NIO: Betting the Company on a Battery You Never Own

I. Introduction & Episode Roadmap

On a humid August morning in 2026, in Quanzhou, a mid-sized port city in Fujian province better known for maritime history than for automotive engineering, a small crowd gathered around what looked like a shipping container with ambitions. It was NIO's four-thousandth battery swap station in China, and the first built on the company's fifth-generation architecture — a rebuilt underlying frame that, for the first time, could accept vehicles from all three of NIO's brands through a single mechanism.¹⁰ A car drives in. A robotic system underneath unbolts a several-hundred-kilogram battery pack, slides it out, slides a fully charged one in, and bolts it back. The driver never gets out. The whole thing takes about as long as a coffee order.

That same week, the company confirmed it had crossed 120 million cumulative battery swaps — a number it had reached 100 million of only six months earlier, in February 2026.¹⁰ By any engineering measure, this is one of the more remarkable pieces of physical infrastructure any automaker has built in the last decade. No other carmaker on earth has attempted it at this scale.

And here is the tension that makes 蔚来 NIO worth two hours of anyone's attention: the company spent more than a decade and well over RMB 100 billion in cumulative losses proving that the technology works, and it still cannot demonstrate that the business underneath it works. In August 2026, NIO began handing the stations away — transferring ownership of 36 swap stations in Wuhan to a state-owned enterprise and retreating into the role of operator and technical manager.⁹ The most distinctive asset in the company's story is being moved off its own balance sheet.

Meanwhile, something genuinely new happened. In March 2026, NIO reported its first-ever quarterly operating profit: RMB 807.3 million on a GAAP basis, RMB 1.25 billion adjusted, on a vehicle margin of 18.1%.² It arrived roughly two years after William Li first told analysts it would. It has now held for three consecutive quarters. Whether that is a genuine inflection or a temporary truce in a price war that has already killed weaker rivals is the question this episode exists to test.

The road map: we start with founding and the Tesla-chasing years, move through the 2019 near-death experience and the government rescue that saved the company, then spend serious time on the battery-swap bet and whether it is a moat or a subsidised cost centre. From there: the multi-brand gamble with 乐道 ONVO and 萤火虫 Firefly, the ownership and capital structure after Abu Dhabi's CYVN Holdings became the largest shareholder, the five-year saga of broken profitability promises, the brutal shakeout underway in China's EV market, and a stalled European ambition that management's own actions tell us more about than any outside commentary could.

Throughout, one pattern keeps recurring, and it is worth naming up front so you can watch for it: every time NIO's capital-heavy strategy has run out of road, someone with a state balance sheet has stepped in. That happened in 2020. It happened again, in a sovereign-wealth form, in 2023. It is happening now, in 2026, with the swap network itself. Whether you read that as smart partnership or structural dependency is, in the end, the investment question.

↑ Back to Top

II. Origins: William Li's Vision and the IPO Gamble (2014–2018)

The founding scene for NIO is not a garage. It is a spreadsheet.

李斌 William Li — Li Bin — was not a car guy by training. He was an internet guy who had already built a company around cars. In 2000 he founded Bitauto, an automotive content and dealer-marketing platform that rode China's car-ownership boom to a US listing, giving Li something rarer in the Chinese EV cohort than engineering credentials: an intimate, granular understanding of how Chinese consumers actually shop for automobiles, and how badly the dealer experience served them. Li had spent fifteen years watching Chinese buyers get treated as marks by dealership finance desks. He founded NIO in November 2014 with a thesis that flowed directly from that: you do not beat Tesla on horsepower or price. You beat it on how it feels to own the car.

That is a softer thesis than it sounds, and it produced a company with genuinely unusual priorities from day one. NIO Houses — part showroom, part clubhouse, part co-working space and children's play area, sited in expensive urban real estate — were not a marketing afterthought. They were the product. The NIO App built an owner community with the density of a social network. Chinese owners refer to themselves and each other as 蔚来车主 in a way that has no real analogue at BMW or Audi dealerships. In an industry where "brand community" is usually a line in a deck, NIO spent real money making it a physical place.

The early flexes were pure halo. In 2016 NIO unveiled the EP9, an electric hypercar that set a lap record at the Nürburgring, and the company entered Formula E in the championship's inaugural season. Neither was ever going to sell volume. Both existed to establish, quickly and expensively, that a Chinese startup could build something world-class — a credibility problem that Chinese premium brands had spent thirty unsuccessful years trying to solve. Judged as marketing spend rather than as a business line, it worked: NIO arrived with a reputation.

But the founding decision that shapes everything downstream was quieter and far more consequential. Tesla had solved the charging problem by building Superchargers — a capital-light-ish network of stalls where the car sits and waits. NIO looked at the same problem and made the opposite choice: it would build stations where a robot swaps the depleted pack for a charged one in minutes.

The engineering logic was defensible. Fast charging degrades battery cells over time and is fundamentally constrained by physics and grid capacity; swapping sidesteps both. It also unlocked a commercial idea NIO would later formalise as Battery-as-a-Service: if the battery lives in the network rather than in the car, the customer can rent it, which strips the single most expensive component out of the sticker price and turns it into a monthly subscription. Elegant.

The cost was that NIO would have to become a property developer and a robotics manufacturer as well as a carmaker. Every station requires land, a grid connection, hardware, an inventory of expensive batteries sitting idle waiting to be swapped, and ongoing maintenance. Tesla's Supercharger network is capital-intensive; NIO's swap network is capital-intensive and carries a working-capital drag in the form of the battery inventory itself. No other global automaker has been willing to make that trade. That is either NIO's great insight or its great error, and the rest of this story is an extended attempt to find out which.

The IPO came before any of it was proven. On September 12, 2018, NIO listed on the NYSE at $6.26 per ADS, raising roughly $1 billion at an implied valuation near $6.4 billion.²⁷ It was, by the standards of the moment, a disappointment — the company had reportedly hoped for a valuation as high as $20 billion, and priced near the bottom of its range.²⁷ Deliveries of the ES8, the seven-seat SUV positioned against Tesla's Model X, had only begun in June of that year.

For investors, the takeaway from this period is not the vision — visions are cheap — but the architecture of the bet. NIO went public as a company that had committed to the most capital-hungry differentiation strategy available in the industry, with a premium-only product line in a market that had not yet proven it would pay premium prices for a domestic brand, and with the losses front-loaded. That structure left almost no margin for a demand shock. One arrived within months.

↑ Back to Top

III. The Near-Death Experience and the Hefei Bailout (2019–2020)

If you want to understand why William Li talks about cash reserves the way a man who has been through a shipwreck talks about lifeboats, 2019 is the year to study.

The Chinese government cut new-energy vehicle subsidies sharply that year. NIO recalled nearly 5,000 ES8s over battery fire risk. Deliveries stalled. The stock, which had briefly traded above $10, collapsed toward $1.19 — territory where NYSE delisting rules start to matter. Layoffs came in waves. Li reportedly earned the nickname "worst-off man of 2019" in Chinese media. The company was, by most external readings, weeks rather than quarters from running out of money.

What saved NIO was not a product cycle or a retail investor rally. It was a provincial government.

