One country — 68% of the world.
Over the past two years, the global EV market traced an extremely asymmetric curve: 2024 global new-energy vehicle sales were ~17.8M; 2025 surged to 20.7M (Rho Motion, +20% YoY), with new-vehicle penetration rising from ~18% to ~24%. But split that aggregate and three worlds barely overlap — China takes 68%, Europe’s BEV share lands at 17.4%, and the US plunged to 5.8% in Q4 after the $7,500 tax credit died. The China curve isn’t carried by subsidies; it’s LFP + PHEV + price war + autonomy democratisation. As the totem of a global leader change: BYD’s 2025 BEV sales of 2.26M units overtook Tesla’s 1.64M for the first time on a pure-BEV basis.
§01 · Overview — what happened to EVs over two years
Image: Wikimedia Commons / CC BY-SA 4.0.
The 24 months from 2024 to 2026 were the global EV industry’s watershed from “policy-driven” to “scale-driven”. Three things define the window: (1) China’s NEV penetration leapt from full-year 2024 ~48% to 59.1% in retail in December 2025 (CPCA; CAAM full-year 47.9%); (2) Tesla was overtaken on the BEV measure by BYD — BYD 2025 BEV 2.26M vs Tesla 1.64M; (3) US policy reset, the One Big Beautiful Bill Act ended the $7,500 EV tax credit on 30 September 2025, with full-year US EV sales of 1.28M (-2% YoY) but Q4 single-quarter penetration crashed from Q3’s 10% to 5.8%.
Tech and cost are collapsing in parallel — BloombergNEF puts the 2025 global pack price at $108/kWh (-8% YoY), with China at $84/kWh; LFP first overtook NMC in global passenger-EV installations, with China LFP at 79% of GWh. CATL 39.2% + BYD 16.4% take 55.6% of the global battery market between them; seven of the top ten are Chinese. On autonomy, Tesla Robotaxi launched in Austin in June 2025 (Model Y only); Waymo reached 450k paid rides per week by the end of 2025, dominating US Robotaxi.
| Metric | Value | Notes |
|---|---|---|
| 2025 global EV | 20.7 M · +20% YoY | Rho Motion · BEV+PHEV |
| Global penetration | ~24% · +4pp YoY | Of light-vehicle new sales |
| BYD vs TSLA | 2.26M vs 1.64M · BEV basis | BYD first overtakes Tesla |
| LFP global share | ~52% · first passes NMC | China 79% · NA 5% |
Bottom Line · The phrase “global EV market” is breaking down — it is now three worlds that don’t talk to each other.
China is a world of scale + price war + autonomy democratisation; Europe is a world of policy × slow BEV ramp; the US is a world of policy reversal + Tesla as near-monopolist. The three markets share almost no product, cost structure, regulation or consumer preference. Internalising that matters more than searching for a “global leader”.
§02 · Global sales — from 17.8M to 20.7M
The global EV sales curve looks like a beautiful monotonic rise at the annual level; zoom into quarters and you see violent policy pulses — every Chinese NEV-subsidy taper, every European CO2 compliance deadline, every US tax-credit pre-buy creates a 15-30% spike in quarterly data. Separating “structure” from “pulse” is what keeps you from being misled by a single quarter.
Sources: Rho Motion · IEA Global EV Outlook 2025 · Counterpoint Research. 2024Q1–2026Q1 is global BEV + PHEV combined (M units); penetration is share of global light-vehicle new sales (right axis). 2026Q1 is preliminary (Rho Motion / IEA).
Quarterly sales · 2024Q1 – 2026Q1
| Quarter | Global M | BEV % | PHEV % | Penetration | Key event |
|---|---|---|---|---|---|
| 2024 Q1 | 3.2 | 62% | 38% | ~16% | Global +18% YoY · PHEV momentum first leads BEV |
| 2024 Q2 | 4.3 | 63% | 37% | ~18% | China PHEV continues to scale; EU imposes provisional anti-subsidy duties on China BEVs in July |
| 2024 Q3 | 4.7 | 64% | 36% | ~20% | China September retail penetration first crosses 50% |
| 2024 Q4 | 5.6 | 65% | 35% | ~22% | EU formal 7.8%–35.3% duties on China BEVs effective in October; China Dec NEV retail 47.97% for the year |
| 2025 Q1 | 4.1 | 65% | 35% | ~19% | China starts Q1 strong · Europe EVs warm up in CO2-target year |
| 2025 Q2 | 5.0 | 65% | 35% | ~22% | H1 cumulative +28% YoY to 9M · Europe BEV YTD first crosses 35% of BEV+PHEV |
| 2025 Q3 | 5.8 | 66% | 34% | ~25% | US Q3 EV sales 438k record (tax-credit pre-buy) · Europe PHEV +38% |
| 2025 Q4 | 5.8 | 66% | 34% | ~24% | US Q4 single-quarter penetration crashes to 5.8% (tax credit expired 9/30) · December single-month 2.1M new global high |
| 2026 Q1* | ~4.3 | 65% | 35% | ~22% | BYD Q1 -30% YoY · Tesla Q1 +6.3% to 358k · China Jan/Feb price war intensifies |
* 2026 Q1 preliminary. Annual totals: 2024 ≈ 17.8M (Rho Motion; some IEA at 17.0M) · 2025 ≈ 20.7M. Tracker differences noted below.
