In January 2018 a promotional video appeared showing a Nikola One hydrogen-electric semi truck gliding silently along a Utah road, the desert scrolling past its slab-sided cab. The truck had no functioning powertrain. It had been towed to the top of a hill on a stretch of road near Grantsville, pointed downwards, and released. The camera was tilted so the grade looked flat. Two years later that clip would be the centrepiece of a short-seller report, and four years after that its author would be a convicted felon.
That video is the single most useful artefact of the electric vehicle startup boom, because it compresses the entire pathology into thirty seconds: a beautiful object, an implied capability, and gravity doing the work that engineering was supposed to do. Between roughly 2014 and 2024, well over a hundred companies announced that they would build electric cars. A handful now build them. The rest are bankruptcy dockets, and the honest version of the story is not that they were all frauds — most were not — but that building cars at scale is a machine for destroying companies, and almost nobody who tried had any idea how thoroughly it would grind them down.
The shape of the graveyard
The boom had a specific cause. Tesla's market capitalisation passed Toyota's in mid-2020, and in the same window the SPAC — the special purpose acquisition company, a listed shell that merges with a private business — became the fashionable route to public markets. The combination was combustible. A conventional IPO in the United States requires an S-1 registration statement in which forward-looking revenue projections are legally perilous. A SPAC merger, at the time, was widely treated as a transaction covered by the safe harbour for forward-looking statements, which meant a startup with no product could publish a slide deck projecting billions in 2026 revenue and hand it to retail investors.
So they did. Nikola, Lordstown, Fisker, Canoo, Arrival, Faraday Future, Lucid, Polestar, VinFast, Xos, Hyzon, Proterra, ELMS, Embark and many others reached public markets through shells rather than through the traditional underwriting gauntlet. Cheap money made the valuations plausible; the Federal Reserve's rate increases from early 2022 made them absurd; and the companies that had budgeted for one more easy funding round discovered there wasn't one.
But the SPAC was the delivery mechanism, not the disease. The disease is that a car is the most difficult mass-produced consumer object in the world, and the gap between a running prototype and a vehicle you can build ten thousand times a month, certify in forty countries, insure, finance, service and legally sell is measured in billions of dollars and half a decade.
Henrik Fisker's first act: the Karma and the A123 domino
Henrik Fisker had the best design résumé in the business. He shaped the BMW Z8, then the Aston Martin DB9 and V8 Vantage — three cars that will be on posters forever. In 2007 he and Bernhard Koehler founded Fisker Automotive, and in 2008 they revealed the Karma: a long, low, impossibly wide four-door with a plug-in hybrid powertrain, a solar roof panel, and proportions that made every other green car of the era look apologetic.
The Karma was not a pure EV. It was a series hybrid — battery drive with a 2.0-litre turbocharged General Motors Ecotec engine acting purely as a generator. Production was outsourced to Valmet Automotive in Uusikaupunki, Finland, the contract manufacturer that had built Porsche Boxsters and Saabs. That was the smart part of the plan: don't build a factory, rent someone else's competence.
The battery came from A123 Systems, the Massachusetts lithium-iron-phosphate specialist that was itself a darling of the era. In late 2011 a batch of A123 packs was found to have an internal fault caused by misaligned hose clamps, triggering a recall and a bruising cash transfer between the two companies. In October 2012 A123 filed for Chapter 11 bankruptcy. Fisker's only battery supplier had evaporated, and there was no second source, because a startup building perhaps two thousand cars a year cannot afford to qualify two battery suppliers.
The rest was almost operatic. The US Department of Energy had approved a loan facility of roughly half a billion dollars under the Advanced Technology Vehicles Manufacturing programme; Fisker drew down somewhere under $200 million before the DOE froze the remainder in 2011 for missed milestones. Part of that money had gone into buying the shuttered GM plant at Boxwood Road in Wilmington, Delaware, where a second, cheaper model — the Atlantic — was to be built. No Atlantic was ever produced there. Then, in late October 2012, Hurricane Sandy flooded Port Newark and destroyed several hundred Karmas waiting on the dock, most of them uninsured against exactly that.
