
There is no verified consumer smartphone or display shipping at volume with graphene as a core component. Not that I can find, and not that anyone's been able to show me. The electronics story is real but concentrated in sensors and R and D. Somewhere in the R and D pipeline there may well be the thing that changes that. There is not a phone on a shelf today where graphene is the reason it works.
That distinction matters, because the hype cycle trained a whole generation of people to expect a graphene phone and a graphene battery and a graphene everything by 2018. What they got was graphene in concrete, thermal paste, car parts, and sensors. Which is a real industry. It's just a different one.
Which brings up the thing I want to raise before we close the commercial section, because I think it's the biggest strategic risk and it barely gets discussed.
China.
China operates a three hundred tonne facility and is scaling aggressively. And the risk here is very specific. Western graphene producers are boutique relative to that. And if Chinese capacity scales the way Chinese capacity has scaled in other materials, you get a solar-panel-style price collapse, where the technology wins globally and the Western companies that developed it go bankrupt.
Which is exactly what happened in solar. The West invented the modern photovoltaic, invested heavily, and then watched Chinese manufacturing drive the price down by an order of magnitude, take the market, and leave most of the Western manufacturers out of business. The technology won. The inventors didn't.
The EU put one point two billion euros into the Graphene Flagship since 2013, which is a serious commitment, and the question is whether that money built capacity that can survive a price collapse, or whether it funded research that will be commercialised by someone else's factory.
I don't know the answer to that. I don't. The solar analogy is the one that scares me, and I don't have a confident read on whether the additive market is different because the customers are stickier, or whether it's the same story with a different material.
Then let's give the market numbers, because they frame the scale question. About one point seven billion dollars in 2025, projected to somewhere between three point six and nine point three billion by 2030. That's a twenty-four to forty-one percent CAGR, depending on who you ask. And the application split in 2024 was composites at thirty-six percent, electronics at twenty-four, energy storage at eighteen, coatings at twelve, sensors at eight.
There's a huge range in that projection, which itself tells you something. Nobody has a confident model. When the forecast for a market five years out ranges from three point six billion to nine point three billion, that's not a forecast, that's a bracket around a guess.
Here's where I've landed. The twenty-year gap between lab and fab wasn't a science problem. It was a decade of unglamorous process engineering, dispersion and standards and pricing, and it got almost no attention because it doesn't photograph well.
The question was never whether graphene's properties justified investment. They clearly did, they were measured from day one. The question was always whether it could be manufactured at sufficient quality, volume, and competitive cost. And the answer, as of right now, is yes for some applications and not yet for others.
Which is a much less quotable sentence than "graphene changes everything" and a much more useful one.
Corn, the last thing I want to say before we move on is that the one thing that struck me doing this is that the pi bonds are doing everything. Strength, conduction, thermal, the optical absorption, all of it comes from that one orbital arrangement. A whole industry is being built on the fourth electron of carbon.
The most overworked electron in materials science.
The manufacturing story is unglamorous, but it's the real one. And speaking of unglamorous quality control, I want to say one thing about how you actually tell whether a batch of graphene is any good, because I was looking into it and there's a whole world of layer count measurement and Raman spectroscopy and atomic force microscopy that determines whether what you bought is what you think you bought.
You could taste it.
I'm sorry, Hilbert, go on.
What I said. I agree with you on the manufacturing side, and the standards stuff especially, because I've lived through it. But you've both missed the thing. The problem wasn't that nobody could agree on a standard. The problem was that nobody could agree on what one layer meant. That's why the standard never stuck. Half the trade was buying three layers and calling it graphene.
You know this because.
Because I used to do quality assurance for a materials supplier in the mid eighties. And I could taste it.
Taste it.
Metallic, with a hint of pencil. And before you say anything, it was a real skill, and I was certified in it. Internal programme. There were eleven of us started. I was the only one who passed.
How does one get certified in tasting graphene, Hilbert.
It was a blind panel. Fifty samples, graded in pairs, and you'd write down the layer count and hand it in. Man called Derek Ruddock ran it, up in the office above the loading bay, and he kept the answers in a biscuit tin. I scored forty-four out of fifty. Derek got forty-six, but he'd written half the samples himself, so it doesn't count.
What happened to the programme.
Discontinued. After me. There was only ever the one graduate and the economics stopped making sense. They printed my certificate on paper that may or may not have contained graphene. I've still got it. It's laminated, in my wallet. The lamination was done by the same company that made the graphene, and I've always thought that certificate is probably the purest graphene I ever tasted.
Of course you have it.
I rejected a whole shipment once. The whole pallet. Because it tasted like three layers. Which it was. And I got a letter about it. I've still got the letter.
The standard problem was real, then.
The standard problem was real. The tasting was real. They abandoned it too soon. There were eleven of us, and then there was one, and then there was none, and I don't want to talk about it further.
Where does that leave us? Twenty years in, the revolution is boring, and that might be the best sign yet. The graphene that's actually shipping is concrete and thermal paste and car parts and sensors. Not a phone, not a battery, not a chip.
I think that's the thing worth holding. Graphene's story is a reminder that the hard part of a wonder material is never the wonder. It's the manufacturing, the qualification, and the supply chain. The science was done in 2004. The industry is being built now, by people solving problems about stirring and paperwork.
Whether that industry exists in the West in ten years, or whether it's a repeat of solar, is open. I'd watch the European and American capacity announcements more than the research papers at this point. The physics isn't the variable anymore.
Twenty years of "five years away" is finally closing, and it's closing because dispersion pastes got good and prices approached carbon black. Not because anyone had a breakthrough.
Which is the least cinematic ending to a wonder-material story, and the most likely to actually happen.
Thanks as always to our producer, Hilbert Flumingtop, who has a certificate in his wallet and a letter he won't show us.
This has been My Weird Prompts. If you enjoyed the episode, leave us a review on your podcast platform of choice. It helps other listeners find the show.
If you've got a question about something that's been "five years away" for twenty, send us your own prompt on Telegram at t dot me slash MWP listener bot.
We'll be back soon.