- BEV vs PHEV tug-of-war. 2024-2025 PHEV momentum was stronger — China PHEV share rose from 30% in 2023 to ~40% in 2025. But BYD 2025 BEV +27.9% / PHEV -7.9% suggests the curve is tilting back to BEV. Europe in this round saw PHEV +38% (2025), but BEV +31% was faster — both rising together. Global BEV share 2025 ~66% · 2023 trough ~63%.
- Tracker methodology divergence. Rho Motion (2025 global 20.7M) · IEA (Global EV Outlook 2025: 2024 17.0M) · Counterpoint (quarterly basis). They differ on whether “EV” includes FCV, commercial vehicles, or only passenger cars. This page uses Rho Motion as the base and notes others in the comparison columns. IEA 2024 = 17.0M vs Rho Motion 17.8M, ~5% gap.
- Extremely uneven regional contribution. Of 2025’s 20.7M global, China is ~14M (68%), Europe ~3.5M, the US 1.28M, others ~1.9M. That means over 80% of global growth comes from China — once China slows, the global headline goes weightless. The single-digit YoY in China NEV in Q1 2026 is already an early warning. ChinaEVHome: China Jan-Nov 2025 = 68% of global.
- Penetration is non-linear. Global penetration went from ~15% in 2023 to ~24% in 2025, +4-6pp every 12 months; 2026 may be the first year of a slowdown (US pullback + China base effect). IEA expects 2030 global penetration 40-45%; BloombergNEF expects ~50%. IEA 40% vs BNEF 50% · 10pp gap on the 2030 forecast.
“If 2020 was EVs’ high-school graduation, 2024-2025 is the start of grad school — slower growth, fiercer competition, structural divergence, subsidy taper — but those are signs of maturity, not decline.”
— Adam Panayi, Managing Director, Rho Motion (late 2025)
§03 · Regions — China · Europe · US · Rest
Splitting 2025’s 20.7M into regions: China alone 68%, Europe 17%, US 6%, others 9%. Even more important is the internal structure of each: China is price war + LFP + autonomy ubiquity; Europe is CO2-target compliance; the US is Tesla + a handful of others, suddenly snapped back to 2019.
Sources: IEA · CPCA · ACEA · Kelley Blue Book / Cox Automotive · synthesized. Stacked area; 2026 Q1 preliminary. “Other” = Japan, Korea, SE Asia, India, South America, Australia.
Regional snapshot · 2025 full year
| Region | 2025 sales M | YoY % | EV penetration | Lead vendors | Key events |
|---|---|---|---|---|---|
| China | ~14.0 | +17-18% | 47.9% (CAAM) / 59.1% Dec retail (CPCA) | BYD · Geely · Wuling · Tesla · Xiaomi | December retail penetration first crosses 60% · CAAM 16.49M NEV |
| EU (EU27) | 1.88 BEV + PHEV ~2.2 | +31% BEV | BEV 17.4% · +HEV 51.9% | VW Group · Stellantis · BMW · Tesla | EU duties on China BEVs in force · ICE + hybrid combined down 35.5% |
| UK + EFTA | ~0.65 | +20% | BEV 19.5% (incl. UK / Norway) | Tesla · VW · BMW · BYD (UK) | UK ZEV mandate 2025 target 22% BEV |
| USA | 1.28 | -2% | 7.8% (Q3 10% · Q4 5.8%) | Tesla 45% · GM 13% · Ford · Hyundai | $7,500 tax credit expired 9/30 · Q4 nosedive |
| Norway | 0.14 | +6% | BEV 95.9% | Tesla 19.1% · Toyota · VW · BYD | April single-month 97% BEV · world’s highest |
| Japan | 0.10 | +5% | ~3% | Nissan · Toyota · BYD entering | BYD March sales doubled · culture and infrastructure both slow |
| Korea | 0.18 | +10% | ~10% | Hyundai · Kia · Tesla · BYD | IONIQ family is the workhorse · subsidies kept |
| Southeast Asia | 0.35 | +60% | ~8-15% (Thailand highest) | BYD · SAIC · Vinfast · Wuling | Chinese brands dominate · Thai policy support |
Multi-source: CPCA / CAAM / ACEA / KBB / Cox Automotive / Rho Motion / IEA. China CPCA (retail) is ~1.5-2M lower than CAAM (wholesale incl. commercial); the table uses the CAAM all-in 16.49M. EU27 BEV figures are ACEA official.
