BYD Company Limited

The thesis, in five lines

BYD — a great business, cheap to a bond buyer, short of a 15% owner’s return

What it is: the world’s largest maker of new-energy vehicles and their batteries, vertically integrated from the cell up — a proven, extraordinary operating company. The verdict: PASS — not a bad business, a return the price does not pay. The one number: discounted at a ~3% long-term Chinese government-bond rate with a 14× terminal cash-flow cap (the house durability convention, ruling #40), the whole-company fair value is about RMB 800–950bn (midpoint ~875bn) against today’s ~RMB 698bn market cap — so the stock trades at ~0.75–0.85× fair value, modestly cheap to a risk-free-return buyer; yet the owner return actually on offer at that price is only about 10–12% a year, short of the 15% we require, with no margin of safety. The one operating metric to watch: gross profit per vehicle × volume — where the battery-cost moat becomes cash; it has just turned down. The open question: not market position (settled) but whether a return on capital near eleven per cent, reinvested almost in full through a state-managed price war, ever clears a 15% owner’s return. How to read this: four parts — how capable the company is, the destination where the value is made, the industry and the capital cycle that set the returns, and the verdict that derives from them.

Evidence tiers, used throughout: FACT (primary filing) · MANAGEMENT CLAIM · THIRD-PARTY INTERPRETATION · ANALYST INFERENCE. Forward figures are analyst inference; source links open in a new tab.

The company

Business Overview

BYD Company Limited is a vertically integrated maker of electric vehicles, the batteries inside them, and the components around them, and it is today the largest manufacturer of new-energy vehicles in the world. In the year to December 2025 it earned revenue of about Renminbi 804 billion and net profit attributable to shareholders of about Renminbi 32.6 billion, having grown vehicle volume roughly 155 per cent across the three prior years. Its distinguishing feature — the fact that decides almost everything else in this report — is that it makes its own battery cells, and the battery is the single most expensive part of an electric car.

The central tension of the name is simple to state. The destination — that BYD is, and will remain, a dominant global electric-vehicle maker — is close to settled. What is genuinely unsettled is the economics of that destination: whether a company that competes by cutting prices, in a Chinese market with roughly twice the capacity it needs and a government now actively managing the price war, can earn a return above its cost of capital through the cycle. The first post-2025 data points make that question sharper, not settled: first-quarter 2026 profit fell 55 per cent year on year, but roughly four-fifths of that fall was a currency swing, gross margin actually held above the full-year 2025 level, and unit volume — down about 16 per cent across the half — had returned to year-on-year growth by June on record exports.

How the money is actually made.

BYD’s model is vertical integration carried to an unusual extreme: it designs and builds its own battery cells, packs, electric motors, power electronics and a meaningful share of its own semiconductors, and assembles the complete vehicle. The economic logic is that in an electric car the profit pools in the cell layer — the independent cell maker Contemporary Amperex Technology Co. Limited earns a battery gross margin near 26 per cent and a return on equity near 25 per cent, well above the fifteen-to-twenty per cent typical of vehicle assembly. Because BYD owns both layers, it captures cell-layer economics a conventional carmaker must pay away.

Revenue is roughly 80 per cent automobiles and related products, with a large listed contract-electronics subsidiary, BYD Electronic (International) Company Limited (Hong Kong 285), alongside energy storage, third-party batteries and, historically, urban rail and semiconductors. The mix is now strongly and deliberately global — overseas sales reached roughly 39 per cent of revenue — and, crucially, overseas is run as a premium business, the mirror image of the domestic price war.

Read on the thesis: a genuinely integrated manufacturer — the capability is real; the question is what it earns.

The record

History

BYD was incorporated on 10 February 1995 (the founding facts are taken from the primary 2002 Hong Kong listing prospectus). It began not as a car company but as a maker of rechargeable batteries, competing against entrenched Japanese producers by replacing automation with a labour-intensive, self-built-equipment process that drove cost low enough to win the mobile-handset accounts of Motorola, Ericsson, Philips, Kyocera and Matsushita. The founder Wang Chuanfu held roughly 28.9 per cent and his cousin Lü Xiangyang roughly 20.7 per cent at the initial public offering.

In 2003 the company bought the Qinchuan automobile works and announced it would build cars — a move that was off-prospectus and sent the shares down sharply. Yet the battery-to-car endgame was, on the investor Li Lu’s firsthand account, in Wang Chuanfu’s mind from the beginning. The company’s character was tested twice: a near-death “three-year adjustment period” in 2010–12, and a trough in 2019. Berkshire Hathaway’s MidAmerican unit took a stake in 2008 — introduced by Li Lu to Charlie Munger and Warren Buffett — and held through a share-price round trip of roughly tenfold up then some eighty per cent down without selling a share. The modern arc followed: the Blade battery in 2020, the exit from pure internal-combustion vehicles in March 2022 (the first global carmaker to do so), and arrival by 2025 at the destination Wang Chuanfu had dated back in 2008.

Read on the thesis: thirty years of hard technology bets delivered on schedule — the record that earns the founder’s forward claims their weight.

The numbers

Financial highlights, and the statements in full

The load-bearing pattern: an extraordinary operating company whose returns, not whose market position, are the open question. The battery cost advantage is real and independently attested — Ford Motor Company’s chief executive stated publicly that BYD’s batteries are “about 30 per cent cheaper” than the merchant price and that Ford “can’t beat BYD’s vertical integration.” Set against that: deeply negative free cash flow, a group return on invested capital near eleven per cent — above a cost of capital near ten per cent, but well below the fifteen per cent return we require — a state-managed price war at home, and a first-quarter balance sheet leaning more on short-term funding.

Revenue rose roughly eightfold over the decade; net profit attributable to shareholders peaked at RMB 40.3 billion in FY2024 and fell about 19 per cent to RMB 32.6 billion in FY2025 — the first profit decline of the modern growth phase, as the price war reached BYD too.

