← Cross-cutsCross-cut · China EV

The involution war

China’s electric-vehicle market is in a price war so brutal its own regulators call it one with “no winners.” Five Teardown studies sit on different rungs of it — BYD, Tesla, Li Auto, NIO and the battery giant CATL. Read together, they don’t just describe the war; they map the handful of positions that survive it, and the much larger middle that doesn’t.

The war, in one number: profit per car

The whole cluster can be arranged on a single axis — how much margin the price war has stripped out. BYD, the company that escalated it, now earns roughly ¥8,500 of profit per car, about a sixth of Tesla’s estimated ~¥60,000. That gap is the price of winning on cost: BYD sells the most NEVs on earth (4.60M in 2025) and still saw net profit fall 19%, its first decline in years. The war is a margin-compression machine, and each of these five sits at a different point on it — from Tesla’s still-fat-but-shrinking premium to NIO, which lost RMB 14.9bn for the year even while booking its first-ever profitable quarter.

Four carmakers, four exits — and one supplier who isn’t playing

Each carmaker has chosen a different way to not die. BYD tries to out-scale the war — vertical integration (in-house Blade batteries, its own chips) lets it set the price floor, and a +151% export surge gives it volume the home market no longer will. Li Auto tried to sit above the battle in a profitable range-extender niche. NIO bets a premium brand and a 3,700-station battery-swap network can hold a high-cost niche the discounters can’t reach. Tesla, the foreign incumbent that started the EV shift, is in retreat as a carmaker in China and increasingly asks to be valued as an AI-and-robotics company instead. And then there is CATL, which isn’t fighting the war at all — it sells the ammunition.

The two safe seats are the extremes

Line the five up and the lesson is stark: the structurally advantaged positions are the two ends of the value chain, not the middle. The most vertically integrated player (BYD) can dictate price; the player least dependent on any single carmaker (CATL, with 39.2% of global battery installations) profits no matter which brand wins the showroom. The proof is in the profit pool — the one company in this cluster printing record earnings, up 42% to RMB 72.2bn, is CATL, the one that builds no cars. The dangerous seat is the mid-tier pure carmaker, exposed to the price war on the outside and to its battery supplier on the inside.

Li Auto is the warning the others should read

For three years Li Auto was the proof that a Chinese EV startup could be profitable — a tidy range-extender niche, no model under RMB 200k, margins above 22%. Then, in a single year, the moat commoditized: gross margin collapsed from 22% to 7.9%, deliveries fell ~19% (the only major Chinese maker to shrink), and Huawei and Xiaomi moved into the same niche. Li Auto is the cross-section’s sharpest cautionary tale — evidence that in this market a product-niche advantage is rented, not owned. It is exactly the fate NIO is betting its swap network and brand can avoid.

Where they agree — and where they split

All five studies accept the same premise: the war is real and structural, not a passing promotion. Industry net margins fell to ~3.9% in early 2025, and Chinese capacity (~15M units) runs far ahead of demand (~10M), so discounting is a feature of overcapacity, not a tactic. They split on the escape routes. Is export the release valve, or just a way to export the margin problem into EU and US tariff walls (BYD’s +151% exports already meet a 17% EU duty and a 100% US wall)? Can a premium niche hold — NIO’s and Li’s studies both land that question as genuinely contested? And is even the picks-and-shovels seat safe, or will involution climb upstream into CATL’s margins too? The demand for EVs isn’t the question. Who gets paid for building them is.

The cluster at a glance

Company2025 scaleMarginProfitabilityRole in the warValuation
BYDHK 1211 / SZ 0025944.60M NEVs (+7.7%) · #1Auto GM ~20.5% (4 qtrs falling)Net −19% · ~¥8,500/carCost-led aggressorRMB 804bn rev
TeslaNASDAQ: TSLA1.64M (−9%, 2nd drop)Auto GM thin · 5.7% opNet $3.79B (−46%)Foreign incumbent ceding China~$1.56T · ~400× P/E
Li AutoNASDAQ: LI / HKEX: 2015406k (−18.8%, only one to shrink)GM 22% → 7.9%Net RMB 1.1bn (−86%)Premium EREV niche, commoditizingStock −51% YoY
NIONYSE/HKEX: NIO326k (+46.9%)Vehicle 14.6% (Q4 18.1%)FY loss RMB 14.9B · 1 good qtrPremium, battery-swap niche~$12.6B · ~1.0× sales
CATL — Contemporary AmperexSZSE:300750 / HKEX:375039.2% global battery share~31% overseas GMRecord RMB 72.2bn (+42%)Picks-and-shovels, above the fray~$300bn · ~28× P/E

Figures as of each study’s stated date (2026-06); see each teardown for sourcing and the full weighing. Margins and multiples are on mixed bases and not strictly comparable; CATL is a battery maker, not a carmaker.

The five studies — and the question each turns on

BYD Company Ltd (比亚迪)HK 1211 / SZ 002594Can a vertically integrated, cost-led champion stay profitable and expand globally while the domestic price war it helped escalate compresses margins and tariffs/controversies meet it abroad?Read the full weighing →Tesla, Inc.NASDAQ: TSLAIs Tesla a car company in structural decline, or an AI/robotics company whose ~$1.5T valuation will be vindicated by robotaxis and Optimus?Read the full weighing →Li Auto Inc. (理想汽车)NASDAQ: LI / HKEX: 2015Is 2025 a forced reset that a cash-rich, still-profitable franchise reinvents past — or the terminal peak of an expiring range-extender edge?Read the full weighing →NIO (蔚来)NYSE/HKEX: NIOIs the Q4 2025 profit a durable inflection for a high-cost premium brand, or one favorable quarter in a decade of losses?Read the full weighing →CATL — Contemporary Amperex Technology (宁德时代)SZSE:300750 / HKEX:3750Is CATL's dominance staying as valuable as it looks — or is the home-market price war quietly eroding a moat the market values above RMB 2 trillion?Read the full weighing →

This is the kind of reading the Desk does for you

A cross-cut takes five teardowns and one price war and asks what they say together. The Desk does the same for the companies you actually own — your thesis, the peers that move it, and the tripwires that would change your mind.

See the Desk →