Fisker Automotive filed for bankruptcy in November 2013 having built roughly two thousand cars. The assets were bought in a 2014 auction by China's Wanxiang Group, which had already bought A123 out of its own bankruptcy — the supplier and the customer ended up under the same roof, in the wrong order. Wanxiang relaunched the car as the Karma Revero under a new company, Karma Automotive, in Irvine, California, where it still builds vehicles in numbers small enough to be counted by hand.
Fisker Inc: the asset-light bet that wasn't
Henrik Fisker's second attempt looked, on paper, like a man who had learned the lesson. Fisker Inc, founded in 2016 and taken public through a SPAC in October 2020, would own no factory at all. The Fisker Ocean, a mid-size electric SUV, would be built by Magna Steyr in Graz, Austria — the contract manufacturer that has assembled Jaguar I-Paces, Mercedes G-Wagens and BMW 5-Series. Fisker would do design, brand and software. Everything capital-intensive was somebody else's problem.
The Ocean was genuinely appealing. It had a rotating central screen, a full-length solar roof on the top trim, a "California Mode" that dropped every window and glass panel at once, and a claimed range at the top of its class. Production started in Graz in late 2022 and deliveries began during 2023.
Then the asset-light model revealed its flaw: outsourcing the metal does not outsource the company. Fisker still had to build a sales operation, a service network, a parts supply chain, a homologation team for every market, and — decisively — the software. Early Oceans shipped with a stack that could brick the car, fail to recognise key fobs, lose the ability to shift, and require dealer-level intervention for faults that should have been over-the-air fixes. Multiple NHTSA investigations followed. Reporting during 2024 described an internal accounting failure in which the company could not reliably reconcile which customers had actually paid for their cars.
In early 2024, with inventory piling up, Fisker cut Ocean prices dramatically — infuriating owners who had paid full price weeks earlier — and abandoned direct sales for a dealer model that had no dealers yet. A rescue investment from a major automaker, widely reported to be Nissan, collapsed. Fisker Inc filed for Chapter 11 in June 2024. Roughly ten thousand Oceans had been built; a large tranche of unsold cars went to a New York vehicle-leasing company for a fraction of their sticker price. The planned follow-ups — the cheap PEAR hatchback, the Alaska pickup, the Ronin GT — never left the studio. The PEAR, incidentally, was to be built in Ohio by Foxconn, in a plant Foxconn had bought from another company on this list.
Lordstown Motors: buying a cathedral to build a chapel
General Motors ended Chevrolet Cruze production at Lordstown Assembly in north-east Ohio in March 2019, idling a plant that had been building cars since 1966 and taking with it the economic spine of the Mahoning Valley. It became a political symbol immediately. In 2019 GM sold the site to Lordstown Motors, a new company founded by Steve Burns, who had just left Workhorse Group, for a sum in the low tens of millions — a fraction of replacement cost, and a genuinely clever purchase.
The product was the Endurance, a full-size electric pickup aimed at commercial fleets. Its distinguishing feature was four in-wheel hub motors supplied by the Slovenian firm Elaphe, which eliminated driveshafts and differentials but placed a great deal of unsprung mass and a great deal of unproven technology at each corner of a work truck.
Lordstown went public via SPAC in October 2020 on the strength of a preorder book it described in the tens of thousands of units. In March 2021 Hindenburg Research published a report arguing that those "preorders" were largely non-binding letters of intent from entities with no plausible ability to buy trucks. An SEC inquiry and a special board committee followed. Burns and the chief financial officer resigned in June 2021, and the company disclosed it lacked the capital to reach volume production.
Then came the inversion that defines the story: Lordstown sold the plant. Foxconn bought the factory in 2022 for a couple of hundred million dollars and agreed to build the Endurance under contract — so the startup that had been created around owning a great American car plant ended up as a tenant in it. Endurance production began in late 2022 and effectively stopped almost immediately; total output ran to a few dozen trucks, some of which were recalled. Lordstown filed for Chapter 11 in June 2023, sued Foxconn, and emerged as a litigation shell under a new name. A plant that once employed thousands building Cruzes had, in the end, produced fewer electric pickups than a dealership sells F-150s in a slow week.