The financial statements in full.

Figures are stamped to primary filings (the FY2025 annual report, the Q1 2026 report filed 28 April 2026, and the June 2026 production-and-sales flash), in Renminbi. For the first quarter of 2026, revenue was 150.23 billion (down 11.82 per cent), operating profit 4.70 billion, net profit attributable 4.08 billion (down 55.38 per cent), and gross margin 18.81 per cent — notably above the full-year 2025 level. About Renminbi 4 billion of the roughly 5.1 billion profit fall was a swing in financial expense from a foreign-exchange gain to a loss — a non-operating item, so the underlying operating deterioration is milder than the headline. Operating cash flow was 2.79 billion and free cash flow about negative 19 billion after roughly 22 billion of capital expenditure. First-half volume was down 15.72 per cent, but June single-month sales were up about 5.5 per cent year on year on record exports, of which overseas was roughly forty-four per cent of the month’s units.

The one number to watch — gross profit per vehicle × volume. If you could keep only one operating figure for BYD, keep this one: it sits exactly where the battery-cost moat turns into cash, it is a driver rather than a scoreboard (net profit and earnings-per-share are both distorted — the latter by the 2025 three-for-one bonus issue), and it is hard for management to flatter because gross margin and unit volume are both audited, mandated disclosures. It also turns before reported profit does. BYD’s automotive gross profit was about RMB 111 billion in 2023, roughly 138 billion in 2024, and about 133 billion in 2025 — the first decline — which works out to gross profit per vehicle of roughly RMB 29,000 on the automobile-and-related base, or about RMB 24,700 on vehicle sales alone (the parts revenue lifts the first figure). Per-vehicle gross profit has fallen about a fifth in two years as prices dropped faster than costs; total automotive gross profit still grew until 2025 only because volume rose faster. The trigger: the thesis weakens when gross profit per vehicle keeps falling and volume stops growing — a condition that is now close, with first-half-2026 volume down 15.7 per cent. This is the single number the annual re-check grades first. [Ruling #27; figures derived from BYD segment and vehicle disclosures.]

Read on the thesis: extraordinary growth, now meeting the price war — the first profit decline of the modern phase, and free cash flow deeply negative; the one metric (gross profit per vehicle × volume) has just turned down and its watch-trigger is near-fired.

The moat

Economic Moats

The moat is vertical integration into the battery cell, and it is now triangulated from three independent vantage points: from inside (the battery chief He Long), from a hostile witness (Ford’s Jim Farley — batteries roughly 30 per cent cheaper than the merchant price, integration Ford cannot match head-on, legacy processes “twenty-five years uncompetitive”), and from a disinterested engineer (the teardown firm Munro & Associates, which confirmed real cost innovation in the pack while finding the edge concentrated in the electrochemistry, not the whole vehicle). Three vantages agreeing is as close to proof as an outside analyst gets.

The size of that edge can be put in numbers. The independent cell leader, Contemporary Amperex, earns a gross margin near 26 per cent, so a carmaker that buys its cells pays roughly one divided by one-minus-0.26 — about 1.35 times — what BYD pays to make the equivalent cell itself. BYD’s cells are therefore on the order of a quarter to a third cheaper, close to the “about 30 per cent” Ford’s chief executive described from the buyer’s side. Because the battery is roughly 30 to 40 per cent of an electric car’s cost, that cell-margin capture alone is estimated to give BYD an 8 to 12 per cent lower total vehicle cost than a rival that buys its cells — an enormous gap in a market where group gross margins sit in the mid-teens and net margins are thin or negative, and the clearest single reason BYD earned a profit in the first quarter of 2026 while the sub-scale pure-plays lost money. Two honest limits: large rivals negotiate below list price and several are now building their own cells, so the gap can narrow (whether they close it is the test); and the captured margin is partly the return on the heavy capital BYD sinks into cell capacity, capital that carries its own cycle risk.

A competitor independently attributes the same edge to integration. On its own 2022 full-year results call, Geely Automobile’s management engaged with the framing that “cost advantage is the core for competition for Tesla and BYD, the focus on integrated solutions or advanced manufacturing” — naming in-house, vertically integrated production as the decisive cost weapon in the market. On its 2023 results call the same management observed that a maker with its own powertrain can earn money on new-energy vehicles where a maker of only battery-electric cars struggles — the reason a vertically integrated producer keeps a cost-and-profit advantage over rivals that buy the same parts on the open market. These are the words of a rival, not of BYD, and are quoted only as third-party corroboration of the input-cost gap the numbers above already show; figures spoken on such calls are management claims until reconciled to a filing. (Source: Geely Automobile 2022 and 2023 results calls, transcripts held in the research corpus.)

Is that still true in 2026? Partly — and the part that is fading matters. The 2023 claim has two layers. The first — that scale plus owning the powertrain and the battery is what makes an electric-vehicle business profitable, while sub-scale makers of only pure battery-electric cars lose money — still holds, and it was never really about hybrids: it is about scale and vertical integration, which is BYD’s battery cost edge described above. The second layer — that the plug-in hybrid is the safe profit haven and the pure battery-electric car is structurally disadvantaged — is decaying fast. The reasons hybrids won in 2023 (short electric range, slow charging, an expensive big battery) are being erased: battery prices fell from about US$94 to US$84 per kilowatt-hour in a year; BYD’s “super” platform flash-charges about 400 kilometres in five minutes and Contemporary Amperex’s newest lithium-iron-phosphate cell recharges in about six. Demand has rotated back to match: in China the pure battery-electric share of new-energy vehicles rose from about 58% in 2024 to about 65% in 2025 while plug-in hybrids began falling in monthly sales from September 2025, and pure battery-electric cars dominate outside China. The 2026 Iran war’s oil-price spike tilts the balance further against a car that still burns petrol. [THIRD-PARTY / ANALYST INFERENCE, dated 2026 — the forces-of-change test, ruling #31.] What it means for BYD: the durable half of the edge — battery-and-powertrain cost integration at scale — travels to pure battery-electric cars too (the flash-charging platform is itself a battery-electric technology), so the shift back to pure battery-electric does not break BYD; but the comfort that half its profit sits in a protected hybrid niche narrows, and BYD’s future profit increasingly rides on winning the pure battery-electric cost war, where the battery moat is the decider.