Three big markets · structural breakdown
- China · 58%+ leads · 14M / 68% · Penetration crosses 60% · price war white-hot. CPCA December 2025 NEV retail share 59.1% — record high; CAAM full-year NEV 16.49M (+35% YoY, 47.9% penetration). BYD’s 4.6M for the year — but from Q4 2025 entered seven straight months of YoY decline; XPeng +126%, NIO +47%, Li Auto -19% — extreme divergence among the three. Structural signal: only BYD / Li Auto / Aito were profitable in 2025; Xi Jinping personally addressed “anti-involution” in August. Industry consolidation arrives from 2026.
- Europe · BEV 17.4% · 2.2M / 11% · CO2-target driven · China-EV duties land. 2025 EU BEV registrations 1.88M, +31%; including PHEV ~2.2M. ACEA shows BEV share 17.4% (+3.8pp YoY); HEV+BEV+PHEV combined first cross half (51.9%). EU formally imposed 7.8%–35.3% anti-subsidy duties on China BEVs in October, then turned to “price-undertaking” negotiation. Structural signal: the CO2 compliance year shifted OEM BEV pushing from “delay” in 2024 to “double down” in 2025; growth is still constrained by purchasing power and charging anxiety.
- US · policy reversal · 1.28M / 6% · $7,500 ends · Q4 crashes to 5.8%. Full year 1.28M, -2% YoY, share 7.8% (8.1% in 2024). Tesla 45% + GM 13% = 58% of the market. Q3 was a record 438k (pre-buy); Q4 immediately crashed to 5.8% penetration. OBBBA was signed on 4 July 2025; the credit expired on 9/30. Structural signal: the US EV curve has dual “policy × product” fragility — Tesla dominates the product side, and no new incentive is likely on the policy side before 2028.
Methodology · CPCA vs CAAM · One China, two completely different sets of numbers.
CPCA reports passenger-car retail (sales to end users); CAAM reports wholesale (including commercial vehicles). The 2025 NEV-penetration gap between them is over 10pp — CPCA December retail 59.1% vs CAAM December wholesale 52.3%; CAAM full year 47.9% is lower still. When you cite a China NEV figure, always note which source — otherwise you’re comparing variables that share a name but aren’t the same thing.
§04 · OEM duel — BYD vs TSLA · Chinese challengers vs Western incumbents
The biggest OEM event over two years isn’t “who sells more”; it’s the leader changing hands — BYD’s 2025 full-year BEV sales were 2,256,714 units, formally ahead of Tesla’s 1,636,000 on a pure-BEV basis. But that doesn’t put Tesla out of the game: per-vehicle gross profit, cash flow, single-brand global scale and the AI assets in Robotaxi/FSD are still far ahead. The two are now two #1s on different tracks — BYD champion of volume, Tesla champion of software / energy / AI.
Sources: company IR / monthly disclosures / Electrek / CnEVPost / company filings. BYD is total NEV (BEV+PHEV); others are deliveries. Tesla 2024-2025 from official IR; 2026Q1 = 358k.
