What Exactly Is Hidden Inside Geely's Q3 Report Cost Structure?

On October 29, Geely released its financial report for the third quarter of 2026.

The report shows that Q3 revenue was 99 billion yuan, up 10% year-over-year (YoY) and up 11% quarter-over-quarter (QoQ); core net profit attributable to the parent company was 5.21 billion yuan, up 32% YoY and up 2% QoQ, hitting a historical high. Within the 5.21 billion yuan core net profit, the core net profit attributable to the parent company per vehicle for the first three quarters was approximately 6,664 yuan, up 37% YoY.

In an era where the profit margin of the entire industry's vehicle manufacturing has fallen to single digits and the profit per vehicle for the vast majority of automakers continues to decline, placing the two figures "10% revenue growth rate" and "37% core profit per vehicle growth rate" side by side is the true message Geely's Q3 report aims to convey — rewriting the cost structure.

To understand this main thread, one must first look at a more basic comparison. In the first three quarters, Geely's total sales volume was 2.235 million vehicles, up 3% YoY; revenue was 272.6 billion yuan, up 13% YoY; core net profit attributable to the parent company was 14.9 billion yuan, up 41% YoY.

The growth rates of sales volume, revenue, and profit increase sequentially, forming a ladder of "Core Profit Growth Rate > Revenue Growth Rate > Sales Volume Growth Rate". This is not only extremely rare in the 2026 automotive market but also forms a sharp contrast with the common situation of peers where "sales volume is maintained by price cuts, revenue is barely held up by sales volume, and profits are squeezed from both ends".

It can be seen that, significantly distinguishing from the traditional habitual understanding of the past, the relationship between Geely's profit margin and sales volume is no longer a simple linear relationship.

The core mechanism behind this is closely related to what Geely mentioned in its Q3 brief as a "Leader in Global AI Intelligent Vehicles". Geely is already using a complete AI vehicle building logic to rewrite the cost structure of traditional automakers, which was "each car model developed independently", into a "one platform, multi-brand reuse" structure.

The basis for this judgment is reflected in the "divergence" of R&D investment in the Q3 report. Total R&D investment for the first three quarters was basically flat YoY, but the R&D investment ratio dropped by 0.7 percentage points to 5.5%.

Investment was not cut, but the expense ratio dropped. This phenomenon is most reasonably explained by the improvement in the reuse rate of R&D results. In practice, Geely's global AI technology system, formed with "1+2+N" as the core, based on the WAM (World Activity Model), has achieved full integration of the intelligent driving domain, cockpit domain, and chassis domain.

After the four major AI architectures modularized the underlying capabilities of the whole vehicle, the cost of flow of technical results between different brands and different models will decrease significantly. When R&D for different brands and different models is no longer "starting from scratch", but rather doing differentiated configurations on a unified "vehicle brain", multi-brands have transformed from a cost burden into a lever for R&D efficiency.

As the absolute value of R&D investment is controlled within a certain range, with the rapid expansion of the number of supported brands, models, and markets, the speed at which the R&D denominator is diluted is faster than the speed of growth of the R&D numerator.

Two other lines on the expense side provide corroborating evidence. Administrative expense ratio decreased 0.1 percentage points YoY to 1.6%, and sales expense ratio decreased 0.1 percentage points YoY to 5.6%. Against the backdrop of high-speed expansion in overseas markets and four brands focusing efforts simultaneously, all three expense lines decreased across the board.

Here, two types of completely different cost improvements must be distinguished. One is optimization "saved" through price suppression in procurement, cutting expenses, and streamlining organizations. Although direct, it decreases with each use, has a clear ceiling, and often comes at the cost of sacrificing long-term capabilities.

The other type is "cost derived from structure", where efficiency improvements brought about by changes will continue to amplify as scale expands — Geely's expense ratio decrease in Q3 obviously belongs to this category.