Why this peer set? These are not simply the highest-volume sellers. They are BYD’s comparable listed competitors for which every figure can be taken from the company’s own primary filing (rulings #22 and #23), chosen to span the whole range rather than a flattering subset: the profitable scale leaders (BYD, Geely, Tesla), the loss-making pure-plays (XPeng, Li Auto, Leapmotor), a well-funded new entrant (Xiaomi’s car unit) and a niche outperformer (Seres). Tesla is the global benchmark and former margin leader; Contemporary Amperex, the battery-cell layer, is shown apart because a supplier’s margin is not comparable to a carmaker’s. The set narrows on some charts (research spending, for one) to the makers that disclose that item on a comparable basis.

The decisive test is to put BYD’s gross margin next to the peer set on one metric over one period — company-level (group) gross margin, first quarter 2025 versus first quarter 2026, each figure the company’s own reported number — so the columns are genuinely comparable. Ordered by first-quarter-2026 gross margin, highest first:

One metric (group gross margin = total revenue minus cost of sales), one period, each figure from the company’s own first-quarter-2026 results release. Composition is not identical and is not adjusted away: BYD’s group figure is diluted by its low-margin electronics and handset-assembly arm (its automobile-only gross margin for full-year 2025 was 20.49%, not broken out by quarter); Tesla’s and XPeng’s are lifted by non-vehicle revenue (regulatory credits, energy, services) — vehicle-only gross margins in the first quarter of 2026 were Tesla 21.1% (including credits), XPeng 12.1% and Li Auto 6.1%. § Xiaomi is its smart-electric-vehicle segment (the group is profitable on phones and other hardware); the −RMB 3.1bn is that segment’s operating loss. Net results are the companies’ reported figures; BYD’s and Geely’s each absorb a large first-quarter foreign-exchange loss. Tesla is reported in US dollars; “bn” means billion. Geely’s Q1 2025 gross margin is the restated figure from its own Q1 2026 filing.

For context, not an automaker peer: Contemporary Amperex Technology (CATL), BYD’s largest battery-cell rival and a supplier to much of the industry, reported first-quarter-2026 revenue of RMB 129.1bn (up 52% year on year) and net profit of RMB 20.7bn (up 48.5%). A cell maker’s margin is not comparable to a carmaker’s vehicle margin, so it is shown here rather than in the table above.

Gross margin alone does not decide who makes money — scale does. XPeng lifted its group gross margin to 20.6% in the first quarter of 2026 and still lost RMB 1.78 billion; BYD earned RMB 4.09 billion on a lower 18.8%. The difference is volume: BYD sells on the order of a million vehicles a quarter, enough to spread fixed cost and research over a base the sub-scale challengers cannot match, so Li Auto, XPeng, Leapmotor and Xiaomi’s car unit each lost money despite respectable margins because their volume could not cover operating cost. BYD’s own group gross margin actually fell year on year (20.07% → 18.8%) as the price war reached it too — so the edge is not the highest margin (Seres at 26.2%, Tesla at 21.1% and XPeng at 20.6% all post higher group figures) but the conversion of a mid-to-high-teens margin into large absolute profit at scale, while every sub-scale pure-play bled. One honest qualifier: Geely, the other legacy-scale Chinese maker, is also profitable at a similar 17.5% margin — so much of “holding profit through the price war” is the arithmetic of scale (fixed cost spread over more than a million vehicles a quarter), which Geely shares, rather than a BYD-specific advantage.

Volume and market share over the years — is the moat widening, or being competed away? The margin table is a snapshot; a moat shows in the trajectory. Three reads — and two of the three counsel caution.

1 · BYD out-grew Tesla — but Geely is catching up. BYD passed Tesla in 2022 and pulled far ahead (4.6m vehicles in 2025 vs Tesla’s 1.64m, with Tesla now declining two years running). But the scaled domestic rival is closing: Geely nearly tripled, from 1.43m (2022) to 3.02m (2025). BYD’s lead over the Western benchmark widened; its lead over its nearest same-scale Chinese rival is narrowing.

2 · BYD’s NEV market share has plateaued. The market itself roughly seven-folded — global NEV sales rose from ~3.0m (2020) to ~20.7m (2025) — so the tide was rising for everyone. BYD’s share of it climbed steeply to ~25% in 2024, then slipped to ~22% in 2025 as the market outgrew it and rivals (Geely, Xiaomi, the broader field) scaled. A widening moat shows a rising share; a share that stalls while the market expands is the moat being competed at the edges. This is the single clearest caution in the moat case — durable, not visibly widening. The chart makes it concrete against peers: Tesla’s NEV share has roughly halved (~17% in 2020 to ~8% in 2025) as it faded, while Geely’s has climbed from low single digits toward ~7% and Xiaomi’s car unit has entered from zero — so BYD is plateauing at the top precisely as one scaled rival fades and others close from below. A share line that flattens while challengers’ lines rise is the visual signature of a moat holding its ground, not extending it.