In April 2020, a consortium of state-linked investors from Anhui province — Hefei City Construction & Investment, State Development & Investment Corporation, and Anhui Provincial Emerging Industry Investment — agreed to inject RMB 7 billion, roughly $990 million, into a newly created entity called NIO China, taking 24.1% of it.⁵ The framing at the time was blunt. Bernstein's analyst, quoted by TechNode, called it what it was: a bailout, not a growth round.⁵

The strings attached were substantial and are still visible in the company today. NIO relocated its manufacturing and R&D centre of gravity to Hefei — a city that has since branded the site "NIO City" and built a significant part of its own industrial identity around the company. And NIO committed to a target: 500,000 units of annual sales by 2024.⁵

Hold that number. NIO's actual group deliveries in 2024 were 221,970 — well under half.²⁶ The company missed the commitment that came with its rescue by a wide margin, and the local government did not, as far as any public disclosure shows, enforce a penalty. That tells you something real about the nature of the relationship: this was industrial policy dressed as private equity. Hefei was buying an anchor tenant for an EV industrial cluster, and NIO was buying survival. The target was a negotiating artifact more than a covenant.

For a long-term investor, this episode establishes two things that matter for everything that follows.

The first is about management's revealed risk posture. The 2019 crisis is the formative trauma of this management team, and you can see its fingerprints in every subsequent decision — the obsession with quarterly cash balance disclosure, the willingness to take dilutive capital at unattractive moments rather than run the balance sheet thin, the 2023 headcount cut executed days before a capital raise closed. Companies that have nearly died once tend to raise money when they can, not when they need to. NIO does this consistently. That is a defensible instinct, but it has a shareholder cost, and we will tally that cost in Section VII.

The second is the pattern flagged in the introduction. NIO's most capital-intensive commitments have repeatedly required state or sovereign capital to stay funded. The Hefei deal was the first instance. Abu Dhabi's CYVN was the second. The 2026 transfer of swap-station assets to provincial state-owned enterprises is the third. Each individually can be explained as opportunistic partnership. Together they describe a business model whose defining feature — building and owning physical energy infrastructure at national scale — has never once been financed by its own operating cash flow.

That is not a moral judgment. Plenty of great infrastructure businesses were built with public capital. But it does change how you underwrite the moat, because a moat that requires periodic external refills is a different asset than one that self-funds. Which brings us to the network itself.

↑ Back to Top

IV. NIO Power: The Battery-Swap Bet — Moat or Cost Center?

Picture a NIO owner leaving Shanghai for a family trip during the 2026 Spring Festival travel rush — the largest annual human migration on the planet. Along the expressway, they pull into one of the 1,049 highway swap stations NIO has built, connecting 550 cities across sixteen major city clusters.¹⁰ Three minutes later, they are back at highway speed with a full pack, while a Tesla driver at the same rest stop is settling in for a twenty-five-minute wait. Over that holiday period, NIO's network delivered more than 2.07 million swap services.

This is the strongest version of the bull case, and it deserves to be stated at full strength: the capability is real, it is at scale, and it cannot be replicated quickly. By August 2026 NIO operated 4,006 swap stations in China alongside a large charging network, had cumulatively performed 120 million swaps, and had invested over RMB 20 billion in the buildout.¹⁰ Ten years of land acquisition, grid negotiation, and robotics iteration is not something a competitor buys off a shelf.

Now let us test the claim, because "hard to replicate" and "economically valuable to shareholders" are different propositions.

The financing structure tells you who really owns it

NIO never fully owned the batteries. In 2020 it spun the battery assets into Wuhan Weineng Battery Asset Co. — a joint venture in which 宁德时代 CATL, Hubei Science & Technology Investment Group, and a Guotai Junan International subsidiary each took roughly a quarter alongside NIO. Successive capital raises have diluted NIO's own stake in that vehicle to roughly 19.4%. In March 2025, NIO and CATL signed a strategic partnership in Ningde under which CATL would invest up to RMB 2.5 billion in NIO Power itself, unify swap technical standards across both companies' networks, and extend CATL's Choco-Swap standard to the Firefly brand.¹¹ Reuters subsequently reported that CATL had been in talks for a controlling stake — a claim NIO neither confirmed nor squarely denied, saying only that the two would "deepen collaboration on capital and business."

Read that sequence carefully. The company's signature differentiator is majority-owned by other people, and its largest supplier — a company that controls roughly 43% of China's battery market — has been steadily increasing its claim on it.²¹ Supplier power is not an abstraction here. It sits inside the moat.

Break-even keeps moving

William Li has said publicly that a swap station needs roughly 60–70 swaps per day to break even. In February 2025, management described the Shanghai network as "close to profitability." Independent analysts, surveying the same disclosures, put network-wide battery-swap break-even at around the end of 2026 — a decade after the first stations went live and years after the original investment thesis implied.⁸ On the Q4 2025 call, management said the service and community division reached profitability in 2025 and expected further improvement in 2026 even while adding more than 1,000 stations.³ That is progress, but note the framing: "service and community" is a broader bucket than the swap network alone, and NIO has not disclosed a standalone operating margin for NIO Power. The absence of that disclosure, from a company that discloses vehicle margin to a decimal place every quarter, is itself information.

The perk that didn't survive the unit economics

Here is the single most useful piece of disconfirming evidence in NIO's own record, because it is the company's own behaviour rather than anyone's opinion.

Free lifetime battery swaps were, for years, the headline reason to buy a NIO. It was the thing owners told friends about. Then it was progressively dismantled: capped at four per month in April 2023, converted into a paid add-on that June, discontinued entirely for new buyers in August 2023, capped further that November, and retiered again in September 2024. In July 2025, NIO owners reported that free-swap benefits were being revoked even for some legacy holders, on grounds of alleged policy abuse.[^10]

A benefit marketed as permanent differentiation was withdrawn in six discrete steps across roughly two years. There is only one plausible reading of that: at the price NIO was charging, unlimited swapping did not pay. The company discovered the true unit economics of its own network in production and repriced accordingly. This does not mean the network is worthless — it means the version of the moat that was originally sold to customers and investors, in which swapping was a free forever-perk that locked in loyalty, was not economically survivable. Any investor still carrying that version of the thesis should mark it down.

Handing the asset to the state

In August 2026, NIO transferred 36 swap stations in Wuhan to Optics Valley Traffic Company, a state-owned entity, moving from owner to "property and technical service provider" — retaining professional operation and technical management while relinquishing ownership.⁹ This is not a one-off: NIO has arrangements with more than 40 state-owned platforms and financial institutions across 25 provinces and regions, covering over 800 swap stations, under a model explicitly described as "state-owned asset ownership plus NIO professional operation."⁹ Financial terms were not disclosed. The company had already slowed its own construction pace over the prior two years specifically to control capital spending, leaning on partners instead — including a June 2026 agreement with Anhui's Zhongan Energy to jointly build 500 stations within a year.¹⁰

The strategic logic is sound: get the capital intensity off NIO's balance sheet while keeping the customer experience. But note what it concedes. NIO is not de-capitalising a business that throws off cash; it is de-capitalising one it could not affordably keep expanding alone. And it is doing so, once again, by leaning on state capital — the same playbook as Hefei in 2020, applied to the asset that was supposed to be the reason NIO deserved a premium in the first place.