Zeekr is the most direct footnote to this mechanism. In Q3, Zeekr delivered 110,000 vehicles, with a cumulative 288,000 vehicles for the first three quarters, experiencing double-digit growth YoY and QoQ for 8 consecutive months. The annual target of 300,000 vehicles has been completed by 96%. Single-month delivery in September broke 37,000, doubling YoY growth.

In the severe environment of this year where brands trade volume for price, Zeekr has charted a path of "simultaneous increase in volume, price, and profit". Its significance does not lie in "selling at a high price", but in that it bears the verification cost of the most cutting-edge technology, while the marginal cost of reusing the verification results approaches zero.

One year after the launch of Zeekr 9X, it still achieved over 7,000 monthly sales in the 500,000-level market, ranking first in sales among vehicles over 500,000 yuan domestically from January to September; Zeekr 7X broke 10,000 for three consecutive months, and Zeekr 007GT maintained monthly sales between 7,000 and 10,000. Zeekr has simultaneously succeeded in the routes of high-end pure electric and hybrid, SUV and sedan. This means the verified results are not exclusive to a single model and can be reused between different powertrain forms and different body shapes.

Specifically in the Q3 report, Zeekr's sales volume accounted for 12.9% of total sales volume for the first three quarters, an increase of 6.3 percentage points compared to the same period last year; revenue share reached 31.3%, an increase of 13.6 percentage points compared to last year, with revenue share far higher than sales volume share.

This is in itself the financial confirmation of the R&D reuse mechanism at the brand level. A platform development and architecture extended from a base generates revenue returns on the Zeekr brand far higher than other brands, while R&D costs have not increased proportionally.

The entire mechanism is not established only on Zeekr; overseas business is another clear aspect revealed in the Q3 report.

Q3 overseas export sales volume was 323,000 vehicles, up 189% YoY and up 19% QoQ; first three quarters surged 169% YoY to 798,000 vehicles, of which new energy vehicle exports reached 484,000 vehicles, up 531% YoY. Q3 overseas sales share was about 40%, overseas monthly sales exceeded 100,000 vehicles for four consecutive months, and Geely's overseas export growth rate was first among mainstream automakers in Q3.

"Acceleration" needs to be understood within the 2026 overseas export environment. Geely faces geopolitical game theory at this time, including the EU's extra countervailing tariffs on China's plug-in hybrid vehicles entering the implementation phase, multiple Southeast Asian countries tightening foreign investment factory policies, and a general increase in global trade barriers, etc.

Yet at this time, Geely's "Industry Symbiotic" asset-light model showed its ability to resist pressure. And the impact of this model on the Q3 cost structure has three real transmission chains.

The first is saving fixed costs.

Building factories with heavy assets means high depreciation, long payback periods, and risks of idle capacity. These sink into the cost structure to become long-term burdens. Although this path is widely adopted, Geely did not take it.

The "123456 Strategy" released in mid-August clarified that the core idea is not to build factories independently, but to build a Geely overseas system, creating five major regions: Greater Europe 600,000, ASEAN 500,000, Latin America/Africa 400,000, Eastern Europe 300,000, and Middle East/Central Asia 200,000, using cooperative capacity to replace own capital expenditure.

On the path, Geely cooperated with Ford, Volvo, etc. in Europe, completing a systematic capacity layout around GEA and SEA vehicle architectures; in ASEAN, relied on the Malaysia Proton factory to build an industrial hub; in Latin America, deeply bound Renault resources; in Eastern Europe/Central Asia, entered quickly via OEM KD methods.

First three quarters, Geely put 13 overseas factories into production. While overseas capacity exceeded 840,000 vehicles, this part of rigid expenditure was also converted into variable costs, making the cost structure naturally lighter.

The second is to avoid tariff costs.

EU countervailing tariffs are essentially a fee added to export costs. The common point of a series of actions such as acquiring partial equity in Ford's Spanish factory, Volvo becoming Lynk & Co's exclusive European distributor, and the first local production result of Renault Geely Brazil Company, the Geely EX5 EM-i, going offline, is that Geely is promoting localization from "sales localization" to "manufacturing and channel localization", upgrading tariff avoidance from "borrowing" to "co-construction", further improving the stability of the cost structure.

