The module layer between the cell and the pack enclosure: both companies looked at it and called it dead weight, structural filler that ate volume and mass and gave nothing back to the electrochemistry. CATL and BYD reached cell-to-pack architecture independently around 2020, then split on almost everything after, which chemistry to bet on and what even counted as a problem worth solving. Six years on, those early bets have hardened into two companies that build batteries barely resembling each other.
BYD’s Blade launched in March 2020 on a lithium iron phosphate (LFP) cathode, and LFP’s olivine crystal structure locks oxygen into the lattice far more tightly than the layered oxides in nickel-manganese-cobalt cathodes. Push an LFP cell into thermal runaway, and the onset comes at 220 to 260°C, compared to 170 to 210°C for NCM, and the heat it dumps is roughly 75 to 85% lower because there’s no oxygen liberation feeding the fire. Field data lands in the same place: five to six times fewer thermal events per million miles than conventional lithium-ion. Wang Chuanfu built the entire Blade launch around the nail penetration test, the de facto proxy for battery safety, in which Blade cells topped out at 30 to 60°C with no fire or smoke. In comparison, NCM cells routinely blew past 500°C under identical conditions.
LFP’s theoretical specific-capacity ceiling sits below NCM’s, and the Blade Gen 1 didn’t hide what that costs: around 166 to 168 Wh/kg at the cell, roughly 150 Wh/kg at the pack, volumetric density of 424 to 448 Wh/L at the cell. Respectable for LFP in 2020, and well short of what premium NCM packs were already delivering.
Qilin CTP 3.0, announced in March 2022 and entering mass production by 2023, ran the opposite play. The flagship NCM variant targeted 255 Wh/kg at the pack level in CATL’s own March 2022 figures, a number its marketing later bumped to 265 Wh/kg. Then, the teardown of the Zeekr 140 kWh pack, built on a 285 Wh/kg NMC cell, measured about 193 Wh/kg at the pack level. That gap between the official claim and the wrench-in-hand measurement is large, and it’s worth carrying into every headline number any of these companies publishes. CATL also sells a Qilin LFP variant at 160 Wh/kg pack level, dead level with its own CTP 1.0, which tells you the structural trick pays off far more on NCM because the chemistry has more headroom to give back.
BYD cools the Blade with a single large plate sitting on top of the cells. The cells are prismatic and absurdly long, roughly 960 mm by 90 mm by 12 mm thick, each one spanning the full pack width, so the plate covers a large area in absolute terms but touches only one face of each cell. Critics have noted that it has less cooling area per unit of cell volume than rival layouts; BYD’s standard reply is that LFP runs cool enough that you don’t need to throw cooling at it.
NCM811 cells, with nickel at 80%, run hotter and are more reactive than the older blends, so CATL dropped its liquid-cooling plates between adjacent cells rather than underneath them. That large-face approach quadruples the effective heat-transfer area over the prior generation, and CATL claims it halves thermal-control time. The plates pull double duty as the structural spacer, which lets CATL delete the internal crossbeams entirely; cells and cooling sandwich into a self-supporting block. Volume utilization hits 72%, which CATL bills as the highest of any production pack.
Prismatic cells swell as they age. Lithium plating, electrolyte decomposition, gas generation in the cell, SEI growth on the anode, all of it adds thickness over thousands of cycles. In Qilin, those swelling cells press on the elastic interlayer cooling plates and pinch the coolant channels shut. Internal resistance climbs with age, too, so the cell generates more heat exactly as the cooling that should answer it gets mechanically squeezed off. That positive feedback loop is what Batterydesign.net flagged as the underexamined risk in an otherwise celebrated design, and whether CATL’s interlayer material maintains channel geometry under realistic expansion loads isn’t documented anywhere in the public domain. Pulling the crossbeams incurs a second cost: the cells vent downward, and with no central structural members, the side-impact load paths become harder to reason about. Traditional packs rely on steel crossbeams that moonlight as intrusion barriers; Qilin’s cooling-cell sandwich has to carry those loads instead, and no one outside CATL has published the margins.