The alliance that hasn't converted

Since late 2023, 长安 Changan, 吉利 Geely, 奇瑞 Chery, 江淮 JAC, Lotus, 广汽 GAC, and 一汽 FAW have all signed on in some form to share NIO's swap standard. This is a genuine attempt to manufacture a network effect: if rival brands' cars use NIO's stations, utilisation rises, break-even arrives, and the standard becomes infrastructure rather than a company feature. On the Q2 2026 call, CFO Stanley Qu confirmed NIO charges other OEMs "an admission fee for access to the power swap network," while acknowledging that details were still under discussion.¹⁸

"Still under discussion," nearly three years after the first agreements, is the tell. NIO has not disclosed meaningful third-party swap volume. Announced alliances are not revenue, and a technical standard that partners have signed but not populated with cars is a press release, not a network effect.

The calibrated conclusion

Weighing all of it: the historical record does not reject the claim that battery swapping is a real differentiator, but it narrows it substantially. What survives is "expensive, genuinely differentiated, operationally impressive, and not yet self-sustaining." What does not survive is "unbeatable moat" — because a moat that required a supplier to co-own it, a state to refinance it, and its own headline customer perk to be withdrawn is not yet generating the excess returns a moat is supposed to produce.

Two things would change that assessment. First, a standalone NIO Power operating margin or break-even disclosure — the fact that NIO has never published one is the most conspicuous gap in its reporting. Second, evidence that the SOE asset-transfer programme meaningfully reduces group capex without degrading swap availability, because the entire premium NIO charges rests on that availability. If stations start getting rationed under state ownership, the pricing power goes with them. And pricing power is the whole reason the network exists — which is exactly where the core business comes in.

↑ Back to Top

V. The Core Business: Lineup, Segments, and Margin Trajectory

Strip away the robots and the clubhouses and NIO is, financially, a company that sells cars. In FY2025, group revenue was RMB 87.49 billion, up 33.1% year over year. Of that, RMB 76.88 billion came from vehicle sales and RMB 10.60 billion from "other sales" — a bucket that includes NIO Power services, accessories, and used cars.² Other sales grew faster, up 41.4%, which is the direction management wants. But the ratio is the point: roughly seven-eighths of the top line is metal moving off a lot. NIO Power is not yet a profit engine standing beside the car business. It is a customer-acquisition and retention expense that happens to collect some revenue.

The lineup maps onto China's premium tiers with some precision. The ES8 large SUV, the ES6 and EC6 mid-size SUVs, and the ET5 sedan and ET5 Touring form the RMB 300,000–400,000-plus volume core — the bracket where a Chinese buyer would otherwise be looking at a BMW 5 Series or an Audi Q5. Above them sits the ET9 flagship sedan, launched from RMB 788,000, which functions much as the EP9 did a decade earlier: a statement of capability aimed at the small slice of Chinese buyers who might otherwise buy a Mercedes S-Class, and at everyone else's perception of the brand.

Underneath the pricing sits Battery-as-a-Service, and it is worth explaining plainly because it distorts every comparison you might want to make. Under BaaS, the buyer purchases the car without the battery and subscribes to the pack monthly. The sticker price drops by roughly RMB 70,000–130,000 depending on capacity. That makes a NIO look cheaper against a comparably-equipped rival on the showroom board, and it converts a chunk of what would have been one-time vehicle revenue into a recurring service stream. It also means NIO's reported average selling price and its true revenue per customer diverge, and that a NIO owner is structurally harder to poach — cancelling a BaaS subscription means giving the car back its battery problem. That is real switching cost, and it is the most underrated part of the swap architecture.

The margin arc is the actual story

Now the number that matters most for anyone underwriting this business. Vehicle margin — the gross margin on car sales specifically — went 9.5% in 2023, to 12.3% in 2024, to 14.6% in 2025, to 18.1% in the fourth quarter of 2025 and 18.5% in the second quarter of 2026.²¹

The reason to trust this trend more than most turnaround stories is that it did not arrive in a single quarter. It compounded across three consecutive fiscal years, through a brutal price war, and it did so while volumes grew rather than shrank — deliveries rose 46.9% in 2025 to 326,028 units.² Margin improvement driven by volume-led fixed-cost absorption, platform standardisation, and component cost-down is a different and more durable thing than margin improvement driven by mix cherry-picking or accounting reclassification. The Q4 2025 improvement, at 18.1% versus 13.1% a year earlier, coincided with deliveries up 71.7%.² That is operating leverage doing what operating leverage does.

But do not round the story up. FY2025 as a whole was still deeply lossmaking: a GAAP operating loss of RMB 14.04 billion and a net loss of RMB 14.94 billion, even with a profitable fourth quarter.² One good quarter at the end of a year does not retire fourteen billion renminbi of annual losses. And the Q2 2026 GAAP picture was still negative — an operating loss of RMB 347.2 million and a net loss of RMB 528.0 million, with the "profit" arriving only on a non-GAAP basis at RMB 206.9 million.¹ The gap between the GAAP and non-GAAP numbers is share-based compensation. Investors are entitled to decide for themselves whether stock issued to employees is a real cost. It is.

The R&D question

The other number worth interrogating: R&D expense fell 18.7% in 2025 to RMB 10.61 billion, and on an adjusted basis fell 22.6%.² In Q4 2025 alone, R&D was down 44.3% year over year.² SG&A, meanwhile, rose 2.2% for the year and fell 27.5% in the fourth quarter.²

A cut of that magnitude in a technology-intensive industry demands scrutiny. There is a benign reading and a worrying one, and the honest answer is that both are partly true. The benign reading: NIO spent years building platforms — the NT3.0 vehicle architecture, an in-house smart driving chip, its own operating system — and platform spending is lumpy. Once the architecture exists, incremental models cost far less. Management has evidence for this: on the Q4 2025 call it disclosed that the second-generation Shenji chip had completed tape-out and entered mass production, with claimed performance equivalent to three Nvidia Orin X chips at significantly lower cost.³ If true, that is a structural cost saving on every vehicle, purchased with the R&D spending of prior years.

The worrying reading: this is a company that promised profitability, missed repeatedly, and then found the profit partly by cutting the line item that determines competitiveness three years out — in a market where 华为 Huawei and 小米 Xiaomi are spending heavily on exactly the software and autonomy capabilities that differentiate premium Chinese EVs. The 44% year-over-year R&D reduction in the very quarter NIO first reported an operating profit is a correlation that deserves to be stated plainly, not smoothed over.

The resolution is empirical and will show up in product: if NIO's 2026–2027 model cycle lands competitive smart-driving and cabin software without the spending, the platform-maturity story holds. If reviewers start describing NIO's software as a generation behind, the cut will have bought a quarter's headline at the cost of the franchise. This is not resolvable today. It is worth watching.

Which raises the question of where the volume that absorbs all that fixed cost is actually supposed to come from — because the NIO brand alone was never going to supply it.

↑ Back to Top

VI. The Multi-Brand Gamble: ONVO and Firefly

By 2023, the arithmetic had become inescapable. NIO had built a national energy network and a full-stack technology base sized for a company selling many hundreds of thousands of cars a year, and it was selling about 160,000. Premium-only positioning in China has a ceiling, and NIO had found it. The choice was to shrink the ambition or broaden the funnel.