3 · On exports, BYD is a fast-follower, not the leader. BYD’s exports exploded — ~0.24m (2023) → 0.42m (2024) → 1.05m (2025). But it was only about #6 among Chinese exporters in 2024; Chery (~1.14m), SAIC/MG (~0.93m), Changan, Geely and Great Wall all led or matched it. BYD is closing on Chery fast (from ~38% of its volume in 2024 toward rough parity in 2025), but the export surge is an industry-wide wave every scaled Chinese maker is riding — a real growth engine, not a BYD-only moat.

Read on the thesis: the trajectory confirms scale dominance over Tesla, but a converging Geely, a plateauing NEV share, and a broad export race — the moat is durable and cost-driven, not visibly widening. It reinforces “great business, competed at the edges,” not “runaway compounder.”

The same scale funds the other half of the moat: research. BYD now out-spends every rival on it — its 2025 research budget exceeds Tesla’s and runs several times the Chinese pure-plays’ — the largest programme in the field, which is what lets a cost leader keep setting the pace rather than only defend.

Is that sustainable? Absolute spend flatters the biggest company, so the sharper test is share of revenue. BYD’s record budget is about 7.9% of revenue — high, but in the normal band for a technology-driven carmaker (Tesla runs about 6–7%), and comfortably funded by scale. The loss-making pure-plays are the opposite case: their smaller programmes are a far larger share of their much smaller revenue, so it is their research that is hard to sustain, not BYD’s. The one thing to watch is BYD’s own ratio: it has climbed from under 5% to 7.9% since 2021, and a share that must keep rising just to stay ahead would be a treadmill rather than a widening moat.

Read on the thesis: the moat is real and triangulated from three sides; the open question is price, not quality.

The stewards

Capital allocation, management and integrity

BYD reinvests heroically — cumulative growth capital expenditure well over Renminbi 260 billion — and the honest read on the return is sobering. On a recomputed basis its return on invested capital is roughly eleven per cent (normalised operating profit after tax of about RMB 34 billion over invested capital of about RMB 300 billion) — above a market cost of capital near ten per cent, but well below the fifteen per cent an owner should require. The upstream cell layer earns far more; most capital goes into the lower-return assembly node in an over-supplied industry, and free cash flow was about negative Renminbi 98 billion in 2025.

The largest single reinvestment is research, and its trajectory cuts both ways. BYD’s research budget has compounded from about Renminbi 2.6 billion in 2012 to 63 billion in 2025, and its share of revenue has climbed from a low near 4.4 per cent (in 2016; it was about 5.5 per cent back in 2012) to 7.9 per cent. The rising absolute spend funds the widest research programme in the industry — a widening-moat engine. But the rising share of revenue is the warning that rides with it: on a fast-moving technology curve a leader can be forced to spend more each year merely to keep pace — an innovation treadmill rather than a durable lead. Which of the two it is turns on whether the gap to rivals keeps widening; today BYD out-spends every domestic peer several times over, but the ratio is worth watching.

On the shareholder side, discipline has been visible: buybacks, a first-ever bonus issue in 2025 to widen the register, and insider share purchases in September 2025.

The people, and how they steward the capital.

Judged on behaviour, the integrity evidence leans intact, much of it from independent or hostile vantages. The founders’ capital is unmistakably in the business (Wang Chuanfu 16.90 per cent, Lü Xiangyang 7.87 per cent, plus the family vehicle). The record: bad-year candour twice a decade apart; a seventeen-year dated promise delivered; a correct on-camera 2018 survivorship prediction; Li Lu’s account that the 2008 Berkshire deal was held so tightly the share price did not move beforehand; and Warren Buffett’s “front-page test,” to which Wang Chuanfu proactively held himself in the trough.

The named qualifications are recorded, not waved away: a euphemistic 2010 description of a dealer rupture; a habit of not marking dated targets; the gap between claiming to “resist irrational competition” while cutting prices hardest — now better explained by the dual-pricing-by-geography finding. The treatment of the roughly one-third minority in BYD Electronic (International) Company Limited is closed at primary level and leans benign: connected sales to the parent sit within the independent-shareholder-approved cap, with the chairman abstaining and the auditor’s pricing letter clean; the one diagnostic that cannot be run — captive-versus-external segment margin — is structurally unavailable because the subsidiary reports a single operating segment.

Read on the thesis: owner-operators reinvesting heavily — but into an ~11%-return assembly layer, below the hurdle, at negative free cash flow.

The destination

What BYD becomes — and whether the price already pays for it

A thirty-year owner’s return is almost entirely the destination, not the quarter — so this is the spine of the analysis, not a closing box. BYD’s size grows in every scenario; even the bear roughly doubles revenue by 2035, so the prize is real and already being captured. The whole dispersion is therefore a margin-and-return bet, not a size bet: what separates the cases is the through-cycle operating margin the business earns, and the owner return that margin throws off at today’s price — not how many cars it sells. What follows sizes that forward — each demand vector where the value is actually made, then (in Part C) the industry structure that decides whether those margins survive, then the scenarios and the value itself — and asks whether today’s RMB 698bn price already capitalises the good end-state.

Read on the thesis: everything before this establishes capability; here the value is actually made or lost.

The vectors

Where the growth comes from — and which parts actually earn

The destination has already started — three drivers are visible today. Exports rose about 71 per cent in the first half of 2026 even as total volume fell, with plants in Hungary (inside the European Union’s tariff wall), Turkey and a third European site under review; flash-charging and the second-generation Blade battery expand the addressable market from roughly half to nearly all of it; and energy storage is sized as potentially rivalling the entire car-battery market by around 2030. Each is the near-term end of one of the vectors below.

Autos (~80% of revenue) — the core, and the margin hinge. Three sub-markets behave differently: China domestic (the price war; the 2026 removal of tax relief for short-range plug-in hybrids), export (the higher-margin pool and the swing on group margin — record shipments and the #1 electric brand in Australia, Brazil and the United Kingdom, though European tariffs threaten exactly that pool), and premium (Denza, Yangwang, Fangchengbao — the answer to “Toyota, or commodity maker?”). The one deciding uncertainty for the whole company: does the battery cost advantage convert to a durable through-cycle auto operating margin, or does the capital cycle compete it away?