Nikola: the truck that rolled downhill
Trevor Milton founded Nikola in 2014 with a proposition that was, in its abstract form, correct: long-haul freight is hard to electrify with batteries, hydrogen fuel cells have advantages in that duty cycle, and the missing piece is fuelling infrastructure. Nikola would sell trucks bundled with hydrogen on a per-mile lease — an elegant idea that solved the chicken-and-egg problem by making the manufacturer the fuel retailer.
The execution was something else. The Nikola One prototype unveiled to an arena crowd in Salt Lake City in December 2016 was presented as a functioning truck; it lacked, among other things, motors and gears. The rolling-downhill video followed in 2018. Nikola went public through a SPAC in June 2020, and for a few days that summer its market capitalisation exceeded Ford's — a company with no revenue valued above one that had been building vehicles for over a century.
In September 2020, days after General Motors announced a deal to take an equity stake and build Nikola's Badger pickup, Hindenburg published its report. The GM deal shrank to a non-binding memorandum and then to nothing. Milton resigned within two weeks. He was convicted on securities and wire fraud counts in October 2022, sentenced to four years in December 2023, and pardoned in 2025.
What is most instructive about Nikola is that it kept going and still died. Under new management it built real trucks: the Tre battery-electric and the Tre fuel-cell, assembled in Coolidge, Arizona, plus a hydrogen fuelling brand and a fleet of tube trailers. It also bought the battery maker Romeo Power. But the Tre BEV suffered a battery-pack coolant fault that led to fires and a recall in 2023, hydrogen stations cost enormous sums to build, freight rates were soft, and the company burned cash at a rate that no truck volume could ever cover. Nikola filed for Chapter 11 in February 2025. The fraud made the headlines; the physics of the business finished the job.
Byton: forty-eight inches of screen and no factory discipline
Byton was founded in 2016 as Future Mobility Corporation by Carsten Breitfeld, who had run the BMW i8 programme, and Daniel Kirchert, who had led Infiniti in China. It was headquartered in Nanjing, backed by FAW, Tencent, the battery giant CATL and local government money, and staffed with a genuinely elite roster of German and Chinese engineers.
Its concept, the M-Byte SUV, stole CES in 2018 with a single feature: a 48-inch display running the full width of the dashboard, from A-pillar to A-pillar, called the Shared Experience Display. There was a second screen on the steering wheel hub and a third between the front seats. It was the purest expression of the era's belief that a car is a phone with wheels.
Byton spent like a company that had already succeeded — a Nanjing plant, offices in Munich, Santa Clara, Shanghai and Hong Kong, a lavish CES presence three years running — while never solving the problem of getting the M-Byte through validation and into series production. Reports of internal disputes between the German and Chinese sides multiplied. In mid-2020 the company suspended operations in China and furloughed most of its staff. A 2021 agreement with Foxconn to take over manufacturing quietly lapsed. A Nanjing court subsequently accepted bankruptcy proceedings. Breitfeld had already left, in 2019, to run Faraday Future — a decision that is its own comment on the state of the industry.
China's shakeout: Weltmeister, Singulato and Bordrin
Western coverage tends to treat the Chinese EV sector as an unstoppable monolith. It is more accurate to say that China ran the same experiment as the United States, at ten times the scale, and killed off proportionally as many participants. At the peak, somewhere near five hundred registered "new energy vehicle" makers existed on paper. Fewer than fifty matter now.
Weltmeister — WM Motor domestically — was for several years the most credible of the challengers. Founded in 2015 by Freeman Shen, a former Geely and Volvo executive, it took its name from the German for "world champion", owned its own plants in Wenzhou and Huanggang rather than renting capacity, and at its height delivered tens of thousands of EX5 SUVs a year, running level with Nio and Xpeng. It then failed to complete an IPO on Shanghai's STAR Market, failed again in Hong Kong, lost a long-running trade-secrets case brought by Geely that produced a substantial judgment against it, and filed for bankruptcy restructuring in Shanghai in October 2023. Owning your factories, it turns out, is not a strategy if you cannot fill them.
