When Finnish startup Donut Lab announced at CES 2026 that it had created the world’s first production-ready solid-state battery, hardware folks like us got very excited, but on looking closer started to wonder what the heck was going on. The claimed specifications were extraordinary: 400 Wh/kg energy density, 5-minute charging, 100,000-cycle lifespan, and availability in Q1 2026, just months from a prototype being unveiled. If true, Donut Lab had solved in stealth and in a matter of months what Toyota, CATL, Samsung and QuantumScape have spent billions failing to achieve.
Here at Negroni, we were very intrigued and also very sceptical. The announcement of a “world’s first” battery technology made with “abundant, green materials found everywhere” sounded too good to be true. On closer inspection, we’re now sure it is too good to be true: and I wanted to share some of our thinking around this, think about what drives a business to make extraordinary claims it can’t back up, and to talk about how blisteringly important it is to do proper technical and financial due diligence before we seriously engage with a company.
Three companies
A closer look at CEO Marko Lehtimäki’s CV reveals something uncomfortable: Donut Lab isn’t an isolated moonshot. It’s the third “world’s first” revolutionary technology company he’s launched in a short space of time. Each company follows a similar playbook, and each raises serious questions about whether the business model is true innovation or something a bit different.
AppGyver (2011-2021)
Marko founded AppGyver in 2011 with a bold promise: “the world’s first professional no-code platform”. The company raised $11 million from reputable VCs including Initial Capital, Karma Ventures, and OpenOcean. In February 2021, SAP acquired the company for an undisclosed sum, calling it “a pioneer in no-code development platforms.” And in this story, this company is a bit of a standout insofar as it did actually do what it promised to do: for a while, especially before the acquisition, it really did provide a decent (if not world-changing or world-first) no-code tool for building React Native applications.
Marko became “Director & Team Lead, SAP AppGyver” and stayed exactly 11 months before departing in January 2022.
What happened next was catastrophic. By 2024, the AppGyver community was furious and desperate, describing the platform as “discontinued,” a “dying platform” with “minimal updates since the SAP acquisition.” Competitors now openly market themselves as AppGyver alternatives because SAP essentially killed the product. The free Community Edition remained nominally alive but feature-frozen, with users describing it as little more than a stepping stone to SAP’s own platform.
The acquisition looked like a success, but the product itself collapsed in very short order, by which time the CEO was long gone.
Asilab (2025)
In May 2025, Marko announced Asilab, a new company he chairs, claiming to have developed “the world’s first true artificial intelligence.” (This “world’s first” thing seems to be a bit of a trope.) The technology, called “Asinoid,” purportedly differs fundamentally from large language models, in much the same way as the Donut battery purportedly differs fundamentally from known battery technologies:
“Current LLMs are like transistors, and we are building the first real computer… Asinoids are today the world’s only AI with their own life, thoughts, continuous evolution and synthetic neuroplasticity.”
The company was founded by “members of the successful AppGyver company,” which is to say the same team. Asilab is “primarily a research and development organisation” seeking corporate partners and employees to commercialise the technology, starting with B2B applications.
There exist no product demonstrations, no peer-reviewed papers, no patents and no technical details beyond marketing copy. But the marketing copy’s great for headlines and investors with AI stars in their eyes: brain-like cognition as described here would make the entire AI industry obsolete.
Donut Lab (2026)
Seven months after launching Asilab, Marko was back with Donut Lab’s solid-state battery announcement. Spun out of Verge Motorcycles (run by his brother Tuomo), Donut Lab claims to have achieved what the entire battery industry has failed to deliver: production-ready, solid-state batteries, available now.
Red flags
The red flags appeared immediately (and you may well have read about them elsewhere; there’s a lot of scepticism out there about this product):
Donut Lab is 100% owned by Verge Motorcycles, which is owned by Marko’s brother. The only “customer” is Verge. Nordic Nano, a company that received Donut Lab investment, has Marko on its board (he also owns a stake in the company), and had previously published specifications for a “bipolar electrostatic capacitor” with identical specs to Donut Lab’s battery (400 Wh/kg, 50k-100k cycles, with a fireproof construction), raising questions about whether this thing is actually a supercapacitor being marketed as a solid-state battery.