Energy storage — the real second leg, now sized. BYD shipped over 60 GWh in 2025 to become the world’s largest stationary-storage integrator by shipment volume (Benchmark, ~13% share, ahead of Tesla) — though it ranks fourth on Wood Mackenzie’s broader ten-criteria business-quality scorecard, behind Sungrow, Tesla and Contemporary Amperex. The two rankings are not in conflict: one measures boxes shipped, the other bankable-business quality. Revenue is plausibly ~US$6–8bn (roughly RMB 43–58bn) in 2025 [analyst inference]; BYD discloses no storage-segment profit, so margin must be read by proxy — and the proxies cut against the simple “same cost engine” story. Contemporary Amperex earns a ~24–27% storage gross margin (higher than its own car-battery margin) and the pure integrator Sungrow ~37%, while since 2023 a flood of cheap merchant cells removed the access advantage, so owning a cell factory alone stopped winning the integration game — the sub-scale cell-makers-who-integrate lost share, and pure integrators (Sungrow, CRRC, HyperStrong) rose from ~20% to ~30% of the market (2023 → H1 2025). But — the point worth stating precisely, because it is counter-intuitive — the champions did not lose: BYD is #1 by shipments and CATL is a top-3 integrator, because they are cost/scale leaders, not merely because they own cells. So the storage moat is cost + scale + software leadership, held by integrated (BYD) and non-integrated (Sungrow) players alike; mere cell-ownership is not the edge it is in cars, where the cell is scarcer and fused into the vehicle. So storage is real number-one volume of unproven profit — a call option on top of the car business, not yet a proven co-equal leg. The single highest-value catalyst would be BYD disclosing a storage profit-and-loss. [FACT/THIRD-PARTY: Benchmark, Wood Mackenzie, BNEF, IEA, Contemporary Amperex FY2025; the margin band and revenue estimate are analyst inference.]

Batteries and chemistry — the moat’s durability. About 90% of the world’s storage cells are already Chinese, and stay cheaper than local production even through tariffs above 100%. BYD is developing both of the next chemistries — sodium (cheaper materials, no lithium) and solid-state (higher energy density; demo ~2027, mass production around 2030). But the honest read (tested 30 July 2026) is not a simple moat extension — it is neutral near-term with a deferred risk. Sodium is a materials-and-scale advantage, so it favours the largest merchant maker (CATL) over BYD's make-it-yourself model — the same “integration inverted” dynamic seen in storage — and its low energy density confines it to storage and two-wheelers; tellingly, BYD keeps its cars on captive LFP and points sodium at storage/scooters, defending the car cost moat rather than betting it on sodium (sodium is also still dearer than LFP per kWh today). Solid-state is a step-change where BYD is a fast-follower, not the leader (Toyota holds the patents; QuantumScape/PowerCo lead the Western pilots; CATL is level with BYD in China) — but auto-scale, cost-competitive all-solid is a ~2030 event for everyone, and the first cells target premium range, not the mass-market cost frontier Blade is anchored to, so Blade capex is not stranded this decade. Net: the cost moat is durable through ~2030 but no longer widening (LFP is near its cost floor), with a genuine post-2030 disruption tail if a rival reaches auto-scale all-solid first — a catch-up-capex risk, not obsolescence. [FORCES-OF-CHANGE, #31; dated 30 July 2026.]

Charging, autonomy and robotics — levers and optionality, weighted up but unproven. BYD’s management frames the next decade around autonomous driving, chips and robotics — aiming to become “the biggest data company” on the back of a 4.6-million-car fleet generating some 200 million kilometres of driving data a day. Two things make this more than a slogan. First, BYD is pushing lidar and high sensor counts down-market faster than anyone — the 2026 Seagull brought lidar to the sub-US$14,000 (A00 mini-EV) class for the first time, with lidar trims from ~90,900 yuan (~US$12,600) — so it is building the largest high-quality-sensor fleet, cheaply, because it makes its own lidar, cars and now its own driving compute. Its May 2026 “God’s Eye 5.0” launch paired a tenfold-redundant, arguably class-leading sensor suite — 1,000-line-plus lidar, dual far-infrared (thermal) cameras and high-speed event cameras — with the self-developed Xuanji A3, China’s first 4nm intelligent-driving chip (~700 TOPS per chip, ~2,100 in a three-chip cluster), co-designed with its L3/L4 algorithms; it also holds an L3 licence. That chip-and-sensor co-design extends the battery-style vertical-integration cost logic into the autonomy stack. Second, its “central brain” unifies the powertrain, cabin and driving chips on one board — a genuine vertical-integration and organisational edge. [FACT / MANAGEMENT CLAIM — BYD “Dare to Be” event, May 2026, third-party corroborated; BYD framed the chip as world-leading, though independent reports call it China’s first.] But this is an edge on the hardware, cost and data-collection axis, not the software one — and that distinction is the whole point. BYD’s own intelligent-driving chief says BYD’s strength is sensors, engineering and integration, while it partners with Huawei, Horizon Robotics, Momenta and DJI on the algorithms and openly concedes user-experience software is a relative weakness [MANAGEMENT CLAIM, official shareholder channel, Jun 2026]. BYD does claim one algorithmic differentiator — a “physical AI” model that predicts surrounding traffic a few seconds ahead rather than only imitating — but with no disclosed performance number it is a claim to watch, not a proven lead. And a telling detail from the same interview: BYD offers the accident-liability “backstop” precisely because city-NOA usage stickiness had stalled near 30% — the product was not yet compelling enough for drivers to use heavily, so BYD is subsidising adoption to feed the very data flywheel the bull relies on. “Software is easy with AI” cuts both ways: if the code commoditises, the differentiator becomes data + compute + deployment (which favours BYD’s fleet) — but rivals have huge fleets and stronger stacks too, so a data lead alone does not win, and BYD gives the assistance away free (now with an accident-liability backstop), which removes the monetisation and adds a tail liability. Read honestly: the data-and-sensor flywheel is real and executing, and it raises the bull ceiling — but with no disclosed autonomy-performance metric and no monetisation, it is carried as upside, not base-case moat credit. Signposts: a disclosed intervention / zero-takeover rate, the share of the fleet shipped with lidar, a first paid autonomy tier or robotaxi/licensing deal, and whether BYD’s own stack (not a partner’s) drives the flagship. [MANAGEMENT CLAIM / FACT (lidar, L3) / ANALYST INFERENCE, dated 2026.]