Singulato — Zhiche Auto — is the purer cautionary tale. Founded in 2014 by Shen Haiyin, it unveiled the iS6 SUV in 2017, signed manufacturing arrangements with established automakers, took local-government money for a plant in Anhui, and reportedly consumed something on the order of ten billion yuan. It never mass-produced a single vehicle for customers. Bankruptcy proceedings followed in 2022.
Bordrin, founded in 2016 by a former Ford engineer and also based in Nanjing, managed the same feat on a smaller budget: two SUV concepts, a plant, a supplier base owed money, and a bankruptcy filing in 2021 amid reports that its founder had left the country. Nanjing alone hosted three of these — Byton, Bordrin and, briefly, others — because local governments were competing to subsidise an industry into existence and were not equipped to tell an engineering programme from a rendering.
Aiways: the exporter with nothing left at home
Aiways deserves its own paragraph because it did something none of its peers managed. Founded in 2017 by executives from Volvo's China operation and FAW, it began exporting the U5 SUV to Europe in 2020 — the first Chinese EV brand to reach European retail customers in any volume, years before BYD and MG made it look easy. It sold cars through Euronics electronics stores in Germany, which was either a stroke of retail genius or a symptom of not being able to afford dealers, depending on your temperament.
What it could not do was sell cars in China, where its domestic volumes were negligible against ferocious competition. Export success on a few thousand units a year cannot fund a car company. By 2023 and into 2024 there were credible reports of unpaid salaries and suspended operations at home, even as the brand's European service arrangements limped on. Aiways is the clearest demonstration that being first to a market means nothing if you have no scale behind you.
Canoo: from subscription to Chapter 7
Canoo began life in 2017 as Evelozcity, founded by Stefan Krause and Ulrich Kranz, respectively the former chief financial officer and chief technical officer of Faraday Future — a lineage that recurs throughout this article. Kranz had run BMW's i division and put the i3 into production, which made him one of very few people in any of these companies who had actually launched a clean-sheet electric car.
The product was distinctive and good. Canoo's Lifestyle Vehicle was a one-box van with a cab-forward silhouette, a flat skateboard platform, and an interior designed by Richard Kim, another BMW i alumnus, as a "loft on wheels" with a wraparound bench. The original commercial model was equally distinctive: no ownership at all, only a monthly subscription including insurance and maintenance.
Then everything got traded away. The subscription model was dropped in favour of conventional sales. A partnership with Hyundai to co-develop a skateboard platform ended. Krause and Kranz both departed. Under new leadership the company relocated from California to Texas, committed to plants in Oklahoma, and pivoted towards fleet customers — winning genuinely prestigious orders from NASA for crew transport vehicles, plus commitments from the US Postal Service and a large agreement with Walmart. It even bought assets from Arrival's collapsing US business in 2024.
None of it generated revenue at any scale. Canoo delivered vehicles in double digits while spending hundreds of millions, and in January 2025 it filed for Chapter 7 — liquidation, not reorganisation. There was nothing left to restructure.
Arrival: microfactories, macro losses
Arrival was founded in 2015 by Denis Sverdlov, a Russian telecoms entrepreneur who had built Yota and served briefly as a deputy communications minister. Its thesis was the most intellectually ambitious of any company here: conventional car plants are ruinously expensive because of stamping presses and paint shops, so eliminate both. Arrival's electric vans would use coloured composite body panels, requiring no paint, assembled by robotic cells in "microfactories" — small, cheap, rapidly deployable plants sited near demand.
If it had worked it would have changed the industry. UPS placed an order for up to 10,000 vans, and Arrival went public through a SPAC in March 2021 at a valuation in the region of thirteen billion dollars, briefly making it one of the most valuable companies ever listed in the United Kingdom by that route.
It never got the microfactory to work. Robotic assembly of a composite vehicle body without hard tooling turned out to require solving a dozen unsolved manufacturing problems simultaneously, and each solution cost time. The company reorganised repeatedly, pivoted from vans to a US-market product designed around the Inflation Reduction Act's tax credits, cut most of its staff, and burned through its SPAC proceeds without series production. Administrators from EY were appointed in February 2024. The intellectual property was picked apart at auction. Arrival is the best evidence that the smartest idea in the sector is still worth nothing if you cannot convert it into a running production line.