When questioned, Marko explicitly denied Nordic Nano was making the battery. But the identical specs raise an obvious question: is Donut Lab’s “solid-state battery” actually an advanced supercapacitor being marketed with battery expectations?
This matters because supercapacitors and batteries have different characteristics.
Supercapacitors have a 100k+ cycle life, can offer that short 5-minute charging the marketing copy mentions, and have a high power density. But they have linear voltage discharge (power drops as it drains), higher self-discharge, and are effectively unusable for the last 20-30% of their capacity.
Batteries have a stable voltage discharge, low self-discharge, and all of their capacity is usable. But they have limited cycles (1,000-5,000), offer slower charging, and suffer from gradual degradation.
If Donut Lab achieved 400 Wh/kg in a supercapacitor, that would still be a really impressive breakthrough. But marketing it as a battery sets false expectations. Verge’s claimed “370-mile range” in a supercap might mean 300 miles at full power, dropping to 150 miles at 60% charge, then becoming unusable. This is very different from a battery that delivers consistent performance until empty.
If the product is a supercapacitor, the descriptions would be technically accurate (it does store 400 Wh/kg) while functionally deceptive (it doesn’t behave like a battery). The VTT tests notably didn’t measure how voltage drops during discharge, a test that would immediately reveal whether this is battery or capacitor behaviour.
Nordic Nano later removed the capacitor report from their website.
Donut Lab commissioned VTT Technical Research Centre of Finland to conduct “independent” tests. Over several weeks, VTT verified fast charging (4.5 minutes to 80%), high-temperature operation (maintained capacity at 100°C), and minimal self-discharge. These tests confirm the battery works reasonably well, although there are some issues with those tests: the battery hit 90°C and test had to be paused/restarted with better thermal contact; in high-temp tests the vacuum seal gave way; and sure, the self-discharge rates were good, but no more performant than a standard li-ion battery. (I’ll give them one point there, though: the self-discharge rates, with 97.7% charge retention after 10 days idle, does suggest that this thing isn’t actually a supercap in disguise.)
But there were several points that VTT conspicuously didn’t test: energy density (400 Wh/kg) and cycle life (100,000 cycles). And these were the two most extraordinary claims made by the company, both trivial to verify. VTT never weighed the cells. No cycle-testing was performed. The chemistry remains undisclosed. Battery researchers analysing the voltage curves say results are consistent with currently-available Li-NMC battery cells, suggesting advanced lithium-ion, not revolutionary solid-state.
Svolt Energy chairman Yang Hongxin was blunt: “That battery doesn’t exist in the world. All parameters are contradictory. Any technician with basic knowledge would recognise it as a scam.” Here at Negroni we are not battery specialists, but we are pretty darn technically proficient, and the tech as described didn’t pass our sniff test even back in January (which is when I started tracking this stuff).
The original claim was Q1 2026 deliveries (by March 31). In January this year at the product announcement, no real battery was available to view: the object presented at CES was a 3d-printed “prototype” of the casing. As of late March 2026, no confirmed customer deliveries exist, despite orders taken at $30,000 per motorcycle with 10-20% “reservation fees” (reservation fees are importantly very different from a deposit, which you can have returned to you when the product doesn’t eventuate); and the company is now saying that the timelines have changed, with a late March date (we’re publishing this on March 25 and I will eat my hat if delivery actually happens in the next few days) for “selected early customers” in Europe (Finland/Estonia); then a Q4 2026 date for new US orders. Even so, Verge has only ~350 bikes planned for the whole of 2026, and Marko is now saying to expect deliveries “after the end of March, likely April“.
Verge Motorcycles, which owns Donut Lab, reported to PwC a €10 million loss on €1 million revenue in 2024. PwC couldn’t complete the audit because the company lacked “almost all financial information on inventory management, sales receipts, research expenditure, property values.” The company has delivered roughly 100 motorcycles since 2018, all while claiming “gigawatt-hour production capacity.” Which kind of speaks for itself.
So what’s going on here? All three companies follow a similar formula. Make “world’s first” claims in hot technology sectors (no-code, AGI, solid-state batteries). Provide absolutely minimal technical disclosure while emphasising secrecy for “IP protection”. Commission friendly testing that validates easy claims while avoiding hard ones. Seek major corporate partnerships or acquisition from established players. Extract value through investment rounds, reservation fees, partnership fees and government grants. Exit before long-term verification becomes unavoidable. And critically, face no consequences, then repeat with next company.
