A follow-up to Donut Lab: a study in deception in venture-backed technology, published 25 March 2026. If you haven’t read it already, go and have a look (please do, it’ll make what’s below make a lot more sense) then come back here.
There is a German company called CT Coatings whose patent portfolio covers screen-printed paving slabs, laminated menus, and warning triangles. It is also, it turns out, the origin of the world’s first production-ready solid-state battery. Only kinda not really.
A quick review for those of you who haven’t followed the link at the top (I’m judging you). In March I looked at a company called Donut Lab’s claims of a novel 400 Wh/kg solid-state, clean materials battery, charging in five minutes, lasting 100,000 cycles and shipping in Q1. And I concluded that the whole thing failed the sniff test.
Donut Lab was Marko Lehtimäki’s third “world’s first” company in fifteen years, after AppGyver, which SAP acquired and then quietly killed once he’d left, and Asilab, which claimed the world’s first true artificial intelligence and has produced no demonstration, no paper and no patent. There was a circular corporate structure: Donut Lab is wholly owned by his brother’s motorcycle company, which was also its only customer, with VTT, an external lab, commissioned to test fast charging and temperature tolerance while conspicuously not measuring energy density or cycle life; and with PwC (very unusually) unable to complete Verge’s 2024 audit for want of almost any financial records at all. I talked about the way that venture-backed deep tech rewards the appearance of breakthrough over the substance of it, because the gap between an extraordinary claim and its disproof is the window in which value gets extracted, and nobody in the loop has much incentive to close it early. I gave three possible endings, put the odds of the battery being real below 5%, and said I’d update when the Q1 deadline passed.
It passed, and so did four more months (I got busy); then in the interval an engineer named Dr Ryan Hughes went and did the due diligence that nobody investing in the company had bothered to do. It’s a very deep dive. His investigation, published on his YouTube channel, is excellent and you should watch it. That investigation changes some of what I was thinking, and means that some parts of my analysis of what the hell was going on have turned out better than the rest.
On the last day of Q1, Donut Lab announced its first production motorcycle. For about a day this smelled like delivery.
It wasn’t. (Shocked?) Hughes surfaced an internal video to reservation holders explaining that the first bikes were going to Verge’s own fleet, to refine the manufacturing process before anything went to a customer.
A vehicle built to refine your manufacturing process is a pre-production vehicle. That is what those words mean.
CEO Marko Lehtimäki then told Finnish media that the 400 Wh/kg cells were not in the bikes, and that the cell VTT had tested, the one Donut’s entire “I Donut Believe” marketing campaign (including merch!) was constructed around, “is not even the cell that’s going to be shipped to customers.”
The independent validation exercise tested a cell the company was not ever planning to ship. I have read that sentence several times now and it is still not making an awful lot of sense to me.
Hughes brought in more than twenty independent battery specialists, among them Julian Zahnow at Fraunhofer, Dr Joachim Sann at Justus-Liebig, Tom Bötticher at Litona, and Dr Juho Heiska at Seinäjoki. Every one of them came to the same place (and I called the chemistry absolutely right in my March post). The cell is lithium-ion.
Two pieces of evidence come from VTT’s own published data, which Donut Lab commissioned and paid for. The first is the voltage curve. The cell sits at 3.7 to 3.8V at half charge, which is where high-nickel NCM lithium-ion cells live. Sodium-ion, which is the only chemistry that fits the claims that Donut Lab were making, does not get meaningfully past 3.5V.
The second finding is the sort of thing that makes me very happy about physics. When a cell charges, ions push into the anode and the anode swells; and the shape of that swelling curve is a signature of the anode material. Graphite produces a characteristic bump between roughly 50% and 70% state of charge, caused by ions reordering themselves between graphite’s layers. The Donut Lab cell demonstrates the bump. Sodium ions are too large to fit between graphite layers at all, so the kink does not merely suggest lithium: it absolutely requires it.
Hughes describes having both a “noisy fingerprint and a photograph of the suspect’s face”, which is fair. The measured energy density lands around 298 Wh/kg: a decent lithium-ion cell, falling quarter short of the claim. And not solid-state, and not made of “clean materials” at all.
In March I wasted a lot of your time wondering whether this was a supercapacitor dressed as a battery. (I have a bit of a thing about supercaps: take me out for a drink sometime and I will bore you.) It isn’t. It’s a battery, and a very ordinary one. Being wrong in a boring way is the most common way to be wrong about hardware.
Back to the paving slabs.
CT Coatings promised Nordic Nano and Donut Lab a screen-printed sodium-ion solid-state cell. What turned up was a lithium-ion pouch cell. Zahnow, who met CT Coatings’ representatives, explains in the video that he formed the view that they did not understand how a battery works, his illustration being that they explained their product contained no rare earths and therefore no lithium. Lithium is not a rare earth. This makes no sense.
The arrangement ran: CT Coatings supplies technology, Nordic Nano manufactures, Donut Lab commercialises. Nordic Nano has reportedly never manufactured a battery cell. Leaked emails show Donut Lab asking CT Coatings when the promised proof of specification would arrive. It never did.
So the chain of custody for a revolutionary battery runs from a firm that patents laminated wine lists, through a manufacturer that has never manufactured, to a commercialiser that didn’t obtain evidence for the thing it was selling. It’s weird all the way down to the bottom, which is stationery, not batteries.