OEM 2025 full-year deliveries
| OEM | 2024 FY | 2025 FY | YoY % | Basis / notes |
|---|---|---|---|---|
| BYD | ~4,272,000 | 4,602,436 | +7.7% | NEV total (BEV 2.26M +27.9% / PHEV 2.29M -7.9%); first time past Tesla on BEVs |
| Tesla | 1,789,226 | 1,636,000 | -8.6% | Global deliveries; two consecutive declines (peak 1.81M in 2023) |
| Geely Group | ~2,170,000 | ~3,000,000 | +38% | Includes Zeekr + Lynk & Co + Galaxy · NEV ~50% |
| Li Auto (LI) | 500,508 | 406,343 | -18.8% | PHEV-led · six straight months of declines · L6 / L7 / L9 |
| XPeng (XPEV) | 190,068 | 429,445 | +126% | Mona M03 + P7+ drove growth · overseas 45,008 +96% |
| NIO (NIO) | 221,970 | 326,028 | +46.9% | Onvo L60 sub-brand contributed · December 48,135 record |
| Zeekr Group | ~430,000 | ~710,000 target | +40%E | Includes Lynk & Co; absorbed by Geely and delisted from NYSE in Dec 2025 |
| Xiaomi Auto (SU7 / YU7) | 139,547 | 400,000+ | +187% | SU7 launched March 2024; YU7 SUV in June 2025; goal hit a month early |
| Rivian (RIVN) | 51,579 | 42,247 | -18% | R1 family · Amazon EDV · R2 mass production delayed to 2026 |
| Lucid (LCID) | 10,241 | 15,841 | +55% | Q4 5,345 record +72% · Gravity SUV ramping |
| VW Group (ID.4 / ID.3 / …) | ~740,000 BEV | ~880,000 BEV | +19% | European BEV #1 · ID.7 / ID.Buzz fill the lineup |
| GM (Ultium) | ~114,000 | ~168,000 | +48% | Equinox EV 57,945 hit · Blazer EV / Silverado EV |
| Ford (Mach-E + Lightning) | ~97,000 | ~79,000 | -18% | Mach-E 51,620 · Lightning 27,307 · Skunkworks low-cost car 2027 |
| Hyundai + Kia | ~320,000 | ~380,000 | +19% | IONIQ 5 (47,039 in US) · EV9 · EV6 · Kona EV |
| Stellantis (incl. Fiat / Jeep / Peugeot) | ~290,000 BEV | ~260,000 BEV | -10% | Weak European sales · Fiat 500e production stop and restart |
| Toyota (BZ4X + hybrid) | ~100,000 BEV | ~150,000 BEV | +50% | HEV-led · solid state in small batches in 2027 |
Multi-source: company IR / monthly disclosures / CnEVPost / Electrek / Electrive / ACEA. Some figures are estimates (especially VW Group BEV and GM Ultium); cross-source variance is ±5-10%. BYD and Geely NEV include PHEV; not directly comparable to BEV-only OEMs.
Four narratives · key takeaways
- BYD: volume champion, but growth is peaking. 2025 full year 4,602,436 NEV (+7.7%) — well below 2024’s +41%. 2026 Q1 -30% YoY (700,463 units), seven straight months of declines. The counter-attack is overseas — Q1 exports 321k, sharply higher YoY; March alone was 120k, 40% of the month’s volume. Structural shift: from “domestic volume king” to “global export king”. 2026 full-year target revision: 5.5M → not announced.
- Tesla: deliveries falling, software / AI still expanding. 2025 full year 1,636,000 (-8.6%); Cybertruck only ~15k (well below expectations). But Robotaxi launched in Austin in June 2025; FSD v14 shipped in October; the Shanghai factory exported 35,000+ in October — a two-year high. Tesla in China was ~625k in 2025 (-5%), holding #2. Tesla Q1 2026 deliveries 358,023 +6.3% YoY.
- Xiaomi’s “Lei law”. SU7 launched in March 2024; 2024 full year 139,547 units; 2025 full year 400,000+ — the fastest brand to 400k in the history of China EVs. YU7 SUV launched in June 2025, monthly sales at the 50k level. The Xiaomi model validates: brand + ecosystem + marketing + engineering supply chain can quickly steamroll traditional EV start-ups in China. 2026 Q1 single month 39,000+ deliveries.
- The extreme divergence of China’s “three musketeers”. 2025: NIO (326,028 +47%) · XPeng (429,445 +126%) · Li Auto (406,343 -18.8%) — three completely different curves on the same track. XPeng turned around with the Mona M03 low-priced BEV + P7+; NIO scaled with the Onvo L60 sub-brand; Li Auto’s “EREV” narrative suddenly faded in 2025 (pure-BEV MEGA underperformed). XPeng overseas 45k beats NIO + Li Auto combined.
- US EV: Tesla + GM, two-horse. Tesla 2025 US share 45% · GM 13% (+48%) · combined 58%. The rest is filled by Ford (Mach-E 51,620 + Lightning 27,307), Hyundai (IONIQ 5 47,039), Honda/Acura (Prologue 36,553 ODM-built on GM Ultium). Rivian -18%, Lucid only 15,841 — US “EV start-ups” essentially have no scale. GM Ultium first crosses 10% share.