Why would someone do this? I’m not really very interested here in the underlying psychology, and whether Marko believes his own claims. But I am interested in whether belief matters when the incentive structure rewards the appearance of breakthrough over the substance of it, and what that means for startups and VCs. Let’s look at the economics.
Asymmetric risk-reward
Traditional fraud is zero-sum. (And, for legal ass-covering, I should point out now that we do not know Donut Labs’ battery claims to be fraudulent: we’re just saying the sniff-test, based on the public sources linked throughout this article, is failing right now.) A traditional fraudster steals money, the victim loses it, and if the fraudster is caught, consequences follow. But venture-backed tech fraud operates in a grey zone where multiple parties benefit from suspension of disbelief.
A founder raises capital based on extraordinary claims, and receives a reward in the form of salary and equity, along with a reputation boost while the company still looks exciting. If acquisition happens (AppGyver), the founder exits with a nice big bag of money before the product fails. If the company then fails, the founder can blame technical challenges, market conditions, or mismanagement by the acquirer – they’re not present, their hands are clean. They can then move to their next company with “successful exit” on their resume, which is a very useful piece of currency when attracting investors to your next thing. Their legal risk is low: optimism isn’t fraud, and “a breakthrough is imminent” is a piece of language that we meet a lot in VC, which is unprovable until a deadline or two passes.
The VCs benefit here too. They (OK, we) need moonshot narratives for LP appeal (“we invest in world-changing technology!”), and there is value in being able to boast that you’ve got the brains to spot an investment in the “world’s first X” during the fundraising cycle. If the company fails, it can be marked as a binary-outcome tech bet, which is expected in portfolio theory. And if the company exits, the VC can claim a successful return regardless of post-acquisition fate.
There is a limited incentive to do proper, thorough due diligence (and one of the things that differentiates us at Negroni from other VCs at our scale is that we spend a huge amount of our time on technical and financial DD, which we believe benefits us and our portfolio companies): for many funds, technical verification is expensive, outside the expertise of the manager, and from the 60,000ft view, kills deals.
The acquirer (in AppGyver’s case) needs good optics on innovation and technology to show shareholders. The acquisition team gets credit for making a “forward-thinking deal”, and integration and failure are likely to happen on a different team’s watch, often years later. It’s easy enough to blame founder departure for product problems, and sunk cost is ratioed against the company’s overall R&D budget.
The media benefits from this too. “World’s first” is an great headline, and the great headlines are the ones we click on. Technical scepticism requires expertise, and looks very boring compared to a breakthrough story, especially when it’s such unalloyed good news: cheap power, green materials, which are naturally “abundant” and without a conflict story! Then, by the time it’s clear that things are not as they have been represented, the story is old and a new “world’s first” has replaced it. No journalist is ever going to be held accountable for breathless early coverage.
What about early customers? In Donut Lab’s case, there are at least 350 people paying reservation fees to be “first” with a revolutionary technology. They’re motivated to believe. (Some of this is the sunk cost fallacy, some of it’s the glow of Being First that early adopters enjoy – no shade, I’m often one of them). By the time the product fails, these fees will be non-refundable (they’re not deposits).
Timelines
The crucial insight is this: the gap between extraordinary claim and irrefutable disproof is exactly the window where value extraction occurs, and for batteries, this gap is 6-18 months.
Here’s what I think feels like a probable timeline.
Month 0: Announce “production ready” with Q1 delivery date
Month 1-3: Take deposits, sign MOUs, raise investment round
Month 4: Miss Q1 deadline, push to Q2 “for final optimisations”
Month 6: Deliver initial units to friendly customers (brother’s company: I don’t think it gets much friendlier than that)
Month 7-9: Customers begin using batteries, initial problems dismissed as “teething issues”
Month 12: Independent teardowns confirm something unexpected: something like discovering it’s advanced Li-ion, not solid-state
Month 15: Cycle life testing shows, say, 5,000 cycles, not 100,000
Month 18: Lawsuits begin, but founder has moved to next company
During months 0-6, the company raises about €10-15 million from €1-2M in customer deposits (350 orders × €27k × 15% deposit/reservation fee); €3-5M in equity investment at inflated valuation; €2-3M in partnership/evaluation fees from OEMs; €1-2M in government grants (the Business Finland program); and, say, €2-3M in licensing fees for a “universal EV platform”.