I argued that the gap between extraordinary claim and irrefutable disproof is the window in which value gets extracted, and that for batteries that gap should run six to eighteen months. I sketched a timeline: independent teardowns confirming ordinary lithium-ion at month twelve, cycle-life results at month fifteen, lawsuits at month eighteen.
I got that wrong. The chemistry was settled in month five, and not by an auditor or by a regulator; and also not by an investor who had written a cheque. It was a smart and somewhat offended power and propulsion engineer with a YouTube channel, a spreadsheet, and the email addresses of twenty people who know what a dQ/dV curve looks like.
This is really the only thing I find particularly cheering in the whole story: the verification capacity in this industry is real, it is fast, and it is distributed, across people who will respond to emails for free because a lot of us feel kind of insulted when we’re presented with bullshit. What it is not is attached to the money. Nobody paid for this. A Big Four diligence engagement would have billed six figures, taken twelve weeks, and, in the versions I have sat through, due diligence may not have caught it, because DD would have approached VTT’s report as validation rather than taking a step back and thinking about what VTT had not been asked to measure.
But the window closing early only helps if the money is still outside it. Here, it wasn’t.
My whole March framing was about venture capital. (I suppose this is probably defensible insofar as venture capital is what I do.) Asymmetric risk-reward, VCs needing moonshot narratives for LPs, acquirers needing innovation optics, everyone declining to look too hard because everyone in the room got something. As it turns out, the venture capitalists were never in the room: and the investment model was the discovery Hughes made that shocked me the most and that’s left the worst taste in my mouth, because it’s nakedly exploitative and pretty nasty.
Donut Lab has over 1,300 shareholders. More than 900 of them hold fifty shares or fewer, which implies individual positions somewhere between a few thousand and a couple of tens of thousands of euros: angel investors putting in their own personal money. A great many arrived through a crowdfunding round for Verge Motorcycles on Finland’s Springvest platform in 2023. When Verge was de-merged and Donut Lab spun out underneath it, the valuation rose and meant that the loss-making motorcycle company became the parent of a half-billion-euro portfolio, on the strength of the untested battery. Then after all the CES press, the valuation shot up to $1.25 billion. An investor letter from Lehtimäki offered “up to 10x in just 12 to 18 months” and encouraged existing holders that it was not too late to add. This looks like absolutely fabulous news if you’re an angel investor: it really looks as if you’ve found the golden goose; and it served as a really great incentive for those angels to put in more money for the company owners to extract.
Now go back and look at the self-validated due diligence.
Back in March I looked at Donut Lab running its own technical DD, rather than commissioning independent validation, as a corner cut by people out of their depth. Hughes reads it as an original and considered scheme, and I think he’s right. Any hardware fund would have insisted on independent cell testing before wiring money. (It’d take a fortnight, and it’s what we would do here at Negroni.) A non-technical retail investor with six thousand pounds on a crowdfunding platform is not in a position to commission VTT, is not able read a dQ/dV curve, and has no way of knowing that the missing cycle-life test is the thing in the report they should be paying attention to. That investor is perfectly placed to be exploited.
Everyone reaches for Theranos when looking at stories like this, and when you look at the science the comparison stands up pretty well; but on the money it flatters Donut. Theranos raised from people who could have checked and chose not to. Donut Lab was structured so the people putting money in were incapable of checking. There’s an ethical vacuum here.
I also said criminal charges were extremely unlikely, on the grounds that no investor has the appetite for the cost and the reputational mess.
Finnish financial and criminal authorities are reportedly investigating. Lauri Peltola, formerly chief commercial officer at Nordic Nano, has filed reports with Finland’s Financial Supervisory Authority, the chancellor of justice, and Helsinki police. Donut Lab and Nordic Nano jointly deny committing any crime or misleading investors, and stand by their announcements. And for ass-covering: nothing has been proven and everyone here is owed the presumption of innocence.
Donut Lab’s own response to the Ziroth video makes two points that deserve airing if we are being scrupulous here. The first is that Hughes has a commercial relationship with CATL, a competitor. But that while that’s maybe a conflict, it does not affect the data: the voltage curve and the expansion kink come from VTT’s own results, which Donut Lab commissioned. The second is that Lauri Peltola, who was interviewed, was not on the battery development team and lacks current knowledge of the work. Nordic Nano has filed its own police report about his conduct. Peltola, for his part, has since published the investor letters he says underpin his complaints, and has filed further reports against CT-Coating and with OLAF, the European anti-fraud office.
So while my reasoning was wrong, I don’t think the conclusion was. Institutional investors don’t press charges because they run on portfolio theory and have a reputation to manage. Nine hundred people who each put in the price of a decent second-hand car have neither. They have nothing to protect and nothing to lose by complaining, and complaining is free. The financing structure that made this possible is the same structure that makes it prosecutable.
We have built a two-tier verification system and called the cheap tier “democratisation”. It is definitely a way to increase access. Specifically, though, it’s access to the deals that could not survive the expensive tier; and that sucks.
So: test the claim that the valuation is based on, and don’t bother about the easy claims, which may be there to distract you. Treat commissioned validation as marketing until proven otherwise; read what the report was not asked to measure; and when a fella with a YouTube channel finds in five months what your diligence process would have missed in twelve, ask what exactly you are paying for.
This piece draws on publicly available information, including the Ziroth investigation and contemporaneous reporting linked throughout. Donut Lab and Nordic Nano deny wrongdoing; the investigations are ongoing and nothing here should be read as a finding of fact against any individual or company. If you have additional information about any of these companies, please get in touch.