Li chose to broaden it, and in doing so quietly conceded something the company had spent nine years denying: the brand-first, price-insensitive thesis that founded NIO could not, by itself, pay for NIO.

乐道 ONVO launched in May 2024 with the L60, a family SUV starting around RMB 250,000, aimed with almost impolite directness at Tesla's Model Y — the single best-selling premium vehicle in China. Li's stated long-term ambition was a "35-55-10" mix across the group, with ONVO carrying the majority of volume. The strategic logic was clean: ONVO buyers would use the same swap stations, amortising the network across a far larger base, while NIO brand kept the pricing power and the halo.

The execution was not clean.

The ONVO stumble, examined

ONVO cleared 10,000 deliveries a month in December 2024. Then it fell off a cliff: 5,912 units in January 2025, 4,049 in February — the lowest since launch — and 4,820 in March, against an internal target of 20,000 a month.¹⁷ Alan Ai (艾铁成), ONVO's president, had publicly declared in early March that he would not "bow his head and admit defeat." On April 2, 2025, he resigned as ONVO president and as a NIO senior vice president, citing the L60's failure to achieve "sales volumes matching its product capabilities" and acknowledging "shortcomings and deficiencies in the marketing efforts."¹⁷ Shen Fei, previously head of NIO Power, took over, with core product and marketing decisions moved directly under Li and president Qin Lihong.¹⁷

How management explained the miss is analytically more interesting than the miss itself. On the March 2025 earnings call, Li attributed 30–40% of the shortfall to "negative public sentiment" — a reference to a period of intense online criticism of NIO's financial position. That is, on its face, blame-shifting: sentiment is not something a management team can be held to account for, which is precisely what makes it a convenient explanation.

To Li's credit, he did not stop there. He also cited lower brand awareness than competitors, order pressure after the initial backlog was fulfilled, insufficient sales experience among new staff, and battery supply constraints. Those are execution failures, they were named specifically, and the executive responsible left. Compared with the industry norm of attributing every miss to "market conditions," a partial and specific admission of internal fault is worth something in assessing management credibility.

But the substance stands: NIO built a new brand, a new dealer channel, and a new sales force simultaneously, and discovered that selling a RMB 250,000 family car to a value-conscious buyer is a fundamentally different discipline from selling a RMB 400,000 statement car to an enthusiast. The company's own president acknowledged as much on the Q2 2026 call, describing ONVO's brand awareness as "comparable to NIO's level 5-6 years ago."¹⁸ That is a striking admission eighteen months after launch, and it bounds how fast the multi-brand thesis can compound.

Firefly, and what actually happened to the mix

萤火虫 Firefly, launched in China in April 2025 and in Europe that August, took the opposite approach: a compact, youth-oriented, deliberately international product designed from the outset for overseas markets and built around CATL's Choco-Swap standard rather than NIO's own.¹¹ It has had a smoother run than ONVO's rocky start.

The recovery in the numbers is real, and it is worth being precise, because the mix picture is frequently misstated. For full-year 2025, group deliveries of 326,028 split 180,722 NIO brand, 112,268 ONVO, and 33,038 Firefly.² The NIO brand remained comfortably the largest of the three — it did not become the smallest. By the second quarter of 2026, group deliveries of 107,658 split 60,945 NIO, 29,124 ONVO, and 17,589 Firefly.¹

Run that forward and the "35-55-10" target looks distant. On Q2 2026 volumes the mix was roughly 57-27-16 — NIO brand still carrying the majority, ONVO at half its intended share, Firefly ahead of plan. ONVO recovered from its 2025 trough to a genuine run-rate, but it has not become the volume engine the strategy requires.

The offsetting evidence, and it matters: ONVO's average selling price reached roughly RMB 240,000 in Q2 2026, and Li noted the brand was one of only eight in China to grow both volume and ASP simultaneously in a market defined by discounting.¹⁸ Growing price and volume together during a price war is genuinely hard and is the strongest available counter to the "ONVO is a failure" reading.

The calibrated conclusion: the multi-brand strategy is working, but more slowly and with more friction than management projected, and the version where ONVO becomes the majority of group volume remains unproven. The relevant KPI is not ONVO's monthly number in isolation but the mix trajectory against 35-55-10 — because if ONVO stalls around a quarter of volume, NIO has added the cost and brand-dilution risk of a three-brand structure without the amortisation benefit that justified it.

Funding all three brands, of course, required capital. A great deal of it, from a shareholder base that has been repeatedly reshaped.

↑ Back to Top

VII. Ownership, Capital Structure, and Governance

In December 2023, NIO's cap table changed hands in a way that would be extraordinary at any Western automaker and barely registered outside specialist press.

CYVN Holdings — an investment vehicle backed by the government of Abu Dhabi — had already put $738.5 million of primary equity into NIO plus a $350 million secondary purchase in July 2023, roughly $1.1 billion combined. Then, on December 27, 2023, it closed a further $2.2 billion subscription: 294 million Class A ordinary shares at $7.50 each.¹² That took CYVN to approximately 20.1% of NIO and made it the company's single largest shareholder, with two board seats.¹³

An Emirati sovereign vehicle became the largest owner of a Chinese premium automaker. It is a small, sharp illustration of how capital flows have rearranged themselves in the last decade — and of how few sources of patient, multi-billion-dollar, loss-tolerant equity remain for a company like NIO.

Who actually controls the company

Here is the part that matters for anyone assessing whose incentives drive capital allocation. CYVN owns roughly a fifth of NIO and cannot direct it.

NIO operates a dual-class structure. Class A ordinary shares carry one vote each. Class C ordinary shares carry eight votes each, and per the FY2025 annual report, William Li together with his affiliates beneficially owns all of the issued Class C shares.⁴ The result is that Li holds roughly 10.5% of the economics and approximately 44.2% of the voting power. CYVN's stake, held entirely as Class A shares with one vote apiece, does not come close.¹³

Founder super-voting structures are common in US-listed Chinese technology companies and are not inherently disqualifying — they can protect long-horizon strategy from short-horizon pressure, which for a company building a decade-long infrastructure bet is a coherent argument. But the specific configuration here deserves clear eyes. An outside investor who wrote a $3.3 billion cheque across three tranches, took two board seats, and became the largest economic owner still cannot force a change of strategy, a change of CEO, or a change in capital allocation if things go wrong. Minority shareholders have even less. That is a governance fact to price, not a scandal to allege — but it does mean the entire investment case rests on continuing to trust one person's judgment, and that trust must be earned by results rather than assumed.

The dilution ledger

NIO has been a persistent consumer of capital markets. A $1.5 billion convertible note issue in January 2021. Further convertible issuance in 2024. And in March–April 2025, an upsized Hong Kong share placement raising HK$4.03 billion, roughly $518 million.¹⁴ That placement came after management had spent two years telling investors profitability was imminent, and it priced at a discount to the market — the standard mechanics of an equity raise by a company that needs the money more than the buyers need the shares.

The pattern is straightforward and should be stated without euphemism: for its entire public life, NIO has funded losses by issuing claims on itself. Every existing shareholder's stake has been repeatedly diluted, and the improvement in margins that began in 2023 has only recently made that pattern look like it might end. It has not yet definitively ended — with GAAP net losses still being reported in the second quarter of 2026, the company remains one bad demand cycle away from needing the market again.