The long, thin Blades replace the steel crossbeams outright. Stacked like blades in a honeycomb and spanning the pack width, they deliver compressive and bending stiffness that ordinary stubby prismatic cells never could, which is how BYD gets the cells themselves to carry the structure. BYD claims 50% better space utilization than a traditional block design; the aluminum casing adds rigidity, and in some implementations, the enclosure ties straight into the body. Validation of the original Blade went well beyond the nail test that made headlines. They crushed it, bent it, ran it to 300°C in a furnace, drove it to 260% overcharge, and Gen 1 came through with no fire or explosion. The nail result mattered most because it showed how the cell behaves under an internal short, the nastiest failure mode there is, and that behavior was a different animal from NCM entirely.
Gen 2 Blade landed March 5, 2026, still LFP but now with fourth-generation cathode materials and a silicon-carbon anode. Silicon stores roughly 10 times as much lithium as graphite does, with the old catch that it swells hard during lithiation and mechanically tears itself apart; BYD’s fourth-gen materials tame that enough to push system energy density to 162 Wh/kg, about 5% over Gen 1 at the pack. It charges from 10% to 70% in five minutes and from 10% to 97% in nine minutes on a 1,000V platform at 10C and 1 MW peak. Cell cost runs about RMB 0.65/Wh. Validation ran 500 flash-charge cycles, standing in for roughly 15 years of use, then drove a nail through the cell while it was actively charging, which it survived without fire or smoke, plus a four-cell simultaneous short that left the neighbors stable.
CATL’s CTP line ran from CTP 1.0 in 2020 (160 Wh/kg LFP, over 55% volume utilization) through a CTP 2.0 waypoint to CTP 3.0/Qilin in 2022 and 2023. Gen 3 Qilin, unveiled April 21, 2026, at the company’s “Super Tech Day” in Beijing, pushes cell energy density to 280 Wh/kg with 10C superfast charging and 3 MW peak power, double the roughly 1,330 kW of the previous generation. The pack comes in at 625 kg, which CATL says is 255 kg lighter and 112 liters smaller than an LFP system covering the same range. The headline safety architecture is “thermal-electrical separation”: every cell gets its own sealed exhaust channel, so a runaway in one cell can’t propagate down shared venting. CATL also rattles off a stack of vehicle-level deltas against a baseline it hasn’t fully published: 0 to 100 km/h shaved 0.6 seconds, 12% shorter overtaking distance, 8% higher moose-test speed, about 1.44 meters off the braking distance, and over 6% lower consumption per 100 km.
The original Shenxing, August 2023, was the world’s first 4C superfast LFP: 400 km from a ten-minute charge, over 700 km total. It’s CATL’s fast-charge LFP track, running alongside Qilin rather than under it. Shenxing PLUS (April 2024) broke past 200 Wh/kg at the pack to hit 205, a figure the industry had treated as LFP’s practical ceiling. The second generation (April 2025) reached a peak of 12 °C and 1.3 MW. Shenxing Pro (September 2025) hit 758 km WLTP with a claimed 12-year/1,000,000 km life and only 9% degradation at 200,000 km. Gen 3 Shenxing (April 2026) runs from 10% to 80% in 3 minutes 44 seconds, from 10% to 98% in 6 minutes 27 seconds, and maintains over 90% capacity after 1,000 cycles. At −30°C, it goes 20% to 98% in about nine minutes, and cold-weather fast charging has been LFP’s chronic weakness, so that last number is the one that earns its place.
The Qilin Condensed Battery, announced on the same April 21, 2026, runs a different electrolyte system entirely, rather than a polished CTP 3.0. The condensed cell uses a gel-like semi-solid electrolyte, which CATL calls biomimetic, sitting between liquid and true solid-state on the spectrum, with no liquid to leak and nothing flammable to ignite. That addresses the combustion risk at the chemical level rather than containing it with engineering. The cathode is high-nickel for density, the anode low-expansion silicon-carbon, and the casing aviation-grade titanium alloy, 60% thinner and 30% lighter than aluminum at three times the unit strength, worth about 20 Wh/kg on its own. Result: 350 Wh/kg and 760 Wh/L at the cell, pack held inside 650 kg. Charging runs from 10% to 98% in 6 minutes 27 seconds at an equivalent 10C, peaking at 15C. Cell cost is about RMB 0.78/Wh, a 20% premium over BYD’s Gen 2 Blade at RMB 0.65, for roughly 2.2 times the energy density.