Sono Motors: the solar car that couldn't be saved
Sono Motors, founded in Munich in 2016 by Laurin Hahn and Jona Christians, was the most likeable company on this list. The Sion was a modest, boxy family hatchback with several hundred photovoltaic cells laminated directly into its body panels, adding a meaningful number of free kilometres per week in decent sun. It was to cost around €25,000 — genuinely cheap — and it had moss in the dashboard as an air filter, which tells you everything about the spirit of the thing.
Sono was substantially crowdfunded, with tens of thousands of reservation holders who behaved less like customers than like members. It listed on Nasdaq in 2021. But the Sion needed a contract manufacturer, a full type-approval programme and hundreds of millions of euros to reach production, and by late 2022 the money was running out.
What followed was extraordinary and futile: a public "#savesion" campaign in which the community was asked to fund the car directly, raising tens of millions but well short of the target. In early 2023 the company cancelled the Sion outright and pivoted to selling solar retrofit kits for buses and refrigerated trucks — a real business, with real customers, and about one percent of the ambition. German insolvency proceedings followed. The solar business survives in some form. The car does not.
Faraday Future: a decade of nearly dying
Faraday Future has been about to fail since 2016 and has not yet finished. Founded in 2014 and funded by Jia Yueting, the founder of the Chinese streaming and gadget conglomerate LeEco, it unveiled the FFZERO1 — a single-seat, batmobile-shaped concept with a slot for your phone in the steering wheel — at CES 2016, alongside plans for a billion-dollar factory in North Las Vegas.
The factory was abandoned before completion. LeEco imploded. Jia was placed on China's list of defaulting debtors and later filed for personal bankruptcy in the United States. An investment from the property developer Evergrande in 2018 turned into a bitter arbitration within months. Faraday bought a modest former tyre plant in Hanford, California instead, hired and lost a succession of executives — including Byton's Carsten Breitfeld, who ran the company for about three years — went public via SPAC in 2021, then disclosed an internal investigation into inaccurate statements about how many reservations it actually held, which led to a regulatory settlement.
The FF 91 Futurist eventually reached customers in 2023, priced above $300,000, with rear seats that recline like a business-class cabin and more screens than occupants. Deliveries have been counted in single and low double digits. The company has survived on repeated dilutive financings and a series of reverse stock splits, each one resetting a share price that keeps returning to the same place. It is the industry's most durable near-death experience: eleven years, several billion dollars, and a product that exists mainly as proof that it can.
The three-wheel loophole and the arithmetic of dilution
ElectraMeccanica, based in Vancouver, tried to dodge the whole problem. Its Solo was a single-seat, three-wheeled electric commuter built in Chongqing by the motorcycle maker Zongshen. Three wheels meant classification as a motorcycle in the United States, which meant escaping the full battery of passenger-car crash and safety standards — a genuine regulatory shortcut, and one that several small manufacturers have used.
The shortcut did not save it. In 2023, after reports of sudden loss of propulsion, the company recalled every Solo it had sold — a few hundred cars — halted sales permanently and offered to buy the vehicles back from owners. A planned merger with a British electric-truck maker was announced and then terminated. What remained was a listed entity with cash and no product, looking for someone to merge with. The lesson is unkind but clear: avoiding regulation does not avoid engineering.
Mullen Automotive occupies a different category — not dead, but instructive. It reached public markets in 2021 through a reverse merger into an existing listed company, acquired a controlling stake in the electric truck maker Bollinger Motors and the assets of the bankrupt Electric Last Mile Solutions, and has since produced commercial vans and trucks in small quantities. It has also executed a remarkable sequence of reverse stock splits, each following a collapse in the share price, funded by convertible instruments that convert at a discount to a falling market. For retail shareholders the arithmetic is brutal: the company can be alive, shipping vehicles, and still destroy essentially all of the equity above it. Survival and shareholder return are separate questions.