The short-seller episode

In June 2022, Grizzly Research published a report alleging that NIO used related party Wuhan Weineng to inflate revenue by RMB 2.6 billion and net profit by RMB 1.8 billion — essentially claiming that batteries sold into the BaaS vehicle were being recognised as sales in a way that overstated NIO's economics.¹⁶ NIO rejected the allegations and formed an independent committee of its board, which by August 2022 concluded the allegations were "not substantiated."¹⁵

That is the record, and it should be characterised accurately. This was a company-commissioned internal review conducted by directors of the company being examined. It was not a finding by the SEC, by a Chinese regulator, or by an independent forensic auditor engaged by a third party. No regulator has to the author's knowledge brought an action on these allegations in the years since, and the company's auditors have continued to issue opinions on its financial statements. A reasonable investor should treat the matter as unresolved-but-unrefreshed rather than affirmatively cleared, and should note that the underlying structural feature that made the allegation possible — a battery-asset JV that is a related party, is not consolidated in the way an outsider might assume, and holds the assets central to NIO's business model — still exists.

Adjacent to this sits NIO Capital, a NEV and energy-technology venture fund co-founded by Li in 2016. It is a legally separate entity with outside limited partners, and there is no public evidence of improper dealing. It belongs in the analysis as a related-party optic and a modest claim on founder attention, not as an allegation.

An activist looking at NIO would find the raw material for a campaign here — a controlled company with a below-market economic stake held by the controller, a persistently diluted float, an unconsolidated related-party vehicle holding the crown-jewel assets, and no disclosed segment economics for the business unit that defines the strategy. What the activist would not find is a lever to pull, because of the Class C shares. That combination — real questions, no mechanism — is the governance summary.

Governance questions matter most when management's word is the primary evidence for a claim. Which brings us to the five years in which NIO's word was, repeatedly, wrong.

↑ Back to Top

VIII. Five Years of Broken Guidance — And Then Delivery

On March 1, 2023, William Li got on a call with analysts to discuss NIO's fourth-quarter 2022 results and made a specific, checkable promise. NIO could reach quarterly break-even in the fourth quarter of 2023, he said, and guided to a fourth-quarter 2023 gross margin of 18–20%.⁶

What follows is the clearest available test of this management team's credibility, and it is worth walking through in sequence, because the shape of the pattern matters more than any single data point.

The original promise, March 2023: break-even in Q4 2023, gross margin 18–20%.⁶

The outcome: Q4 2023 vehicle margin came in at 11.9%. Full-year 2023 vehicle margin was 9.5%. Group break-even was not achieved, and the margin miss against the guided range was not a rounding error — it was roughly half the promised level.

The reset, March 2024: guidance moved to 15–18% vehicle margin for full-year 2024, with no explicit break-even quarter named. Note the retreat: the specific, dated commitment was replaced with an annual range.

The re-commitment, September 2024: management publicly reiterated confidence in Q4 2024 break-even.

The outcome, November 2024: on the third-quarter call, vehicle margin was 10.7% — below the low end of the annual range guided eight months earlier — and revenue missed guidance. The break-even language shifted again, this time to "profitable by 2026." Full-year 2024 vehicle margin finished at 12.3%, under the 15–18% range.²⁶

The reset again, March 2025: company-wide break-even by Q4 2025, alongside an 18% automotive margin goal. In the same period Li told staff that if NIO had "only one goal this year, it is to achieve profitability in the fourth quarter."¹⁷

The doubling-down, November 2025: Li declared there was "no Plan B" for the Q4 2025 target.⁷ Sell-side opinion split — CMB International skeptical, JPMorgan judging it achievable.⁷

The delivery, March 2026: NIO reported GAAP operating profit of RMB 807.3 million and non-GAAP operating profit of RMB 1,251.3 million for the fourth quarter of 2025, on vehicle margin of 18.1% and net profit of RMB 282.7 million — the first quarterly profit in company history.² CFO Stanley Qu framed it precisely: non-GAAP operating profit "for the first time on a quarterly basis."²

And then it held. Q1 2026 and Q2 2026 sustained it, with Q2 2026 delivering non-GAAP operating profit of RMB 206.9 million on vehicle margin of 18.5%, deliveries up 49.4% year over year, and cash reserves of RMB 56.7 billion — up from RMB 45.9 billion at the end of 2025.¹²

How to weigh this

There is a lazy bull reading and a lazy bear reading, and both are wrong.

The lazy bear reading is that management lied for five years and therefore cannot be trusted. But look at what happened underneath the missed dates: vehicle margin improved in every single year of the sequence — 9.5%, 12.3%, 14.6%, 18.5% — through the misses, through the price war, through the layoffs. The direction of travel was correct the entire time. Management was wrong about when, not about whether. And the destination was eventually reached and then held for three quarters, which is more than most of the sell-side expected in November 2025.

The lazy bull reading is that the promise was kept and the credibility question is closed. It is not. Five years of specific, dated, publicly reiterated break-even commitments, missed repeatedly, is not a single miss explicable by macro noise. It is a pattern in how this management team forecasts: consistently, materially over-optimistic on timing, and willing to re-commit to a new date immediately after missing the last one. A "no Plan B" declaration in November 2025 reads differently when you remember it was the fourth such declaration.

Nor did the pattern stop at profitability. NIO set a 2025 delivery target of 440,000 units and delivered 326,028 — 74% of target.¹⁹ That was a better completion rate than Li Auto's 63% or HIMA's 59%, but worse than XPeng, Leapmotor, and Xiaomi, all of which exceeded their targets.¹⁹ The forecasting bias is systemic, not confined to the P&L.

The fair conclusion is a narrowing, not a rejection. The historical record does not reject the claim that NIO can execute a margin-led turnaround — it demonstrably did. It does reject the claim that management's stated timelines should be taken at face value, and it means that any forward guidance from this team deserves a discount and a longer assumed lag. Investors should price both facts, not just the one they prefer.

The confirming-or-falsifying event to watch is specific: whether vehicle margin holds at or above 18% through the third and fourth quarters of 2026 as China's price war reintensifies. There is already a warning light. On the Q2 2026 call, CFO Qu disclosed that material costs had risen by an average of RMB 14,000 per car since late in the fourth quarter of 2025, with another RMB 2,000–3,000 expected in the second half.¹⁸ That is a meaningful headwind against an 18.5% margin, and it sits directly beside the margin achievement rather than in a distant risk paragraph, because that is where it belongs.

One coda that illuminates how this team operates under pressure. In November 2023, as reserves fell and financing costs rose, Li announced in an internal letter that NIO would cut roughly 10% of positions, hitting battery and smartphone teams — the NIO Phone project — hardest. "Duplicate" and "inefficient" roles would go, and projects that would not contribute to financial performance within three years would be deferred or cut.²⁸ Li told employees the coming two years would bring "the most intense competition during the transformation of the automotive industry," and compared the journey to "a marathon on a muddy track."²⁸ The announcement came days before the CYVN capital injection closed. Cut costs visibly, then raise capital: whether you read that as good sequencing or as optics, it tells you this is a management team that thinks carefully about what the market sees and when.

He was right about the competition, at least.