CATL showed a 500 Wh/kg condensed prototype back in April 2023 and flew a four-ton electric aircraft on condensed cells in 2024, so the program runs older than the 2026 reveal lets on. The 350 Wh/kg Qilin Condensed is the production-ready, vehicle-grade version, aimed where mass and volume constraints turn punishing enough to swallow the cost premium, which today means high-end EVs and the flying-car and aviation fringe. CATL calls it semi-solid, not all-solid-state, and the distinction earns its keep: all-solid-state hits manufacturing walls at scale that semi-solid mostly walks around.
For pure reference, the five product lines are up like this:
| Parameter | BYD Blade Gen 1 | BYD Blade Gen 2 | CATL Qilin CTP 3.0 (NCM) | CATL Qilin Gen 3 | CATL Qilin Condensed |
|---|---|---|---|---|---|
| Chemistry | LFP | LFP + Si-C anode | NCM | NCM | High-Ni + Si-C, semi-solid electrolyte |
| Cell energy density | ~166–168 Wh/kg | ~162 Wh/kg (system) | ~285 Wh/kg (cell); ~193–255 Wh/kg (pack, varies by source) | 280 Wh/kg (cell) | 350 Wh/kg (cell) |
| Pack energy density | ~150 Wh/kg | 162 Wh/kg | 255–265 Wh/kg (official); ~193 Wh/kg (measured) | Not separately published | Controlled within 650 kg pack |
| Volumetric density | 424–448 Wh/L (cell) | Not published | 72% volume utilization | Not published | 760 Wh/L (cell) |
| Fast charge | Not specified | 10%→97% in 9 min (10C, 1 MW) | 10%→80% in 10 min | 10C, 3 MW peak | 10%→98% in 6 min 27 sec (10C, 15C peak) |
| Cycle life | 3,000+ | Not published | Not published | Not published | Not published |
| Cell cost | Not published | ~RMB 0.65/Wh | Not published | Not published | ~RMB 0.78/Wh |
| Cooling | Top-surface plate | Top-surface plate | Between-cell large-face | Between-cell + thermal-electrical separation | Not detailed |
| Thermal runaway onset | 220–260°C (LFP) | 220–260°C (LFP) | ~170–210°C (NCM) | ~170–210°C (NCM) | Electrolyte non-flammable |
| Volume utilization | 50% better than traditional | Not published | 72% | Not published | Not published |
| Key applications | BYD internal brands | Sealion 06, Seal 07 | Zeekr, Li Auto, Xiaomi, Aito | Announced, not yet deployed | High-end EVs, aviation |
BYD’s safety claim lives in the material: LFP’s olivine structure makes catastrophic runaway chemically improbable because the oxygen that feeds the fire isn’t there to release. That’s a materials-level guarantee, and it doesn’t care whether the cooling works or the BMS is even awake. It holds even if every engineering control fails at once.
CATL’s standard NCM Qilin takes the opposite bet. The chemistry runs hotter, so CATL out-engineers it with independent per-cell exhaust channels and large-face cooling between the cells, the NP 2.0 management layer riding herd on all of it, enough control to keep the reactivity bounded. It’s a legitimate position, and CATL’s February 2025 certification as the first battery to meet China’s “No Fire, No Explosion” standard, more than a year ahead of the 2026 effective date, shows the controls do their job in testing. What’s unproven is whether engineering-level safety will remain as durable a moat as chemistry-level safety once energy densities keep climbing and operating conditions get rougher. The condensed battery sidesteps the whole question by moving the guarantee back into the chemistry, a non-flammable electrolyte that kills the combustion path regardless of what happens to the cell. That lands the Qilin Condensed closer to BYD’s LFP than to CATL’s own NCM Qilin, except it pulls it off at 350 Wh/kg.
CATL ran 37.9% of the global EV battery market in 2024 on 339.3 GWh installed, more than double its nearest rival, and that share edged to 38.1% through October 2025. In China alone, its March 2026 installed share hit 45.54%. BYD held 17.2% globally in 2024, slipping to 16.9% through October 2025, with a 17.83% share in China in March 2026. Between them, the two control roughly 55% of the global supply and about 69.5% of the Chinese domestic market.
CATL supplies about 70% of vehicles priced above $35,000 worldwide, leveraging NCM’s density advantage in the premium and performance tiers,r where cost and complexity pay off. BYD burns roughly 90% of its battery output internally and sells only 10 to 20% outside, mostly to Toyota, which makes its battery tech a captive weapon rather than a revenue line. CATL’s model runs the other way, supplying Tesla, the major German and American automakers, Hyundai, and dozens of Chinese OEMs.
