On September 16, at CATL's direct-sales experience center in Beijing, a 2019 GAC Aion S that had covered 770,000 km (about 478,000 miles) was lifted onto a teardown platform. As a ride-hailing vehicle, it had been charged and discharged almost daily for six years, yet its battery health remained at 88%. Another vehicle, a Geely Geometry A with 700,000 km (about 435,000 miles), showed a battery health of 90%. Upon opening, neither battery showed signs of water ingress, corrosion, or abnormal swelling, contradicting common assumptions about aged batteries, according to D1EV, a Chinese EV news outlet.
The closed-door teardown came just nine days after Li Auto announced a full pivot to self-developed batteries. In the context of the industry's "de-CATL" movement, the event seemed like a response from CATL: batteries with similar specs at launch can diverge greatly after six or seven years, because the true determinants of battery life are capabilities that cannot be written into a specification sheet.
The limits of self-development
Li Auto's decision to switch battery suppliers for its MEGA model was straightforward: CATL's 5C ternary batteries were nearly out of stock. CATL's capacity utilization hit 94.86% in the first half of the year, with high-rate products in short supply. For high-volume models, missing supply is more dangerous than higher prices. But scarcity was only the trigger; batteries are among the most expensive components and directly affect range, fast charging, and space. Whoever defines the battery controls cost and product experience.
Li Auto began cell R&D in 2020, adopted Sunwoda cells for its 2024 L7, and later formed a joint venture and took a stake in Sunwoda. However, its "self-development" does not mean building its own factories; Li Auto defines the formula, packaging, and BMS algorithms, while Sunwoda manufactures. Xiaomi's Longjing battery, produced by CALB and Sunwoda, is defined by Xiaomi, which controls manufacturing standards and quality. Huawei's Giant Whale battery platform also does not produce cells but sets rules for its HarmonyOS suppliers.
Most automakers, however, pursue "not only CATL" rather than "no CATL." Leapmotor uses multiple suppliers but still relies exclusively on CATL for its flagship D19. Xiaomi's high-spec models and HarmonyOS vehicles continue to source from CATL. Li Auto goes further, attempting to exit CATL's supply chain entirely, but that path is not feasible for everyone. Exclusive cells require large sales volumes, and custom formulas demand years of data and talent. Moreover, for cutting-edge products like 6C, high-nickel, and silicon-carbon cells, few suppliers can mass-produce reliably, and CATL remains a selling point for many models.
What CATL is defending
In the first half of this year, CATL's share of China's passenger EV battery installations rose to 46.7%, up 5.6 percentage points year-on-year. Yet, from September 15, CATL's stock fell over 9% in two trading days, reflecting concerns about future supply relationships. Automakers are reclaiming product definition, battery packs, and BMS, which are close to the vehicle and better suited for automakers to control. CATL aims to prove that even if these capabilities are taken away, some things inside the battery pack remain difficult to replicate.
Battery aging involves SEI film rupture and repair, lithium plating on the anode, and cathode particle damage, each consuming available lithium and active material, leading to capacity loss, increased resistance, and swelling. Adjusting graphite, coating particles, and combining additives require iterative refinement across materials, equipment, and production data. Thus, identical specs do not mean identical capabilities; specification sheets can state what a battery achieves, but not how much margin remains under extreme conditions. For example, during fast charging, internal temperature differences can exceed 10°C, but sensors are limited in placement and may not capture the hottest spots. Cells also expand and contract with each cycle, thickening over years, so structural design must accommodate this change.
Of course, the invisible is not forever unlearnable. Sealing, corrosion protection, thermal management, and expansion allowance are mechanical and systems engineering, and companies like Li Auto, Xiaomi, and Huawei are incorporating these lessons into their design and testing processes. The truly difficult part is the underlying cell chemistry and the consistency across mass production. CATL's chief manufacturing officer, Ni Jun, said a vehicle's range and lifespan depend on the weakest cell in the pack. Making one excellent cell in the lab is different from ensuring billions of cells maintain the same quality.
Once a battery fails, it is hard to determine whether the issue lies in the cell, the pack, or the charging strategy. Geely's Viridi and Sunwoda had a dispute over this: Viridi claimed some cells delivered between 2021 and 2023 had voltage platform and internal resistance consistency issues, while Sunwoda said the problems were related to Viridi's pack design and charging strategy. The dispute does not prove second-tier suppliers are generally inferior to CATL, but it shows that long-term battery performance is not determined by a single factor; it is a coupling of cell, packaging, thermal management, and control strategy.
As automakers reclaim definition rights, they also take on more quality responsibility. When problems arise, allocating responsibility across components becomes harder. The new national standard GB 38031-2025, effective July, adds safety tests after fast-charge cycling, and a September group standard requires batteries to undergo mechanical, environmental, and cycle aging before safety tests. This is progress for the industry, but for CATL, the brand trust it once provided is being converted into common contract terms and test metrics, thinning its premium. Higher barriers do not mean every company can achieve the same yield and cost.
Ultimately, the "de-CATL" discussion will be settled by degradation curves, warranty claims, recall records, insurance data, and used-car residual values in the coming years. CATL's moat is not an insurmountable wall but a distance that requires time, capital, and vast vehicle fleets to close. The teardown of an old battery reveals the craft accumulated over a decade. Now, the industry is dissecting "de-CATL" to see how automakers are reclaiming initiative layer by layer. The first to be taken was exclusive supply status, then product definition, packs, and BMS. Now, companies like Li Auto are venturing into cell formulas. But from defining a battery to mass-producing billions of cells that endure years of use, there remains a gap that only real-world fleets can validate. CATL is losing its near-default status, and passenger vehicle battery margins will decline with competition. Yet "CATL is no longer the only choice" does not mean "CATL can be easily replaced." The former has already happened; the latter still requires crossing the thresholds of cell chemistry and manufacturing consistency. Time will answer.