- European incumbents’ “passive electrification”. VW Group is European BEV #1 (~880k). Stellantis weak at -10%; Fiat 500e production stop and restart. German BMW / Mercedes EQ family stays at the 200-300k range with high-end models. But Chinese rivals BYD / MG / Geely Polestar are penetrating faster via the price-undertaking deal; 2026 will be the test year for European local brands. Polestar 2 / 3 globally ~80k.
“China OEMs’ edge isn’t that one of them beats Tesla; it’s that ten of them simultaneously pull the EV cost curve down — upstream on scale, midstream on supply chain, downstream on autonomy. That kind of system-level capability cannot be replicated in Europe or the US within a year.”
— A China sell-side auto analyst, Q4 2025 investor meeting
§05 · Battery — LFP overtakes NMC · price under $108
The battery is the EV industry’s most important value-redistribution over the past 24 months. Three things happened in parallel: (1) LFP first overtook NMC in global passenger-EV chemistry installs; (2) BNEF’s 2025 global pack price fell to $108/kWh (China $84, US $121, Europe $131); (3) CATL + BYD took 55.6% of the world battery market between them. Together they mean: over the next five years the “energy density” narrative cedes to “cost + scale + vertical integration”, and Toyota / CATL / QuantumScape solid-state timelines cluster in 2027-2028.
Source: BloombergNEF Lithium-Ion Battery Price Survey (annual). 2025: BEV pack $99 · storage $70 · LFP avg $81 · NMC avg $128. All-in weighted $108/kWh.
Global battery market share · 2025 full year
| Vendor | 2025 installs GWh | YoY % | Share | Key customers |
|---|---|---|---|---|
| CATL | 464.7 | +35.7% | 39.2% | Tesla · BMW · XPeng · Li Auto · NIO · VW · Ford · Hyundai |
| BYD (FinDreams) | 194.8 | +27.7% | 16.4% | Mostly captive · partial external (Tesla Model Y LFP) |
| LG Energy Solution | 108.8 | +11.3% | 9.2% | Tesla (NA) · GM Ultium · Hyundai · Ford |
| CALB | ~55 | +22% | ~4.6% | GAC · XPeng · Great Wall |
| Panasonic | ~48 | -5% | ~4.0% | Tesla 2170 (Fremont · Reno) |
| SK On | ~45 | +8% | ~3.8% | Ford · Hyundai · VW |
| EVE Energy | ~35 | +40% | ~2.9% | BMW large-cylindrical orders · ESS |
| Samsung SDI | ~32 | -8% | ~2.7% | BMW · Rivian · Stellantis |
| Gotion | ~28 | +15% | ~2.3% | VW partnership · overseas plants |
| Sunwoda | ~22 | +30% | ~1.8% | Geely · SAIC · Easy-Electric |
| Top 10 total | ~1,035 | — | 86.9% | 7 Chinese in Top 10 · ~65% combined |
Sources: SNE Research / CnEVPost / company disclosures. CATL + BYD combined 55.6%; Chinese battery makers ~70% combined. 2025 global installs ~1,190 GWh, +30%+ YoY.
LFP vs NMC · chemistry shift
| Metric | Value | Notes |
|---|---|---|
| LFP share · 2020 → 2025 | 10% → 52% | 2020 global LFP <10% · 2024 ~40% · 2025 first overtakes NMC at 52%. China 79% · Europe 13% · North America 5% · others 57%. |
| LFP pack price | $81/kWh | 2025 global LFP average; 37% below NMC’s $128. China LFP cost compressed further to ~$65-70. |
| Global weighted pack price | $108/kWh | 2025 global weighted average, -8% YoY; from ~$1,200 in 2010, cumulative drop 91%. |
Solid-state roadmap
| Vendor | Tech route | Pilot | Commercial | Key milestones |
|---|---|---|---|---|
| Toyota | Sulfide all-solid | 2027-2028 | 2030 | 1,000km range · 10-min charging · launches in top-trim Lexus |
| CATL | Sulfide + semi-solid hybrid | 2027 | 2028-29 | Sample validation in 2025; CTO Wu Kai publicly confirmed timeline |
| BYD | Sulfide | 2027 | 2028 | LFP scale still the priority; solid-state is high-end supplement |
| Samsung SDI | Sulfide | 2027 | 2028 | BMW large-cylindrical → solid-state transition |
| SK On | Sulfide + polymer | 2028 | 2029 | Timeline pulled forward by one year in 2025 |
| QuantumScape | Oxide / ceramic | 2027 (C-sample) | 2028+ | B-sample in 2026 · partnered with PowerCo (VW) |
| Stellantis / Factorial | Sulfide | 2026 | 2027 | Dodge Charger Daytona is the first; China unknown |
“Pilot” means a pilot line; “commercial” means scaled mass production. All 2027-2028 timelines depend on scaled-process breakthroughs; historically the battery industry has slipped milestones like these by 1-2 years.