By month 12, when the truth emerges, the founder has paid themselves a salary of perhaps €500k (I am spitballing), and extracted €2-3M in secondary shares (still spitballing, feels about right). They’ve built relationships with the next set of investors and established a “track record” of launching bold ventures.
Downside risk
So what’s the downside risk?
Reputational damage is a risk, but “failed fast” is celebrated in our (broken?) startup culture. There is potential civil liability. But in a company that’s structured correctly, which, given the careful complexity of the family and business relationships, I’ll bet this is, that liability will be limited to company assets. And finally, criminal fraud charges (extremely rare and pretty darn unlikely – no investor in the world has the appetite for getting involved in something like that with all its attendant costs and PR burden; it also requires proof of intent, and the crossing of regulatory lines).
In AppGyver’s case, we see this exact pattern: Marko extracted value through acquisition, left within a year, and the product failure happened on SAP’s watch. He faced no consequences. His reputation as a “successful tech entrepreneur” remained intact enough to launch two more “world’s first” companies.
And this is what makes this difficult to unpick: the more sophisticated the enterprise, the more it resembles aggressive entrepreneurship. Unsophisticated: “This battery exists and works” (when no battery exists). The business making this statement will fail immediately. Sophisticated: “Our breakthrough battery will be production-ready Q1 2026,” “Initial testing shows promising results,” “Chemistry details are confidential for IP protection,” “Timeline may shift based on optimisation needs,” “Early customers are experiencing teething issues that we’re addressing”. This business has a while to run.
Each statement is technically defensible. The battery exists (it’s just not solid-state). Testing showed results (just not the extraordinary claims). Delays happen in manufacturing. The pattern suggests something is fishy here, but each individual statement is plausibly deniable.
Perhaps the most damning evidence comes from Verge Motorcycles’ 2024 audit. PwC, one of the Big Four accounting firms, could not complete the audit because the company lacked “almost all financial information on inventory management, sales receipts, research expenditure, property values.”
This isn’t a simple accounting oversight. This is a company claiming that they have Gigawatt-hour battery production capacity, an order book worth €10+ million, a revolutionary solid-state battery in production and partnerships with 200+ OEMs – and yet they’re unable to show auditors basic financial records about inventory, sales, or research spending.
For context: PwC failing to complete an audit is extremely rare. It means the auditors couldn’t get sufficient evidence to sign off on even the most basic financial statements. This typically indicates one of three situations: records don’t exist (incompetence); records are deliberately withheld (obstruction); or records would reveal fraud (concealment).
None of these scenarios is compatible with a legitimate company on the verge of commercialising revolutionary technology that will “transform the entire automotive industry.”
There’s a crucial difference between breakthrough ambition and systematic deception, and these are things we look at when we’re assessing whether to add a startup to our portfolio. Let’s compare a legitimate company with a breakthrough technology with a company that looks more problematic.
The legitimate company is transparent about challenges and uncertainties with customers and with investors. It publishes research allowing peer review if it’s doing something truly novel. The company is likely to be conservative on timelines, under-promising and over-delivering. It should be able to demonstrate incremental progress, making one major claim at a time; and founders should stay through commercialisation phase.
A problematic company (and we’ve met these companies when being pitched here at Negroni) often only reveals easy-to-verify claims, obfuscating others. There’s no peer review, just a “trust us bro” narrative, and probably something in there about IP protection getting in the way of full transparency. Timelines are aggressive, and slip quietly and repeatedly. The company will demo only things that can’t be easily disproved.
The most telling difference here is verification: legitimate innovators believe in their product, are passionate about bringing people on board, and have an enormous incentive to prove their technology works. They welcome scrutiny because vindication is their goal. Elizabeth Holmes didn’t avoid independent testing because she was protecting IP: she avoided it because Theranos didn’t work.
Donut Lab commissioned VTT to test fast charging and temperature tolerance (easy to pass), while avoiding exposing energy density and cycle life (which would expose the fundamental claim as false). That’s not the behaviour of a company confident in their breakthrough.