Why building cars at scale destroys companies
Strip away the personalities and the same five forces appear in every one of these stories.
Capital intensity is not a metaphor. A clean-sheet vehicle programme — design, engineering, validation, tooling, homologation — runs to well over a billion dollars before a single customer car exists. A plant capable of a few hundred thousand units a year costs a similar amount again. Stamping dies for a single body panel cost millions and take a year to cut. Startups routinely budget for the visible costs and are ambushed by the invisible ones: test fleets, durability mules, cold-chamber and hot-desert programmes, and the dozens of complete vehicles that must be destroyed in crash testing.
Certification is a gauntlet, and it repeats per market. In the United States a manufacturer self-certifies against the Federal Motor Vehicle Safety Standards, but self-certification means you must actually run the tests, including electric-specific requirements for post-crash electrical isolation and an audible warning for pedestrians at low speed. The EPA and CARB have their own processes. Europe requires whole-vehicle type approval under UNECE regulations, with different content. Every additional market is a new programme, new documentation, and new hardware.
Suppliers do not want your business. A Tier 1 supplier tools up for a programme on the assumption of hundreds of thousands of parts. At a startup's volumes the per-piece economics are terrible, so the supplier demands non-recurring engineering charges paid in advance, high minimum order quantities, and often cash up front, because your credit is worthless. You become the least important customer in the building, first to be deprioritised when semiconductors or cells are short — which is exactly what happened across 2021 and 2022.
The working-capital trap is the actual killer. An established automaker pays suppliers on extended terms and turns finished inventory quickly, so growth funds itself. A startup prepays for parts, holds inventory for months, sells each car below its build cost, and then waits to be paid. Every additional unit sold consumes cash. Growth accelerates death. Rivian and Lucid both lost sums well into six figures on every vehicle in their early quarters, and both had the balance sheets to absorb it; Fisker and Canoo did not.
Then the car keeps costing you money. You owe warranty coverage for years, which means a parts distribution network, trained technicians, diagnostic tooling and a reserve on the balance sheet. In much of the United States, franchise laws complicate direct sales, but a franchise network is only attractive to dealers who expect volume. And modern EVs are software products: an over-the-air update pipeline, a connected-car backend, a certificate authority, an app. Fisker's Ocean did not fail because the sheet metal was wrong. It failed because the software was not finished and there was no organisation capable of finishing it.
Underneath all of this sits a simpler truth. A show car is roughly one percent of the work. The remaining ninety-nine percent is design validation builds, production validation builds, supplier part approvals, run-at-rate trials, and the grinding process of making the ten-thousandth car identical to the first. It is unglamorous, it cannot be compressed with money past a point, and it does not photograph well at CES.
What the survivors actually had
Rivian is the clearest counter-example, and the reasons are unromantic. RJ Scaringe founded it in 2009 and spent nearly a decade in near-total silence before showing a product — a decade of engineering rather than a decade of announcements. In 2017 he bought a complete, functioning car plant in Normal, Illinois from Mitsubishi for a sum in the mid-tens of millions, inheriting a paint shop, a body shop and, crucially, a workforce that had built cars. In 2019 Amazon and Ford invested, and Amazon ordered 100,000 electric delivery vans — a captive, contracted, non-speculative order book. The R1T pickup reached customers in September 2021, beating Ford, GM and Tesla to the electric pickup market. The November 2021 IPO raised close to $12 billion in one of the largest US listings in years, and a joint venture with Volkswagen announced in 2024 brought both cash and validation of Rivian's software and electrical architecture. Rivian has lost enormous sums. It has never lacked the money to keep going.
Lucid's survival has a single cause: Saudi Arabia. The company started in 2007 as Atieva, a battery-pack engineering firm — it supplied the packs for every car in Formula E's early seasons, which is an underrated credential — and became a carmaker only in 2016. Peter Rawlinson, chief engineer of the Tesla Model S, joined and eventually ran it. The decisive moment was 2018, when the Public Investment Fund put more than a billion dollars in and later took majority control. That let Lucid build its own plant at Casa Grande, Arizona, and deliver an Air sedan with an EPA range rating above 500 miles and a powertrain efficient enough that Aston Martin licensed it. Lucid's volumes have consistently disappointed and its leadership has changed. Its shareholder simply does not need it to be profitable this decade.