Key raw materials · supply chain
| Material | Price | Notes |
|---|---|---|
| Lithium carbonate (Li2CO3) | ¥95,200 /t · +25% YTD | Nov 2025 rebound peak · still below 2022 high of 600k |
| Nickel (LME) | $18,239 /t · +8% YoY | Indonesia quotas tighten · LFP squeeze on demand |
| Cobalt | ~$29k /t · +30% YoY | DRC supply disruption · weak demand |
| Graphite | under control · export controls | China started 2023-12 · suspended to 2026-11 in Nov 2025 |
Policy risk · export controls · China’s gallium / germanium / graphite / superhard-material export controls are 2025’s biggest tail risk.
2023-08 gallium / germanium → 2023-12 graphite → 2024-12 full embargo on the US → 2025-10 expansion to rare-earth magnets → 2025-11 suspended to 2026-11 after trade talks. This policy line needs continuous monitoring through November 2026. If controls are re-imposed, North American / European battery anode-material supply faces 3-6 months of real shortage (China graphite is ~80% of global supply).
§06 · Autonomy — FSD v14 · XNGP · ADS 3.0 · God’s Eye
Over the past 24 months autonomy moved from “Tesla in a class of its own” to “Tesla + Waymo (NA) + Huawei / XPeng / Li Auto / BYD (China) — six players standing”. The core narrative isn’t whose algorithm is best, but how fast autonomy becomes ubiquitous — BYD’s God’s Eye is included for free on cars at the ¥70k level; Huawei ADS 3.0 ships at ¥500k; XPeng XNGP achieved national mapless coverage in May 2024. Autonomy has moved from premium differentiator to 2026 “standard equipment”.
- Tesla FSD v13 → v14 · Robotaxi launches. Late 2024 saw FSD v13 (HW4 / AI4 only); October 2025 brought FSD v14 (incl. v14.3), with the core upgrade being a large-scale rewrite of the underlying neural-network architecture and Robotaxi’s 1.25M miles from Austin and the Bay Area being fed back to the consumer build. Robotaxi launched in Austin in June 2025, still on Model Y, with year-end coverage planned at 8-10 cities. FSD v14 · Robotaxi cumulative 1.25M miles.
- Waymo · 450k rides per week. As of December 2025, Waymo paid rides reached 450,000+ per week, with 14M cumulative rides in the year (3×+ 2024). By March 2026 it rose to 500,000/week with a 3,000+ vehicle fleet doing 4M passenger miles/week. Coverage: SF / Phoenix / LA / Austin / Atlanta / Miami / Dallas / Houston / San Antonio / Orlando. Waymo cumulative autonomous miles 100M+ (Jul 2025).
- XPeng XNGP · national mapless. May 2024 announced XNGP 100% mapless national coverage — the first Chinese start-up to hit that milestone. From 2025 it has shipped iterative E2E (end-to-end) neural-network builds, with data scale > 300M km. December 2025 received an L3 testing licence; Li Auto / Huawei followed soon after. XPeng full-year deliveries doubled (+126%); the “autonomy drives sales” narrative is fully validated. XPeng + Huawei + Li Auto · first batch of L3 in China.
- Huawei ADS 3.0 + Qiankun Intelligent. Huawei ADS 3.0 (Qiankun) launched April 2024, on a “LiDAR + light map + end-to-end foundation model” path. The four-Jie family (Aito M9 / Stelato S9 / Luxeed R7 / Maextro) ships it. ADS 3.0 selling points: cross-layer expressway, irregular-junction passage, “parking-spot to parking-spot” coverage. Huawei’s approach + Tesla’s pure-vision form the two main schools. Aito M9 2025 sales 150k+ · ADS 3.0 + HarmonyOS cabin.
- BYD God’s Eye · autonomy democratisation. February 2025 BYD launched “God’s Eye” ADAS, included for free on entry models like the ¥69,800 Seagull. It covers expressway NOA + auto-parking, in three trims (A/B/C). Independent tests show the C variant (entry) is still behind the leaders in predicting other road users, with scenario robustness to be proven, but BYD’s scale + price has accomplished the revolution of “pulling autonomy into the ¥70k price tier”. God’s Eye on 21+ models, free across the line.
