Forget about XPENG Motors

On April 15th, in Shenzhen.

He Xiaopeng stood on the press - conference stage, with a full - size flagship SUV, the XPeng GX, behind him.

This scene is not unfamiliar.

Four years ago, at another press conference, there was also a flagship SUV, and He Xiaopeng personally took the stage.

That car was called the G9, which was touted as the best SUV under 500,000 yuan. Just 48 hours after its launch, there were urgent changes to the SKUs, pricing, and configuration logic. That night, He Xiaopeng was questioned face - to - face within the company. He later said, hyperbolically, that it was like a warrior losing his head.

Four years later, the GX is priced at 399,800 yuan, with the same price for both the pure - electric and range - extended versions.

If you look at the GX in the same way as the G9, you might think it's just another attempt by XPeng to move upmarket. But taking a step back, you'll see something else: a company has finally learned to let go of its past self.

1. The Cost of a Believer

In the past decade, if you had to use one word to describe XPeng, the most appropriate one would be "believer".

It adhered to the pure - electric route, avoiding range - extended and hybrid technologies. It pursued full - stack self - developed intelligent driving and didn't use supplier solutions. It aimed to be the first to make the urban NGP work, the first to mass - produce 800V technology, and the first to send a flying car into the sky. He Xiaopeng repeatedly said: Do what is difficult but right.

This stance was an advantage during the industry's upward phase. In 2021, XPeng delivered nearly 100,000 vehicles annually, and its stock price reached an all - time high. The capital market was willing to buy into the story of pure - electric and intelligent driving, and He Xiaopeng seemed like a prophet.

The turning point came quickly.

In September 2022, the XPeng G9 was launched. There were more than a dozen SKU versions, and the base model split all the core selling points into optional packages. Consumers felt like they were doing a college entrance exam math problem when looking at the configuration table. Social media exploded within 24 hours. He Xiaopeng held an emergency meeting overnight. 48 hours later, the G9 had a "second launch", cutting versions, changing configurations, and effectively reducing the price by 20,000 - 30,000 yuan.

A flagship car full of promise failed to prove itself but instead showed that there were problems with internal management. That month, XPeng's Hong Kong - listed shares dropped by about 73% in a single month, and its market value shrank by more than HK$250 billion from the year's peak, falling below the HK$100 - billion mark.

The G9 incident was not a slip - up. The old XPeng's problems had always been there: Too much faith in technology and too little sense for products. The R & D department wanted to make the best cars, but the sales department didn't know how to sell them to the most people.

Then came 2023. XPeng fell out of the top tier of new - energy vehicle startups, and its monthly delivery volume once dropped to 5,000 - 6,000 units. He Xiaopeng initiated the largest organizational restructuring in the company's history, bringing in Wang Fengying, the former vice - chairperson and general manager of Great Wall Motors, to serve directly as the president, making changes step by step.

The firmer the belief, the more painful the turn.

2. Many Stories at XPeng

From the end of 2025 to April 2026, in just a few months, the changes at XPeng accelerated. They were not isolated but more like a series of dominoes.

What toppled the first domino was a decision that XPeng would never have made in the past.

On January 8th this year, at XPeng's new - year press conference, four new models were launched at once. The P7+ and G7 were unveiled in both pure - electric and super - range - extended versions simultaneously, and the official put forward a new slogan: One car, two capabilities.

Only those who understand XPeng's history can feel the weight of this term. The company was a firm guardian of the pure - electric route in the past, and He Xiaopeng had more than once publicly shown his disdain for range - extended and hybrid technologies. Li Auto achieved a monthly sales volume of over 50,000 units with range - extended vehicles, but XPeng never followed suit.

Now it has. The person who once said that pure - electric was the future has included range - extended technology in the product line. It's not a tentative move but a major route on par with pure - electric.

This is equivalent to admitting one thing: Pure - electric alone is not enough to support the next stage of scale growth and global expansion.

Once this is acknowledged, the subsequent cards are turned over one by one.

The first knot untied by range - extended technology is overseas expansion. The density of charging piles overseas is far lower than in China. Range - extended vehicles are not a compromise there but a necessity. With the range - extended version as a foundation, XPeng's globalization can truly take off.

By the end of 2025, XPeng had entered 60 countries and regions, with 380 overseas stores. The P7+ was launched simultaneously in 36 countries. In 2025, the overseas delivery volume nearly doubled to 45,000 units, and the overseas revenue accounted for more than 15%. In the first quarter of 2026, the expansion continued into the UK, New Zealand, Mexico, and Latin America. The official no longer talks about "going overseas" but "taking root globally".