↑ Back to Top

IX. Competitive Landscape: China's EV Bloodbath

In April 2026, China's new-energy vehicle retail penetration hit 61.4% — meaning that in the world's largest car market, roughly three of every five vehicles sold at retail were electric or plug-in hybrid. That is a technology transition running years ahead of any Western forecast made in 2020.

And it has been a catastrophe for the people who built it.

China's automotive sector profit margin sank to a record low of 3.2% in the first quarter of 2026 and stood at 3.8% for the first half, with industry profit down about 20% year over year.²⁹ In the first half of 2026, only three brands — BYD, Xiaomi, and Leapmotor — were profitable.²⁰ The sector's own trade body cut its 2026 outlook to an 11% decline.²⁰ This is not a growth market any more. It is a consolidation-phase market, and the economics of those two things could not be more different: in a growth market, everyone's mistakes get bailed out by volume; in a consolidation market, the marginal producer dies and the survivors buy their share cheap.

Let us war-game NIO's position against each threat vector.

The scale problem: BYD

比亚迪 BYD sold 4,602,436 new-energy vehicles in 2025 — roughly fourteen times NIO's group volume.¹⁹ More importantly, BYD makes its own batteries, its own semiconductors, and a large share of its own components. Vertical integration at that scale produces a cost position NIO structurally cannot match, and it gives BYD the ability to initiate price cuts that competitors must absorb out of margin.

The nuance worth noting: BYD itself missed its 2025 target of 5.5 million units, hitting 83.7%, and posted its first monthly sales decline of the year.¹⁹ Even the giant is feeling the market. But a giant with a cost advantage that slows down is still the price-setter for everyone below it.

The direct rivals: a reshuffled premium field

Among the original "new forces" cohort, the standings have inverted in ways that are uncomfortable for NIO.

理想汽车 Li Auto sold 406,343 units in 2025 against a 640,000 target — a 63.5% completion rate.¹⁹ The company that pioneered range-extended EVs in China and got to profitability first has been losing ground. 小鹏汽车 XPeng went the other way entirely: 429,445 units in 2025 against a 350,000 target, a 22.7% overshoot and an all-time high, and in October 2025 XPeng overtook Li Auto in market capitalisation.¹⁹²³ XPeng executed a genuine turnaround from a near-death position in 2023 — the same trough NIO was in — and did it faster.

That comparison is uncomfortable and instructive. Two companies with similar starting points, similar capital constraints, and similar timeframes; one delivered a volume-and-narrative recovery two years before the other. NIO's turnaround is real, but it is not exceptional within its own peer group, and any bull case built on "nobody else could have done this" does not survive contact with XPeng's record.

Then there is 小米 Xiaomi. A consumer electronics company with no prior automotive experience launched the SU7 sedan in 2024, sold more than 400,000 vehicles in 2025 against a 350,000 target, crossed 500,000 cumulative deliveries by December 2025, and became one of only three profitable NEV makers in China.¹⁹²⁰ The SU7 attacks precisely NIO's ET5 and ET7 sedan positioning.

The structural threat: Huawei's asset-light model

The most analytically interesting competitor does not build cars at all.

华为 Huawei's HIMA ecosystem — spanning AITO/赛力斯 Seres, Luxeed with Chery, Stelato with BAIC, Maextro with JAC, and Shangjie with SAIC — sold 589,107 vehicles in 2025, with the AITO brand alone topping 420,000 and leading China's luxury brand rankings.¹⁹²² Huawei supplies the smart cockpit, the driver-assistance stack, the brand halo, and increasingly the retail channel through its own stores. It owns no factories.

This is the sharpest possible contrast with NIO's philosophy. NIO built everything: the platform, the chips, the operating system, the retail houses, and a national energy network of physical stations. Huawei licenses intellectual property and lends its brand, and is outselling NIO's premium positioning by a factor of nearly two, in the same segment, at a fraction of the capital intensity. If the market rewards capital efficiency over vertical control — and in a 3% margin industry, it eventually must — that is a structural problem for NIO's entire thesis, not a cyclical one.

HIMA missed its own one-million-unit target badly, achieving 58.9%.¹⁹ Overreach is universal in this market. But missing an ambitious target with almost no capital deployed is a very different financial event than missing one after building 4,000 stations.

The cheap-and-profitable path: Leapmotor

零跑汽车 Leapmotor sold 596,555 units in 2025 against a 500,000 target — 119% completion — and reached its first annual profit after a substantial prior-year loss, remaining profitable through the first half of 2026.¹⁹²⁰ Roughly 20%-owned by Stellantis, Leapmotor took a lower-cost, less differentiated route to the profitability NIO spent years longer reaching. That does not invalidate NIO's approach, but it does establish a base rate: in this market, profitability was achievable without a battery-swap network, without NIO Houses, and without a decade of losses.

Supplier power

CATL controls roughly 43% of China's battery market.²¹ It is NIO's largest supplier, a co-owner of NIO's battery-asset vehicle, an investor in NIO Power, and the operator of a competing swap standard. Few automakers anywhere have a supplier so deeply entangled in their strategic core.

Five Forces, honestly applied

Rivalry: extreme. Sub-4% sector margins with more than 30 NEV-focused manufacturers and only three of them profitable is the textbook definition of destructive competition.²⁰

Buyer power: high. Chinese consumers are price-sensitive and switch brands readily; BaaS lock-in helps NIO at the margin but does not change the industry condition.

Substitutes: rising. BYD and Li Auto's plug-in hybrid and range-extender offerings are a hedge against pure-EV charging anxiety — and they hedge against the very problem NIO's swap network exists to solve, which is an uncomfortable substitution risk for the moat.

Supplier power: concentrated, as above.

Barriers to entry: here the received wisdom failed. Everyone assumed capital intensity would deter entrants. Instead a phone maker and an IP licensor with no factories both entered successfully within three years and are outselling NIO. The barrier that NIO's capital spending was supposed to build turned out to protect against the wrong kind of competitor.

Against that backdrop, an overseas growth story would be worth a great deal. NIO has one. It is not going well.

↑ Back to Top

X. Europe and Global Ambitions: A Stalled Bet

In 2021, NIO opened a NIO House in Oslo. It was a genuinely bold move — the first serious attempt by a Chinese premium EV brand to sell directly to European consumers on brand rather than price, in the country with the highest EV adoption rate on earth. Norway was chosen deliberately: no import tariffs on EVs, a wealthy and technologically curious buyer base, and a government that had already done the work of making electric ownership normal. NIO followed with Germany, the Netherlands, Sweden, Denmark, Belgium, and Portugal, and built a small network of swap stations to support them.

Five years later, the ambition is on ice, and the evidence for that is not analyst speculation — it is management's own operational decisions.

Swap-station construction in Europe slowed to a crawl: nine new stations across nine months of 2024, versus twenty in eight months of 2023. The UK country manager departed in 2024 before the market ever properly launched. The Hamburg flagship store closed. Denmark was shifted to a distributor model, which is what a company does when it no longer wants to fund its own retail presence. And NIO told European staff there would be no model updates until late 2027 and no further swap stations.²⁴

That last item is the one that settles the question. A carmaker that tells its own European organisation there will be no new product for more than two years has, functionally, stopped competing in that market. In an industry where model cycles run three to four years and Chinese domestic refresh cadence is faster still, a two-and-a-half-year product freeze in the world's second-largest premium car market is not a pause. It is a withdrawal that has not been announced.