- Autonomy’s “cost-curve collapse”. 2023 L2+ autonomy cost ~¥20,000/car (LiDAR + compute SoC); 2025 BYD entry-level God’s Eye estimated ~¥3,000-5,000/car, a 70%+ reduction. Drivers: pure-vision maturity + LiDAR localisation + chip substitution (Horizon Robotics, Black Sesame) + end-to-end algorithm convergence. Autonomy went from “premium differentiator” to “standard across the line”. LiDAR cost: 2020 ~$1,000 → 2025 ~$200.
Path comparison · Tesla · Waymo · Chinese
| Player | Tech path | Current level | Scale | Business model |
|---|---|---|---|---|
| Tesla FSD | Pure vision · end-to-end · HW4 | L2+ (consumer) / L4 (Robotaxi) | FSD MAU 500k+ · Robotaxi 1 city | $8k one-time / $99/mo subscription · Robotaxi by mileage |
| Waymo | Multimodal · LiDAR + HD map | L4 commercial | 3,000 cars · 10+ cities · 450k rides/wk | Ride-hailing · Uber integration |
| Huawei ADS 3.0 | LiDAR + light map + end-to-end | L2+ / L3 licence (2025-12) | “Four-Jie” family · ~400k cars | Vehicle-integrated · not separately monetised |
| XPeng XNGP | Pure vision + optional LiDAR · E2E | L2+ · L3 in application | ~400k cars · national mapless | Vehicle-integrated · low-price penetration |
| Li Auto AD Max | LiDAR + end-to-end VLM | L2+ · L3 testing licence | ~400k cars · urban NOA | Premium-trim differentiation |
| BYD God’s Eye A/B/C | Tiered · A LiDAR · B vision · C basic | L2 / L2+ | Across the line · 21 models | Free standard · sales accelerator |
| NIO NAD / NOP | LiDAR + cloud compute | L2+ | NIO + Onvo across the line | Subscription ¥380/mo |
| Xiaomi Pilot | Pure vision · E2E · HAD | L2+ | SU7 / YU7 · ~400k cars | Vehicle-integrated · HAD as premium option |
“Autonomy isn’t the new moat — it is becoming a feature like power, ABS or air-conditioning. The real moat returns in 2026 to data / compute / end-to-end algorithms, plus Robotaxi operational scale — that is a new race line.”
— A China autonomy supplier CTO, Q4 2025 industry forum
Robotaxi race · Tesla vs Waymo · December 2025 snapshot: Waymo 450k rides/week; Tesla Robotaxi just starting in Austin.
Waymo currently leads Tesla Robotaxi by roughly 2-3 years — scale, city coverage, operational data and regulatory compliance are all more mature. But Tesla has FSD data feedback from a daily million-user base, an advantage in algorithm-iteration speed and cost. The decisive period is likely 2027-2028: Tesla deploys Cybercab purpose-built vehicles (lower per-vehicle cost) against Waymo’s Jaguar I-Pace / Zeekr custom builds. The China Robotaxi market is led by Baidu Apollo Go / Pony.ai / WeRide — faster on policy cadence but still small-scale (Apollo Go in Beijing/Wuhan ~100k rides/week).
§07 · Outlook — three scenarios for the next 24 months
The past 24 months have made the EV industry’s main lines roughly clear: (1) LFP + scale + price war compresses gross margin, (2) BYD / Tesla / Chinese start-ups / Waymo — a four-pole structure, (3) US policy retreat / European protectionism rising / Chinese consolidation. Looking forward, the real variable isn’t “can the technology be done”, it’s can the market absorb it + can policy stabilise. The three scenarios below are the most-likely tracks over the next 24-36 months.
- Base · Prob 50% · China consolidation + global hits 30M in 2027. China’s “anti-involution” policy lands; 2026-27 brings 10-15 second/third-tier OEMs into consolidation; BYD / Geely / Xiaomi / Huawei-system + the three musketeers entrench; global 2026 sales ~23-24M (+15%); 2027 ~28-30M. Europe BEV share lifts to 21-22% in 2026; the US holds at 8-9% under Tesla / GM. LFP share keeps rising to 60%; pack price $95. Allocation: overweight CATL + BYD + leading autonomy suppliers; Tesla rerated by software + Robotaxi + energy; pressure persists on US start-ups (RIVN / LCID).