Cross-vector: every leg leans on the same asset — captive battery-cell and power-electronics manufacturing at automotive scale. Storage is the only non-core leg that is already a number-one business; charging and autonomy are car-sales enablers; semiconductors is the one piece of optionality with independent traction (BYD is China’s leading automotive power-chip supplier, though its value is mostly captive — cheaper chips lift the auto margin rather than form a separate profit pool).

Read on the thesis: every leg leans on the same battery-manufacturing engine; storage is the one already at scale.

The centrepiece

The industry and the capital cycle — the question that sets the returns

The single most important lens is the capital cycle, and it counsels caution. The Chinese new-energy-vehicle industry carries roughly twice the capacity it needs (about fifty per cent utilisation); more than a hundred brands compete; a three-year price war has erased on the order of US$69 billion of industry revenue; and capital still floods in even as first-half 2026 retail sales fell about 13% year on year. This is the classic late-cycle shape — an industry competing its own returns toward zero — and the cautionary analogue sits inside BYD’s own past: the Chinese solar boom of 2011–12, which it entered and lost money in.

The shakeout — who survives. Only a handful of some thirty makers earn anything. A full-year-2025 third-party count put it at about three; the first-quarter-2026 issuer primaries show a somewhat different profitable set — BYD, Geely, Tesla and Seres — but the shape is the same: a long tail of loss-makers waiting to be cleared.

The counterweight: BYD is the low-cost survivor most likely to cross the graveyard rather than enter it — a reading supported by the founder’s correct 2018 prediction that only “ten to twenty per cent” of the new entrants would survive, and increasingly carried by exports rather than the home market.

Does it normalise, or stay a price war? This is the question that decides the return, and the historical record answers it more sharply than a scenario table can: consolidation does not confer good returns — exit does. Whether the survivors then earn above their cost of capital turns on how the field clears. Across eight historical consolidations the pattern is consistent:

The rational cases share hard entry barriers, real differentiation, and genuine exit of losers; the value-destroyers share commodity-like products and blocked exit — bankruptcy protection, subsidy, or bailout. Chinese new-energy vehicles sit between the two: entry barriers are rising (software, battery, scale) and the product is genuinely differentiated, but price is drifting toward commodity, and autos is the sector where Chinese local governments most resist exit.

The policy overlay. Beijing’s “anti-involution” campaign is the swing factor — it is trying to stop the price war rather than let the cycle clear it, which is itself a source of unpredictability. In June 2026 the regulators again summoned automakers over “irrational” pricing; the 60-day supplier-payment rule and a demonstrated willingness to force capacity retirement in solar could engineer the rational outcome, but it is a scalpel, not a switch.

The reframe (dated 2026, ruling #31): the BYD case should not rest on “the industry becomes a rational oligopoly” — that is the weaker claim. It rests on the Toyota question: can the lowest-cost, most-integrated player out-earn a structurally poor industry? China’s white-goods winners prove that exact playbook — vertical integration, component control, premiumisation, exports — can work. [THIRD-PARTY / ANALYST INFERENCE.]

Source tiers (ruling #23): the electric-vehicle penetration path and the storage growth path are cited to ingested primaries (IEA Global EV Outlook 2026; IEA Batteries and Secure Energy Transitions). The Chinese-industry specifics (brand count, how many makers are profitable, the overcapacity ratio) and the eight historical consolidation analogs are third-party leads — authoritative for context, but there is no single issuer filing behind an industry-economics fact, so they are labelled as leads rather than dressed up as primaries.

Read on the thesis: the industry is competing returns away today; consolidation alone does not confer returns — exit does; the bet is the Toyota question, not the oligopoly.

The scenarios

Bear, base and bull — as stories, with the signposts to watch

Bear (30%) — the winning-but-commoditised number one. BYD keeps winning volume, but the industry never clears: exit stays blocked, the price war grinds on, the supplier-float unwind bites cash flow, tariffs wall off the export margin, and assistance and storage never earn. Autos become a low-return utility; the achievable owner return stays below the cost of capital. Signposts: net margin stuck in low single digits through 2027, overseas prices falling toward domestic, free cash flow still negative, the survivor count still above twelve.

Base (45%) — BYD is the Toyota. BYD is the lowest-cost survivor and out-earns a structurally poor industry even if that industry never becomes a rational oligopoly; export and premium lift the blend; storage is a solid second leg; through-cycle auto margin normalises to about 6–8%. The achievable owner return at today’s price is about 10–12% a year — below the 15% hurdle. Signposts: margin holding as volume grows, export plants past 50% utilisation, storage disclosed above 15–20% of group profit, the field thinning toward six to eight makers.

Bull (25%) — the energy-and-mobility champion. Storage rivals autos, the data flywheel and driver-assistance monetise, sodium and solid-state extend the cost lead, and the market re-rates BYD from price-war carmaker to quality energy compounder. Even here the achievable return tops out near 12.5% a year — still short of the hurdle at today’s price. Signposts: storage margin above auto margin for a year, a first paid autonomy tier with a disclosed take-up, a named robotics product with outside orders, group returns on capital rising despite diversification.