Polestar cheated in the most sensible way possible: it was never a startup. It began as Flash Engineering, a Swedish touring-car team founded in 1996, became Volvo's performance partner, and was absorbed by Volvo in 2015 and relaunched as a standalone electric brand in 2017. That meant every hard problem was already solved by someone else. The Polestar 2 is built on Geely's CMA architecture alongside the Volvo XC40 in Luqiao; the Polestar 3 shares a platform and plants with the Volvo EX90, including Volvo's South Carolina site; homologation, purchasing, warranty and dealer infrastructure all lean on the parent. Polestar has still had a miserable financial time — a SPAC listing in 2022, delayed accounts, restructuring, and a change of chief executive to Michael Lohscheller, who had previously run Opel and, briefly, VinFast. It survives because Geely wants it to.
VinFast is the same lesson in a different currency. It is a subsidiary of Vingroup, Vietnam's largest conglomerate, founded in 2017 by Pham Nhat Vuong. It began by building licensed BMW-derived petrol cars in Haiphong with Pininfarina styling — an apprenticeship in manufacturing before attempting anything original — then pivoted to electric-only in 2022. Its first US product, the VF 8, was reviewed savagely on quality and software. Its North Carolina plant has slipped repeatedly. And it has kept going, because Vuong has repeatedly pledged personal wealth to it, and because a large share of its output goes to a taxi operator he also controls. Captive demand plus a patron is not a business model most companies can copy, but it is a survival model.
The pattern is unmissable. The survivors did not have better ideas than Arrival or nicer products than Sono. They had access to patient capital measured in billions, a manufacturing base they either inherited or bought secondhand from someone competent, a real supply chain relationship, and a product a customer could actually take delivery of.
Tesla and BYD were never the template
Both survivors-in-chief are routinely cited as proof that a startup can win, and both prove something narrower.
Tesla nearly died twice — in December 2008, and again during the Model 3 "production hell" of 2018 when it was assembling cars in a tent in the Fremont car park. It survived the first crisis on a Daimler investment and a DOE loan, and the second on the willingness of capital markets to keep funding it. Crucially, it started tiny: the original Roadster was a Lotus Elise glider with a battery, built by Lotus in Norfolk, which meant Tesla learned the business on someone else's chassis. Then it bought the NUMMI plant in Fremont — a Toyota-GM joint venture facility — for a nominal sum in 2010. Cheap inherited capacity, again.
BYD is even less of a startup story. Wang Chuanfu founded it in 1995 as a battery manufacturer, spent eight years becoming one of the largest cell producers on earth, and only then bought a small state-owned carmaker in 2003 to get a production licence. By the time BYD became the world's largest EV maker it had thirty years of vertical integration behind it — cells, semiconductors, motors, and eventually its own ships. It stopped building pure combustion cars in 2022 from a position of total strength. Nobody founded a company in 2018 and replicated that.
How to read the next one
There will be more. Solid-state battery startups, autonomous trucking companies, electric aviation, hydrogen ventures and the next generation of Chinese exporters will all arrive with beautiful renders and a market projection with a hockey stick in it. The graveyard above is only useful if it makes you better at the next assessment, so here are the questions it answers.
Ask who is building the vehicle and whether they have built one before. Ask whether the company owns, rents or has merely announced a plant, and whether that plant has ever produced a saleable car. Ask what the order book actually is — a binding contract with a named fleet buyer, or a refundable $100 deposit counted as demand. Ask how many months of cash remain at the current burn, and what the last financing cost in dilution. Ask whether a single customer has taken delivery, paid full price, and driven it home.
And then go and look at the thing itself. Every company in this article had a compelling object at its centre — Fisker's proportions, Byton's screen, Arrival's paintless composite panels, Sono's solar skin. The object was never the problem. Sit in one at a show, admire the idea, and understand that between that car and a car you can buy sits four years, two billion dollars, and a production line that has to run correctly ten thousand times in a row.
