When the earnings season rolls around, the most eye-catching figure isn't how many cars were sold by whom, but how much money was burned in the lab.

BYD: Earns 3, spends 2—truly a tough guy

BYD's annual research and development amounted to 63.4 billion yuan, ranking first in the A-share market, with 120000 engineers and over 70000 patents. Net profit of 32.6 billion yuan, for every 3 yuan earned from outside, 2 yuan goes into the laboratory and the remaining 1 yuan is pocketed profit. Money has been invested in vehicle intelligence: "Eye of the Heavenly God" assisted driving, fifth generation DM, second-generation blade battery, megawatt flash charging, both soft and hard, self-developed full stack.

Geely: Revenue of 345.2 billion yuan, the most decisive AI transformation

Geely's total revenue for 2025 is 345.2 billion yuan, a year-on-year increase of 25%; After deducting non core attributable net profit such as foreign exchange, the core attributable net profit was 14.41 billion yuan, an increase of 36%. The total sales volume was 3.025 million vehicles, and the penetration rate of new energy exceeded half for the first time. R&D investment was 17.62 billion yuan, a year-on-year increase of 29%, and cash reserves reached a historical high of 68.2 billion yuan.

But Geely's biggest blow is not numbers, it's strategy: completely stopping the development of non AI intelligent cockpits and completely shifting towards AI native architecture. I don't want the previous gameplay anymore, I'll redo it from scratch. Supporting this determination is the Star Wisdom Computing Center 2.0, with a computing power of 23.5 EFLOPS, which firmly ranks first among Chinese car companies. The route taken by Geely is "computing power base+technology decentralization" - first smashing out the infrastructure, and then delegating advanced intelligent driving to Lynk&Co and Galaxy.

Great Wall: Revenue hits new high, profits plummet in half

The revenue of Great Wall in 2025 is 222.8 billion yuan, a year-on-year increase of 10.2%, surpassing 220 billion yuan for the first time. But the net profit attributable to the parent company was only 9.865 billion yuan, a year-on-year decrease of 22.07%, and the net profit after deducting non recurring expenses fell sharply by 37.5%. A typical example of 'increasing income without increasing profits'.

Where did the money go? Sales expenses amounted to 11.273 billion yuan, a year-on-year increase of 43.93%; Research and development expenses amounted to 10.432 billion yuan, an increase of 12.13%. Despite declining profits, research and development continues to grow in double digits - this is the posture of Great Wall's "die hard" technology. The research and development focus is on the VLA assisted driving large model, and its biggest highlight is to demonstrate the decision-making process through the CoT thinking chain, solving the "black box trust crisis" of intelligent driving. Previously released data showed that the intelligent R&D team has over 5000 people, with software development accounting for 70%.

Wei Xiaoli: New forces are not soft hearted either

By 2025, Ideal Auto will invest approximately 11.3 billion yuan in research and development, with half going to AI. Over the past three years, the cumulative investment has reached 33 billion yuan, with an average of one billion yuan burned every three days. CEO Li Xiang said that the company wants to become a "embodied intelligent enterprise" and will mass produce self-developed chips by 2026. Xiaopeng has nearly 9.5 billion yuan in research and development throughout the year, with nearly half focused on AI, selling software subscriptions while charging technology licensing fees to the public.

Interestingly, NIO is one of the few car companies this year that has seen a decrease in research and development investment, and has been cutting hard to reduce costs and increase efficiency. NIO's total revenue for 2025 is 87.488 billion yuan, with an annual R&D expenditure of 10.605 billion yuan, a year-on-year decrease of 18.7%. But AI has not let go either. In October 2025, the World Model 2.0 version was launched, which focuses on "open interaction". At the same time, the intelligent driving department has been adjusted to a "4 × 100 relay baton" mode, merging the intelligent driving organization and the general AI organization.

No one can see through the outcome of this bold gamble

By 2025, the industry's profit margin will only be 4.1%, with less than 7 out of 11 mainstream car companies experiencing profit growth. Geely exchanged 17.62 billion yuan in research and development for a profit growth rate that outperformed its revenue; Great Wall achieved a new high in revenue with 10.421 billion yuan in research and development and 11.273 billion yuan in sales, but faced pressure on profits; BYD exchanged 63.4 billion yuan in research and development for technological leadership. If electrification is the "heart transplant surgery" for cars, then intelligence is the "brain transplant surgery". The 2025 financial report is announcing a fact: the top car companies of the future will be AI companies first, followed by automotive companies. Whoever completes this identity transformation first may win the next decade.