Every case sits at or below the 15% required return — even the full bull tops out near 12.5% a year at today’s price, which is the whole point: the dispersion is about margin and return, not size. Weights are set by the engine (ruling #32) and reviewed by the owner. The bear is a touch heavier than a naive read because the analog record makes the industry-level bear the stronger structural case — offset by BYD being a credible low-cost champion. Baseline FY2025 (fact): revenue RMB 804bn, operating profit RMB 40.2bn (the ~50.5bn cited in an earlier draft was the FY2024 figure — corrected), operating cash flow RMB 59.1bn, about 4.6m vehicles. Anchored to IEA Global EV Outlook 2026 and IEA Batteries and Secure Energy Transitions, both ingested.

Read on the thesis: size grows in every case, so the whole dispersion is a margin-and-return bet — and no case clears the 15% hurdle at this price.

The value & the price

Fair value at the bond, and the 15% return the price does not pay

The price, struck as fact.

The figures below are struck valuation context, published as fact — the price and the multiples the market was paying on the stated date. They are not a recommendation, a target or a view; prices are struck from the exchange feed on the valuation date.

Owner earnings are a range, and the range is the point. On a maintenance-capital basis — the assumption that true maintenance capital expenditure runs at roughly depreciation and amortisation — they are on the order of positive RMB 34 billion; on a free-cash-flow basis (treating growth capital expenditure as committed) they were about negative RMB 98 billion in FY2025. Normalised through the cycle, and after netting the roughly RMB 5–6 billion of interest that the supplier-float unwind costs as it reverses, the figure we carry is about RMB 38–42 billion. Two cautions ride with it: whether that growth capital expenditure is discretionary or committed is the single assumption the economics most depend on; and the maintenance-capex-equals-depreciation assumption is the key downside sensitivity — if true maintenance capital is materially higher, the normalised base falls toward RMB 20–30 billion.

Fair value — the whole company, discounted at the bond. Under the house convention (ruling #40), intrinsic value is a fair value: the destination’s owner earnings and free cash flow discounted at a long-term risk-free rate — here the ~3% long-term Chinese government bond, the local risk-free for a renminbi-earning business — with the terminal year capped at 14× free cash flow (a durability cap, not a growth multiple). Valued whole-company first (per Li Lu; the per-share figure is derived, not the unit of analysis): the present value of interim free cash flow to 2035 is about RMB 300 billion (the early years near zero as the float unwinds), and the terminal — 14 × roughly RMB 82 billion of normalised free cash flow, discounted back ten years — is about RMB 855 billion, for operating value near RMB 1,155 billion. Subtract adjusted net debt of about RMB 320 billion — carrying the payables and Di-Chain supplier float as debt, not accepting the reported net-cash optics — and equity fair value is about RMB 800–950 billion (midpoint ~875bn). Against the RMB 698 billion market cap, BYD trades at roughly 0.75–0.85× fair value — about 15–25% below what a risk-free-return buyer would pay. On this measure the market is not over-paying; a Buffett-style buyer who discounts at the bond would call it modestly cheap. (Sensitivity: at a US-dollar risk-free near 4.5%, the number the H-share investor might use instead, fair value falls to about RMB 620–720 billion — roughly parity with the price.)

The return the price actually pays — the go/no-go. Fair value tells you the stock is not expensive; it does not tell you the return. An owner’s return is roughly the rate at which intrinsic value compounds plus the cash yield. BYD compounds intrinsic value at about its return on invested capital times its reinvestment rate — and the recomputed return on capital is about eleven per cent (normalised operating profit after tax near RMB 34 billion over invested capital near RMB 300 billion; above a ~10% cost of capital, below the 15% hurdle), reinvested almost in full, plus a roughly one per cent dividend and a mild tailwind as the price drifts toward fair value. Run through the destination, that implies an achievable owner return of about 10–12% a year in the base and about 12.5% in the full bull — in every case below the 15% we require. The cross-check agrees: discount the same destination at 15% and it is worth about RMB 290–300 billion, so the RMB 698 billion price implies an internal rate of return well under the hurdle, consistent with the 10–12%.

The verdict — Pillar 4 fails at the price. The achievable return, roughly 10–12%, falls short of the 15% required return, with no margin of safety. The two measures point different ways, and the honest thing is to name it: BYD trades below its risk-free fair value yet still fails the 15% test. That is the framework working as designed — value the business at the bond, but demand a return, not merely a discount to fair value — and it is the sharpest statement of what we do differently: the market prices a ~10–12% return and is content; we require 15%. The buy level — where the achievable return clears 15% with a margin of safety — is a market capitalisation near RMB 290–300 billion, about HK$37–40 a share (against roughly HK$88–93 now), a little over half today’s price. That is the framework’s break-even entry, the price at which the return math clears the hurdle — not a position-sizing instruction.

Where the value sits — the facets. The same business can be decomposed on either discount basis, and the composition is what matters — it barely moves between them. Shown here on the owner-earnings basis capitalised against the 15% required return — the value at which the return would clear the hurdle, and therefore the ~RMB 290–300 billion buy level — so the parts sum to the whole. (At the risk-free rate the levels scale up toward the ~RMB 875 billion fair value, but the shares — autos ~83%, storage the bull lever, autonomy ~0 in the base — are the same.) For each segment: what it becomes (its thesis), why it is worth that, and its share of the total. The captive battery and captive-chip cost advantage is folded into the autos line (it raises the auto margin) rather than added as a separate leg — the double-count guard that lets the parts tie to the whole. Splits are analyst inference (BYD discloses segment revenue only broadly, and no storage or semiconductor profit); the point is composition, not decimal precision.