Right now, this is working for Donut Labs because multiple parties – VCs, potential acquirers, the media -benefit from looking the other way:
What happens next
As of late March 2026, we’re approaching the moment of truth. Donut Lab claimed Q1 2026 deliveries. The deadline is days away. I can think of three different ways this might go.
The first, and best case. Deliver and it works. Verge delivers motorcycles with working solid-state batteries, independent teardowns confirm 400 Wh/kg and the solid-state chemistry. Long-term testing begins on cycle life, and I look like an idiot and this entire article becomes a cautionary tale about scepticism. I’d lay odds of <5% on this happening.
The next case, deliver and it doesn’t work. Verge delivers motorcycles with batteries that work but that aren’t solid-state, and independent analysis confirms that the batteries are advanced Li-ion, not a revolutionary solid-state technology. Cycle life testing shows a more realistic 5,000-10,000 cycles, not 100,000. Legal disputes begin, the company pivots messaging and Marko moves to his next world’s first venture.
The most likely case in my opinion, they don’t deliver at all. The Q1 deadline is missed, and pushed to Q2 for “optimisation”. There’s a small batch delivery to Verge, the CEO’s brother’s company only, so no independent verification. The timeline continues slipping throughout 2026, the company eventually folds or pivots, the customer fees to secure a reservation are lost. And Marko has moved on and is already chairing his next world’s first company. It’s the lowest-risk path: you string things along, extract maximum value during the “maybe it’s real” phase, then let the company fail while positioning the next venture.
The broader problem
This isn’t just about Donut Lab, or even about Marko Lehtimäki specifically. At Negroni, we’ve seen behaviours like this in founders with companies that are a lot smaller, or that have produced much less buzz. It’s about an incentive structure in venture-backed deep tech that rewards extraction over creation.
When the same person can launch a company with questionable claims, extract €10M+ in value over 12-18 months, exit before any malfeasance is proven, face zero legal or reputational consequences and then immediately launch next “world’s first” company with fresh backing and repeat the process, we have to conclude that the system is optimising for bad faith.
The solution isn’t better fraud detection (though that helps). The solution is realigning incentives.
I like good governance and good regulatory systems, both of which do work to keep us all honest; and regulators could help a lot here by changing some of our paradigms. Deposit-taking on undelivered tech products could be treated as securities, requiring escrow – of course, in the Donut Lab/Verge instance the deposits weren’t even deposits, but reservation fees, but should probably be treated the same. It would be helpful to outsiders if disclosure of related-party transactions (your brother’s company being your only customer, for example) was mandatory. Failure to complete audits should trigger investigation. And possibly the most effective bit of regulatory change we could make here would be to create personal liability for founders making false claims to raise capital.
Investors can do better too. We should demand independent technical validation before investment; or at least invest more in due diligence. It’s good to structure deals with milestone-based payouts, not upfront capital, in an environment where incentives can skew roadmap. And of course, any decent investor should be conducting reference checks on the founder’s previous exits. Ask: “Why did you leave your last company so quickly after acquisition?”
Trust is built in drops, but lost in buckets
Marko Lehtimäki’s own words from his “Ethical by Design” keynote: “Trust is built in drops, but lost in buckets.”
He’s right. The venture ecosystem’s trust has been built over decades of legitimate innovation: startups that kept promises, founders who stayed to commercialise their inventions, breakthroughs that survived scrutiny. That trust is what allows entrepreneurs to raise capital on ambitious visions.
But when that trust is systematically exploited – when founders learn they can make extraordinary claims, extract value, exit before validation, and face no consequences – the bucket has a hole in it.
There are three companies here, and three is enough to show a pattern. AppGyver collapsed after Marko left. Asilab has shown no product. Donut Lab’s deadline is days away, and nothing smells right. At what point does “serially optimistic entrepreneur” become “systematic malfeasance”?
The line exists. This company is standing on it. And the incentive structure that got it there is still in place, waiting for the next founder to walk the same path. I’ll be updating this when that Q1 delivery deadline passes.
This article is based on publicly available information and represents analysis of a pattern of claims and outcomes. Readers are encouraged to examine primary sources and draw their own conclusions. If you have additional information about any of these companies, please drop us a line.