The third is to amplify the R&D amortization denominator.

The same set of R&D investment, selling 100,000 vehicles domestically and 300,000 vehicles globally, results in completely different amortization costs per vehicle.

Geely has currently created "4 sets of 3" global best-seller cars. China Star uses Coolray, Emgrand, Boyue to stabilize the fuel base; Galaxy uses Xingyuan, Galaxy E5, Starship 7 to form global volume synergy; Lynk & Co uses 08, 02, and 900 to drive; Zeekr focuses on 7X, 8X, 9X. The R&D investment for every global model is continuously diluted on top of the Q3 export volume of 323,000 vehicles.

Connecting the three chains together, one will also find the tightness of Geely's asset-light model in internal collaboration.

The reason why this model can have such a big impact on the cost structure is that the prerequisite is still that the R&D side has achieved cross-brand, cross-market reuse. If a new adaptation plan needs to be developed every time entering an overseas market, the fixed costs saved by the asset-light model will soon be eaten up by adaptation costs.

Regarding high pricing ability in overseas markets, Zeekr 9X has already begun sales in developed European countries such as Germany, France, and Sweden, with a starting price nearing 1 million; Zeekr brand export sales volume in August and September both broke 15,000 vehicles, with exports contributing over 40% to the monthly sales for those months, firmly ranking first among Chinese high-end brand exports.

What these reflect is not the optimization of the cost structure, but the confirmation of revenue quality. The cost side makes the denominator smaller, the revenue side makes the numerator larger, and the two jointly support the increase in profit per vehicle.

Of course, Geely's AI vehicle building rewriting of the cost curve is currently still mainly concentrated on the decrease in R&D expense ratio and reduction in vehicle amortization, and has not yet significantly become an independent variable on the Q3 income statement. However, the scaled supply of AI to consumer experience is currently happening and generating positive impact.

The progress of i-HEV Intelligent Engine Hybrid can be concretized as another facet of this logic. This technology, through an architecture of "Electric Drive Dominated, Oil-Electric Synergy", achieves over 80% electric drive, with actual fuel consumption of 2.22L/100km, more than 12% lower than the Japanese hybrid benchmark.

Currently, the China Star fuel series is comprehensively popularizing i-HEV. As of September, cumulative sales reached nearly 60,000 vehicles, occupying over 40% of Chinese brand HEV vehicle sales. Its cost significance lies in that, even if the fuel vehicle overall market declined severely, China Star still sold 293,000 vehicles in Q3, and with the sales performance of 874,000 vehicles in the first three quarters, it firmly ranked first in fuel vehicle sales. Obviously, this base is not sustained by price cuts, but by technological upgrades.

At the end of September, Geely released Megawatt-level Intelligent Charging technology, with a single gun power reaching 2.2 MW; 800V high-voltage platform has covered the 120,000 to 300,000 yuan mainstream price range, with main models such as Galaxy TT and the new Lynk & Co 20 already equipped; in terms of charging infrastructure, an "Person, Vehicle, Pile, Cloud, Network, Storage" AI charging network was built, achieving strategic cooperation with NIO, and carrying out open sharing and industrial synergy in the charging and battery swapping fields.

As of early October, Qianli Haohan accumulated 1.073 million users, cumulative mileage reached 3.14 billion kilometers, activation rate reached as high as 93.9%, accumulated avoidance over 12.18 million times; Flyme Auto accumulated installations in the first three quarters exceeded 1.326 million vehicles, global users exceeded 3.59 million. In Q4, Qianli Haohan G-ASD will undergo a comprehensive upgrade, supporting city roaming NOA, end-to-end AEB, and natural voice vehicle control.

Qianli Haohan's 1.073 million users, Flyme Auto's 3.59 million global users, 800V covering mainstream price ranges, Megawatt intelligent charging and NIO's charging synergy, these are all counted as infrastructure and user entry points for scaled implementation of AI capabilities, and also determine that AI is transforming from "indirect contribution on the cost side" to "attributable contribution on the revenue side".