- Bull · Prob 25% · Global acceleration + Robotaxi commercialisation. The US new administration doesn’t restore the EV tax credit before 2028, but Tesla Cybercab scales in 2027 and Robotaxi reaches 50+ cities; China exports continue at +30%, BYD overseas at 3M; Europe BEV crosses 25% in 2026 and 30% in 2027; India / SEA / South America add 3M combined. Global hits 32M in 2027 with 35% penetration. Allocation: full overweight China EV + autonomy companies; autonomous-software valuations rerate; LiDAR / end-to-end AI / NdPr rare earth benefit.
- Bear · Prob 25% · Trade war + China overcapacity spillover. China overcapacity spills over → EU duties rise to 45-50% and the US triggers Section 301 → China 2026 exports fall 30%; China domestic demand stays weak under continued real-estate drag + anti-involution production cuts; NEV YoY only +5%; global 2026 ~22M (under expectations). BYD forced to cut staff and close plants; second/third-tier start-ups exit en masse. Tesla pivots to energy / Robotaxi to hedge. Allocation: avoid second-tier China start-ups; favour Tesla / BYD (resilient); European local OEMs benefit short-term but stay squeezed long-term; battery raw-material winter extends.
Key watch points
- Watch #1 · China NEV Monthly Retail Penetration. CPCA monthly. If from Q2 2026 it stays below 55% for two consecutive months (vs December 2025’s 59.1%), the narrative could flip from “penetration continues” to “ceiling hit” — and the entire China EV chain valuation faces re-rating. Trigger: CPCA monthly < 55% × 2 months.
- Watch #2 · BYD Monthly Exports. BYD March 2026 single-month exports were 120k (+65% YoY) — the key variable for “domestic slowdown → overseas hedging”. If any month in 2026 falls below 80k while domestic YoY stays negative, BYD valuation faces a double squeeze. Trigger: exports MoM < 80k while domestic YoY < 0.
- Watch #3 · Tesla FSD/Robotaxi City Expansion. April 2026 currently has ~2-3 cities, year-end target 8-10. If Q3 2026 still has only 3-4, Tesla’s “software / AI” valuation logic needs revisiting; if it crosses 10 cities with 1,000+ vehicles per city, the Waymo lead window closes. Trigger: Robotaxi city count · Cybercab in mass production.
- Watch #4 · US EV Market Structure. After the 30 September 2025 credit expiry, Q4 penetration fell from 10% to 5.8%. Whether full-year 2026 holds 7-8% is the floor for the US EV industry not breaking; if it falls below 5%, Rivian / Lucid / Ford EV investment-payback assumptions need a rewrite. Trigger: KBB Q2/Q3 2026 monthly penetration < 6%.
- Watch #5 · EU — China EV Price Undertaking. EU-China negotiations to replace duties with a “price-undertaking” arrangement haven’t closed. If a deal is signed mid-2026 (with floor pricing), Chinese exports to Europe become “higher-margin but volume-capped”; if talks fail and duties continue, Chinese OEMs will route around to local manufacturing (BYD Hungary · Geely MG · Chery Zetor). Trigger: EU-China MOU signing · local capacity utilisation.
- Watch #6 · Battery Price / Lithium Recovery. BNEF expects further pack-price downside in 2026, but lithium rebounded to ¥95,200/t in November 2025. If lithium rises another 30%+ in 2026 (back to ¥130,000), the pack-price decline curve breaks, the $95 target becomes hard, and EV gross-margin recovery stalls. Trigger: battery-grade Li2CO3 > ¥130,000/t for a quarter.
Bottom Line · Portfolio Takeaway · Allocation logic for the EV industry in 2026 has shifted from “pick the track” to “pick who survives”.
- Core holdings (traditional path): BYD + CATL + Tesla — three legs corresponding to “scale · battery · software” moats. Even with short-term pressure in 2026, all three have through-cycle cash flow.
- Start-up picks (alpha source): Xiaomi · XPeng · Huawei-system (Seres · Hongmeng-Auto); avoid NIO / Li Auto, dependent on a single price-tier lineup, until they finish their refit.
- Waymo (Alphabet) and Tesla Robotaxi AI valuation is the most-likely “new AAPL / NVDA-level” story for 2026-2028 — premium space well beyond the auto industry itself.
- Avoid: US start-ups (RIVN / LCID) — business model unproven post tax-credit; European EV start-ups (Polestar / Fisker bankruptcy reorg) with fragile capital structure; Japanese / Korean battery makers passive against the LFP wave.
- Tail hedges: lithium miners / graphite / rare-earth magnets / charging infrastructure (EVGO / ChargePoint) — non-linear-payoff assets if a supply-chain risk pulse appears.