The scoreboard confirms it. NIO's 2025 European registrations totalled roughly 1,100–1,300 units across all markets — a rounding error against 326,028 group deliveries.²⁵ Germany, the largest of those markets, fell to 325 units from 398 the year before, and NIO-brand registrations across Norway, the Netherlands, Germany, Sweden, and Denmark fell 31% year over year.²⁵ Firefly's European launch on August 14, 2025, beginning with Norway and the Netherlands, was the one genuinely new development, and by year-end only around 130 Firefly units had been registered across Europe.²⁵

Why did it fail? Several reasons compound. NIO's entire value proposition rests on the swap network, and a swap network with a handful of stations in a country is not a value proposition — it is a liability, since a NIO in Europe is a car whose defining feature barely exists. European buyers had no brand familiarity to draw on and NIO had no volume to fund building it. The EU's countervailing duties on China-built EVs from late 2024 raised landed costs. And crucially, the domestic price war made every euro of European marketing spend an opportunity cost against defending share at home.

The honest read: this should be treated as substantially deferred optionality, not abandoned optionality — NIO retains the entities, the homologations, the dealer relationships, and now a product in Firefly that is genuinely designed for European roads and European garages. But it should carry close to zero weight in any current assessment of the business. Optionality that management has explicitly defunded for two-plus years is not optionality an investor is being asked to pay for; it is a call option with a long-dated expiry and no premium being spent on it.

There is a broader lesson buried in the European retreat, and it applies well beyond NIO.

↑ Back to Top

XI. Business & Investing Lessons

Capital-intensive differentiation is a double-edged sword, and the second edge takes a decade to appear.

The battery-swap network is simultaneously NIO's most distinctive asset and its heaviest anchor. It gave the company a genuine reason to exist in a market where two hundred EV startups did not, and it gave owners a daily experience no rival could match. It also required a provincial government bailout to fund in its early years and a second, distributed handoff to state-owned enterprises a decade later to sustain. The general principle: when differentiation requires permanent capital rather than accumulating advantage, check whether the asset compounds or merely persists. A brand compounds. A software platform compounds. Four thousand physical stations that must be maintained, upgraded through five hardware generations, and stocked with depreciating batteries do not compound — they must be continuously refinanced. The 2026 asset transfers are what that distinction looks like when it finally shows up on a balance sheet.

Guidance discipline compounds too — in the wrong direction.

A single missed quarter is noise. Five consecutive years of specific, dated, publicly reiterated break-even targets, each missed and immediately replaced with a new one, is a data set. The instructive part is that NIO's underlying trend was genuinely improving the entire time — margins rose every year through every miss. Management was directionally right and repeatedly wrong about timing, and it cost them credibility they then had to buy back with three consecutive profitable quarters. The lesson for investors is to separate the two questions completely: is the trend real and is the timeline reliable. At NIO the answers were yes and no, and conflating them produced expensive mistakes in both directions — for bears who dismissed the margin trend, and for bulls who bought each new break-even date.

Moving down-market is a different business, not a bigger one.

ONVO had NIO's engineering, NIO's supply chain, NIO's charging infrastructure, and NIO's balance sheet — every advantage a new brand could want — and still fell from 10,000 monthly deliveries to 4,049 in two months and cost its president his job. Selling a RMB 250,000 family SUV requires a different sales motion, a different service cost structure, and a different marketing language than selling a RMB 400,000 statement car. Capability does not transfer across price tiers as cleanly as org charts suggest. When a premium company announces a mass-market brand, the correct prior is that it will take twice as long as guided.

Global expansion optionality is not real until it survives a downturn.

The European retreat is the cleanest possible demonstration. When the domestic price war intensified and cash got tight, the international growth story — the one that carried real weight in valuation conversations in 2021 and 2022 — was the first thing cut. Not reduced: frozen, with a stated two-and-a-half-year product hiatus. Growth optionality that exists only while the core business is comfortable is not optionality. It is a discretionary marketing budget with a narrative attached, and investors should value it accordingly until it has survived at least one cycle of the core business needing the cash.

A related lesson on state capital. There is nothing wrong with taking government money — Hefei's investment saved a company that is now a significant employer and technology base, and both sides got what they wanted. But a business model that has required state or sovereign capital at each of its three most important junctures carries a distinct risk profile: it concentrates relationship and policy risk in a way that competitors with asset-light models do not carry, and it means the terms of future support are negotiated rather than market-determined.

↑ Back to Top

XII. Bull vs. Bear, Risk Radar, and What to Watch

The bull case, stated at its strongest

Three consecutive quarters of operating profitability through mid-2026, with vehicle margins at 18.5% — a level that puts NIO in the same neighbourhood as the peers who got there first. That improvement compounded across three fiscal years rather than appearing in one, which is the signature of structural cost improvement rather than a one-off. Deliveries grew 49.4% year over year in the second quarter of 2026 while average selling prices rose across all three brands, and NIO's average transaction price of RMB 406,000 in the quarter — over RMB 430,000 in July — ranked first among mainstream premium brands in China.¹¹⁸ Growing volume and price simultaneously during a price war is the single most bullish operating fact in this entire story, because it is the empirical signature of pricing power.

The swap network, whatever its standalone economics, is being de-capitalised through SOE partnerships without visible degradation of the customer experience so far, and fifth-generation stations cost roughly RMB 1.4 million in capex — about RMB 100,000 cheaper than the previous generation — while serving all three brands from one mechanism.¹⁸¹⁰ A maturing three-brand portfolio now spans price tiers from Firefly's compact segment to the ET9's RMB 788,000 flagship. Cash reserves of RMB 56.7 billion, up nearly RMB 11 billion in two quarters, provide real runway.¹ And management has guided to positive operating and free cash flow continuing through the third and fourth quarters of 2026.¹⁸

The bear case, stated at its strongest

The profitability turn arrived roughly two years after it was promised, and it coincided with a 44% year-over-year cut to R&D in the very quarter it was first reported.² It also arrived in an industry where sector-wide margins sit near 3–4% and only three of roughly thirty NEV makers are profitable — meaning NIO's margin gains exist at the pleasure of competitors who have shown repeated willingness to cut price.²⁰²⁹ The RMB 14,000 per-vehicle material cost increase since late 2025, with more expected, is already eating into the 18.5% figure from the input side.¹⁸

The multi-brand strategy is behind plan: on Q2 2026 volumes the mix sat near 57-27-16 against a 35-55-10 target, meaning ONVO is carrying roughly half the share the strategy requires, and management itself describes ONVO's brand awareness as five to six years behind the NIO brand's.¹⁸¹ Europe is functionally frozen. GAAP profitability has still not been achieved on a quarterly basis. And the company has twice needed state or sovereign capital injections and is now executing a third, distributed version through provincial SOEs — a dependency that concentrates policy and relationship risk in ways an asset-light rival like Huawei's HIMA simply does not carry.

Through Helmer's 7 Powers

Scale economies: weak. At 326,028 units against BYD's 4.6 million, NIO is subscale in the dimension that matters most in a commodity-ising market.¹⁹²

Network economies: the most interesting case, and unproven. The swap network should generate them — more stations make each NIO more valuable, more NIOs make each station more economic. The alliance with seven other automakers was the attempt to make this a true industry network. Three years on, no meaningful third-party volume has been disclosed and access fees remain "under discussion."¹⁸ Latent, not realised.