Rewriting the cost structure also requires a business prerequisite: multi-brands must not step on each other. If the four brands consume each other within the same price range, the efficiency of R&D reuse will be offset by internal competition, and the scale advantage of asset-light overseas expansion will also be eaten up by channel duplicate construction.

Data from the Q3 report shows that this prerequisite is already in place. Galaxy has seen consecutive QoQ growth since August, with double-digit YoY and QoQ growth in September, exceeding 130,000 in a single month. Galaxy E5 broke 10,000 monthly sales for four consecutive months; Galaxy Starship 7 increased continuously to break 22,000 in September; Xingyuan reached 1 million vehicles produced within two years of launch, breaking the global speed record for single model million-vehicle production.

Lynk & Co has stabilized QoQ growth since August, with September sales volume exceeding 18,000 vehicles, new energy ratio exceeding 90%, and new energy ratio in the first three quarters exceeding 70%. China Star September sales volume broke 100,000 vehicles, with double-digit YoY and QoQ growth. Zeekr September deliveries broke 37,000 vehicles, doubling YoY.

The fact that the four brands achieved simultaneous YoY/QoQ improvement in Q3 is no coincidence.

Corresponding to the move to set up a sales headquarters in early August to integrate marketing resources of its brands, and facts such as Galaxy TT achieving delivery of over 10,000 vehicles within 30 days of launch, Galaxy Warship 700 achieving over 10,000 orders within 12 minutes of launch, with high-spec version accounting for 90%; Lynk & Co 07 GT achieving delivery of over 10,000 vehicles within two months of launch, and the new Lynk & Co 20 achieving over 14,600 orders within 1 hour of launch.

It can be found that Geely's integration at the organizational level allowed the four brands to avoid internal consumption during the most intense period of the price war, concentrating resources on their respective favored price ranges and market segments.

A deeper organizational change than setting up a sales headquarters occurred in mid-August. An Conghui succeeded Li Shufu as Chairman of the Board of Geely Automobile. This company has formally moved from the "founder-driven" startup and development phase to the "system and team-driven" mature phase.

How to transition from founder governance to system governance is a question all successful enterprises must answer. The founder's strong judgment is an irreplaceable asset in the early stage of the enterprise, but the larger the enterprise scale, the more complex the business, and the more globalized the market, the lower the marginal return of personal judgment, and the higher the marginal value of system and team decisions. Geely chose to complete this transition at this point, essentially a confirmation of its own development stage: it is already too big to operate relying on just one person's judgment.

Li Shufu's "resignation" is not an exit, but a role transition. He withdrew from the position of "enterprise operator" and instead became a "system designer" and "value guardian". The significance of this transition for Geely is indirectly but clearly projected in the financial data of the Q3 report.

A company operated personally by the founder often relies on personal judgment and ad-hoc decisions for cost optimization; efficiency is high but not reproducible; a company driven by systems and teams has cost optimization as an endogenous result of processes and mechanisms; it is slower but sustainable. Geely's Q3 expense ratio declining across the board, R&D reuse rate increasing, and four-brand collaboration improvement are not results that one person's decision can bring at once, but products of a mechanism operating continuously.

The relationship between this matter and the cost structure is not direct, but it is fundamental enough. Rewriting the cost structure requires long-term investment, and long-term investment fears uncertainty at the governance level most. When a company's strategic direction no longer sways with personnel changes, mechanisms like R&D reuse, platform modularization, and asset-light overseas expansion that take three to five years to take effect are only then possible to be executed stably.

Looking at Geely's Q3 report value from this angle again, it does not lie in proving that Geely can still grow, but in proving that Geely's cost structure no longer relies on revenue growth rate.

Functions can be chased, but the mechanism of cost structure cannot. This is the earliest and hardest-to-copy layer of evidence that "Global Leader in AI Intelligent Vehicles" has in finance.