Revenue fell 9.7% year-over-year to RMB 362 billion (about $52 billion) in 2024 as lithium carbonate slid from roughly 100,000 to 75,000 yuan a ton, dragging CATL’s average selling prices down 25.3%, the cost of running a third-party supply model straight through a chemistry-price collapse. Net profit still climbed 15% to RMB 50.7 billion (about $7 billion), gross margin widening 5.3 points to 24.4% and peaking at a record 31.17% in Q3 2024. Q1 2026 net profit of 20.74 billion yuan was up 48.52% year-over-year. CATL raised HK$41 billion in a Hong Kong secondary listing in May 2025, the largest global IPO of the year, and sits on over RMB 300 billion in cash, backed by a 20,000-strong R&D organization and 43,354 patents worldwide.
Projected manufacturing capacity hit roughly 4,800 GWh by 2025 against demand near 1,200 GWh, a four-to-one glut grinding the roughly 50 sub-tier manufacturers toward consolidation. CATL holds 68% of China’s NCM market and 37.2% of its LFP market, making it dominant in both chemistry families.
BYD plans 20,000 flash-charging stations by the end of 2026, with 1,500 kW peak per gun, open to every EV brand, offering a year of free charging for first-batch owners, and sites that bundle solar and local storage with the grid connection to dodge big grid upgrades. CATL is building 4,000 “super swap-integrated” stations that combine charging and battery swapping, aiming for over 85% equipment utilization. The five-to-one station gap matters less than the fact that BYD is opening its network to all comers, which flips it from a captive battery supplier into a public-infrastructure player. This perch can sway which car someone buys, regardless of which chemistry wins on the bench. CATL’s swap bet needs cross-platform standardization that nobody has reached yet; NIO proved swapping works within one brand’s walls, and nobody has proved it scales beyond that.
CATL has thrown over 1,000 researchers and more than 10 billion yuan at solid-state, with a sulfide-electrolyte prototype at 500 Wh/kg, small-batch production targeted for 2027 and large-scale for 2030. Its condensed semi-solid is already shipping, an intermediate rung BYD doesn’t have. BYD has been in solid-state since 2013, built its first vehicle-grade cells (20 to 60 Ah) in 2024, targets 400 Wh/kg, and plans to demonstrate vehicles in 2027, with mass production in 2030. SMM projects all-solid-state shipments at 13.5 GWh by 2028 and semi-solid shipments at 160 GWh by 2028, against global solid-state penetration of around 0.1% in 2025, rising toward roughly 4% all-solid-state by 2030 and about 10% including semi-solid by 2035. Underneath all of it, global lithium-ion demand is projected to be near 2,800 GWh by 2030, EVs growing at roughly 11% CAGR from 2024 to 2030, storage at 27%, and consumer electronics at 10%.
CATL’s first sodium-ion generation, July 2021, delivered 160 Wh/kg with 80% charge in 15 minutes and operated down to −20°C, marking the start of the Naxtra line that now forms a third chemistry track. The 2025 Naxtra reached 175 Wh/kg, the highest sodium-ion density anywhere, with GWh-scale mass production aimed at the end of 2026 once four manufacturing bottlenecks are cleared: extreme water control in production, gas generation in hard-carbon anodes, aluminum-foil adhesion, and self-forming anode systems. Sodium’s case is low-cost, cold-climate stationary storage, where lithium’s price and temperature sensitivity become liabilities.
BYD owns a defensible spot in mass-market LFP, real fast charging now, and a RMB 0.65/Wh cost structure that’s brutal to undercut, with a charging network that reaches well past its own badges. CATL is the harder company to box in: dominant share in both NCM and LFP, the only vehicle-grade semi-solid battery actually in production, and a lock on the premium segment. BYD is only now elbowing into the market with silicon-carbon LFP. They collide most directly in the premium mass-market band, RMB 200,000 to 400,000, and that’s where the chemistry-versus-engineering safety question gets settled for real. It comes down to one variable: whether CATL’s interlayer cooling holds its channel geometry as those NCM cells swell over a real vehicle lifetime. Get that wrong, and the densest pack on the market is the one that ages the worst.
Related reading: Xiaomi, now straddling flagship phones and its own EV line, leans on exactly these battery suppliers.