Counter-positioning: this was the original power, and it has partially inverted. NIO counter-positioned against Tesla's charging model with something Tesla could not copy without abandoning its architecture. But Huawei then counter-positioned against NIO with an asset-light licensing model NIO cannot copy without abandoning its architecture. The disruptor got disrupted on capital intensity.

Switching costs: real and underrated. BaaS subscribers face genuine friction leaving, and the community and service ecosystem adds more. This is probably NIO's most durable power, and it is not the one the company markets.

Branding: real within a niche. Achieving the highest average transaction price among mainstream premium brands in China is a meaningful brand result and the best single piece of evidence for the original thesis.¹⁸ But brand strength has not translated into the volume required to amortise the cost base, and ONVO's struggle shows the brand does not stretch downward.

Cornered resource: the swap patent estate and station real estate qualify partially, but NIO is progressively transferring the real estate to third parties, which trades the cornered resource for capital relief.

Process power: the strongest evidence here is the margin trajectory itself. Improving vehicle margin from 9.5% to 18.5% across three years while growing volume is a manufacturing and procurement achievement that requires accumulated organisational capability. This is the power NIO has most clearly built rather than bought.

Risk radar

Demand and price war. The mechanism is direct: NIO's 18.5% margin was achieved during a period when its own mix improved. Further price cuts led by BYD or Xiaomi would force NIO to choose between volume and margin, and the material-cost increase already disclosed means it enters that choice with less cushion than the headline suggests.

Refinancing and dilution. GAAP losses persist. The history of convertible notes and discounted placements means any relapse would most plausibly be financed by issuing equity again, and NIO's share count history shows the company has never hesitated to do so.

Execution. ONVO demonstrated that multi-brand scaling is not automatic; Firefly's international rollout carries the same category of risk, compounded by the European organisation being told there is no new product coming.

Governance and control. Li's roughly 44.2% voting power against roughly 10.5% economic ownership means that if strategy falters, minority shareholders — including a sovereign investor holding 20.1% — have no practical mechanism to force a change.⁴¹³

Geopolitical and listing structure. NIO carries a triple listing — NYSE primary, HKEX under 9866.HK, and a Singapore secondary. That structure was built partly as insurance against US delisting risk for Chinese ADRs, and it means headline risk can move the stock without any company-specific trigger. On the other side, EU trade measures on China-built EVs constrain the European path independent of anything NIO does.

Related-party and disclosure. The unconsolidated battery-asset vehicle, the absence of standalone NIO Power segment economics, and the unrefreshed 2022 short-seller allegations together mean an investor is asked to take the network's economics substantially on trust.

The three KPIs that matter

Everything above reduces to three things worth tracking. Not calculating — tracking, quarter by quarter.

1. Vehicle margin through the next price-war cycle. Does 18%-plus hold through the second half of 2026 and into 2027, or was the fourth quarter of 2025 through the second quarter of 2026 a temporary peak assembled from favourable mix and a lull in discounting? Given the disclosed material-cost headwind, this is the single most informative number NIO reports.

2. Brand mix versus the 35-55-10 target. Specifically ONVO's share of group volume. If ONVO climbs toward half of deliveries, the multi-brand strategy is working and the network amortisation thesis is intact. If it plateaus near a quarter, NIO has bought three-brand complexity without the volume that justified it.

3. NIO Power's standalone economics and the pace of SOE asset transfers. Whether the company ever discloses a segment-level operating result for the swap network, and how quickly stations move to state ownership without availability deteriorating. This is the clearest available test of whether the defining bet of NIO's existence becomes a genuine asset or remains a subsidised cost centre.

↑ Back to Top

XIII. Epilogue

Where does NIO stand in September 2026?

It is a company that spent a decade and well over RMB 100 billion in cumulative losses proving a contrarian proposition about how electric cars should be refuelled. It nearly died in 2019 and was saved by a provincial government that wanted an industrial anchor tenant. It nearly stalled again in 2023 and was recapitalised by an Emirati sovereign fund that got a fifth of the equity and almost none of the control. It promised profitability five times and delivered on the sixth, roughly two years late, and has now held it for three quarters — an achievement that fewer than one in ten of its domestic competitors can claim.

It has also, in the same year it finally turned a profit, begun giving away the physical asset that made it distinctive, freezing the international ambition that once carried a meaningful share of its valuation narrative, and cutting the research budget that determines what it will sell in 2029.

Those are not contradictions. They are the same decision viewed from different angles: a company that has stopped trying to win every argument at once and started trying to survive profitably. Whether that is maturity or retreat depends almost entirely on what happens next.

The open question for a long-term investor is not whether NIO can be profitable for a quarter or three. It has now demonstrated that it can, and the margin arc underneath it — three straight years of improvement, achieved during the worst pricing environment in the industry's history — is the most credible thing in the entire story. The question is whether that profitability survives the next leg of China's price war without another round of state-capital dependency, without brand-diluting product missteps in a three-brand portfolio that is already behind plan, and without the stalled global ambition quietly becoming permanent.

The battery you never own is, in the end, an apt metaphor for the whole enterprise. NIO built something genuinely novel, at enormous cost, and then discovered that the most sustainable version of it was one where somebody else held the asset and NIO simply ran it. That may turn out to be the smartest capital-allocation decision the company ever made. It is also, unmistakably, a smaller claim than the one it started with.

↑ Back to Top

XIV. Outro

The 4,000th station in Quanzhou is still swapping batteries. The robot doesn't know who owns it.

NIO Inc.

Table of Contents

NIO: Betting the Company on a Battery You Never Own

I. Introduction & Episode Roadmap

II. Origins: William Li's Vision and the IPO Gamble (2014–2018)

III. The Near-Death Experience and the Hefei Bailout (2019–2020)

IV. NIO Power: The Battery-Swap Bet — Moat or Cost Center?

The financing structure tells you who really owns it

Break-even keeps moving

The perk that didn't survive the unit economics

Handing the asset to the state

The alliance that hasn't converted

The calibrated conclusion

V. The Core Business: Lineup, Segments, and Margin Trajectory

The margin arc is the actual story

The R&D question

VI. The Multi-Brand Gamble: ONVO and Firefly

The ONVO stumble, examined

Firefly, and what actually happened to the mix

VII. Ownership, Capital Structure, and Governance

Who actually controls the company

The dilution ledger

The short-seller episode

VIII. Five Years of Broken Guidance — And Then Delivery

How to weigh this

IX. Competitive Landscape: China's EV Bloodbath

The scale problem: BYD

The direct rivals: a reshuffled premium field

The structural threat: Huawei's asset-light model

The cheap-and-profitable path: Leapmotor

Supplier power

Five Forces, honestly applied

X. Europe and Global Ambitions: A Stalled Bet

XI. Business & Investing Lessons

XII. Bull vs. Bear, Risk Radar, and What to Watch

The bull case, stated at its strongest

The bear case, stated at its strongest

Through Helmer's 7 Powers

Risk radar

The three KPIs that matter

XIII. Epilogue

XIV. Outro

References