The read: autos is ~83% of the value; storage is the biggest — but still unproven — bull lever; autonomy/robotics is priced at zero in the base. Because this decomposes the value (captive batteries and chips sit inside the autos line, not bolted on top), the parts sum to the whole. Read against the 15% required return, today’s ~RMB 698 billion price is well above this ~RMB 300 billion figure — it already capitalises autos-as-Toyota plus a large slice of the storage and optionality bull, which is why the return it offers falls short of the hurdle even though the same business is cheap to a bond buyer. That is the crux the parts make visible. [ANALYST INFERENCE; full workings: byd-value-facets-sotp.]

Integrity and the balance sheet: the short-seller “hidden debt” thesis was tested against the filings. The supplier financing it points to is large and real — but it is disclosed in both the Hong Kong and mainland accounts, under standard terminology and with the purpose stated in BYD’s own notes — so it is a working-capital and balance-sheet compromise, not concealment; the integrity gate clears. The sharpened bear is free cash flow: the anti-involution 60-day payment rule (the revised SME-payment regulation, effective June 2025) is unwinding the supplier float, and the balance sheet swung from net cash to net debt in 2025 to fund a capacity wave. Sized (30 July 2026): the remaining one-time cash drain to a genuine 60-day norm is roughly RMB 130–200bn over 2026–27 — large, but a one-time transition cost, not an operating loss (BYD is still profitable), and BYD pre-funded it with a ~HK$43.5bn (~RMB 40bn) equity placement in March 2025 — management saw the squeeze coming (a capital-allocation positive). Liquidity is thin but adequate: it now depends on continued access to debt (short-term borrowings +72% in Q1 2026), and free cash flow should turn positive around 2027–28 as capex rolls off its peak. Read: material-but-survivable, not a liquidity crisis. The signals that would flip it to avoid — a second equity raise, any going-concern or qualified auditor opinion, or a failed / expensive short-term-debt rollover — are the things to watch in the late-August H1 report. And the “hidden debt” question was resolved against the primary filings (30 July 2026): the derecognised endorsed notes are 100% bank-acceptance bills — no commercial-acceptance, no recourse retained (Ernst & Young unqualified opinion), so the concealment thesis fails and the integrity gate clears; what remains is disclosed supplier-financing leverage (the Di-Chain platform, now being wound down, and an opaque ‘external current-account’ payables line) that a valuation should carry as adjusted debt — roughly the RMB 320 billion netted above — rather than accept the reported net-cash optics. [FACT / ANALYST INFERENCE, forces-of-change #31.]

Read on the thesis: cheap to a bond buyer, but the return the price pays falls short of 15% — no margin of safety, though the integrity gate clears.

What could break it

Business Risks & AI Exposure

The principal risks are the capital cycle (returns competed down in an over-supplied, state-managed home market), the balance sheet (first-quarter 2026 short-term borrowings up about 72 per cent and notes payable up about 116 per cent, with roughly Renminbi 119 billion of derecognised notes an un-adjudicated forensic item), the research-and-development capitalisation shift, and the extrinsic overlay — the United States has effectively closed to BYD by statute (a Department of Defense listing with a procurement ban from mid-2026), and the European Union is tariffed.

On artificial-intelligence exposure the evidence points to durability, for a specific reason. BYD’s moat is physical — battery electrochemistry, vertical manufacturing, capital, scale and a compounding fleet-data asset. But two independent sources — BYD’s own smart-driving chief Yang Dongsheng, who concedes the company “won’t always be the most advanced” in artificial intelligence and stays open and partnered on algorithms, and Ford’s Jim Farley, who names Huawei, Xiaomi and NIO as the in-car software leaders — agree that BYD is not the software leader. The correct reading: the moat lives in the hardware and manufacturing stack; this report explicitly declines to credit BYD with a software moat it does not claim.

Read on the thesis: the sharpened bear is free cash flow and the state-managed war, not integrity.

The verdict

Opportunity type: PASS — a great business, cheap to a bond buyer, short of a 15% owner’s return

This is not a bad-business pass. BYD is a genuinely extraordinary, proven-moat operating company; the framework declines it at today’s price for one reason only — the return. The positional destination — dominant global maker of new-energy vehicles and their batteries — is close to settled and already reached. What is unresolved is the economics of that destination: whether a return on capital near eleven per cent, reinvested almost in full in a state-managed home market carrying roughly twice the capacity it needs, ever throws off a 15% owner’s return. Valued at the bond the stock is modestly cheap — about 0.75–0.85× fair value — but the return that price actually pays, roughly 10–12% a year, is below the 15% we require, with no margin of safety. Great business does not mean buy at any price; the entry that clears the hurdle sits near RMB 290–300 billion (about HK$37–40).

The eight-pillar scorecard, and the three gates.

The three hard gates all pass: integrity (management judged on behaviour, clean); harvesting (the gate targets extracting price from trapped customers — a relentless price-cutter is the opposite, so it does not bite); value trap (the business grows, it is not a melting ice cube). Beyond the gates the pillars are weighed, not counted:

The edge: psychological at best, possibly absent — the most-covered stock on earth offers no informational edge, and over an economic destination this hard to forecast, “I can hold when others can’t” is hard to distinguish from holding what you cannot value.

The falsifiable tests — what would most change this: (1) primary: BYD’s operating margin and free cash flow stabilise above the cost of capital as the industry consolidates — first decisive read is the H1 2026 interim (does the export-led re-acceleration carry margin, not just units?); (2) the mass-market-only gross margin, if it is falling hard while premium and export mix carry the blend; (3) the capitalised-development balance amortises unimpaired rather than being written down.

Panic test: halved tomorrow — disoriented, not glad, because the economic destination cannot yet be confidently valued.

Read on the thesis: a great business the framework declines at this price — worth owning at the destination, but the entry is not here yet.