As overseas expansion unfolds, the product launch rhythm must keep up. In the past, XPeng took a year to develop a single model, often with long product gaps that dragged down sales. By 2026, this problem had basically disappeared. Four models were launched in January, and in March, the pure - electric version of the X9, the range - extended version of the G6, and a revised version of the P7 were successively introduced. At the beginning of April, the 2026 version of the MONA M03 was launched. The intensive new product launches essentially represent a shift from a single - thread to a parallel mode in the supply chain and R & D pipeline.

After the product line expands, XPeng's technology base begins to have independent value.

In the past, XPeng's technology labels were XNGP and urban intelligent driving, which were essentially just functional selling points in the configuration table. Now it's different. The second - generation VLA, the world model X - World, the self - developed Turing AI chip, and the fact that Robotaxis and mass - produced vehicles share the same model base have elevated intelligent driving to the company's AI infrastructure.

The GX is equipped with four self - developed Turing chips, with a local computing power of 3000 TOPS. He Xiaopeng defined it as China's first pre - installed and mass - produced Robotaxi prototype. Since the Turing chips were mass - produced and installed in vehicles in the third quarter of 2025, the cumulative shipment has exceeded 200,000 units, and the annual shipment target for 2026 is nearly 1 million units.

Once the technology base is independently formed, the revenue logic naturally changes. On March 13th this year, the first jointly developed model by Volkswagen and XPeng, the Volkswagen 08, officially went into production, standard - equipped with XPeng's VLA intelligent driving system and the Turing AI chip. XPeng has started to sell technology, exporting chips, intelligent driving solutions, and electronic and electrical architectures as products.

It was hard to imagine three years ago that a Chinese new - energy vehicle startup would sell its self - developed intelligent driving solution to one of the world's largest traditional automakers. In the 2025 annual report, XPeng's service and other revenues reached 8.34 billion yuan, accounting for nearly 11% of the total revenue, a year - on - year increase of 65.6%, and the profit margin of related businesses was as high as 68.2%.

All these changes combined make it inevitable for the company's identity to change.

Since April 1st, XPeng Motors has officially changed its name to XPeng Group. Autonomous driving, Robotaxis, humanoid robots IRON, flying cars, and physical AI are all included in the same strategic basket. He Xiaopeng said during the Two Sessions: 2026 will be the first year for AI to enter the physical world.

The name "XPeng Group" turns this statement into the company's name.

Looking at these changes together, they seem more like a causal chain:

Admitting the value of range - extended technology clears the way for overseas expansion; overseas expansion speeds up the product rhythm; a faster rhythm allows the technology base to emerge; the ability to sell technology changes the revenue structure; and a changed revenue structure makes the name "XPeng Motors" no longer sufficient.

The old XPeng's creeds of pure - electric, intelligent driving, and only selling cars are being rewritten step by step.

3. A Bet That Can't Be Lost

If we consider XPeng's series of changes, today's XPeng seems to believe that the next round of differentiation in the automotive industry will not be determined by the scale of vehicle sales but by the depth of AI capabilities.

Is there a chance of winning this bet? Yes, and it's not small.

Currently, globally, XPeng is the only Chinese company that can sell its intelligent driving system to mainstream international automakers. Volkswagen chose XPeng not because it had no other options but because XPeng's technology has reached a certain threshold. This endorsement is more valuable than any sales figures.

In 2025, XPeng's R & D expenditure was 9.49 billion yuan, a year - on - year increase of 47%. The annual net loss was 1.14 billion yuan. In Q4 2025, it achieved a single - quarter profit for the first time, with a net profit of 380 million yuan. Despite this, it maintained a high R & D intensity. He Xiaopeng didn't slack off as sales recovered; instead, he accelerated his bets.

But this bet also has its dangerous side.

Technology output is a long - term process. The production of the Volkswagen 08 is just the beginning, and there is still a long way to go from a single cooperation project to stable technology revenue. The market won't give XPeng an infinite time window. Each of the multiple fronts in physical AI, such as Robotaxis, robots, and flying cars, is an independent and tough battle that will continuously consume funds and talent.

By the end of 2025, XPeng had 47.66 billion yuan in cash on hand, which seems sufficient. However, burning multiple fronts simultaneously will consume resources much faster than expected. The numbers of 60 countries and 380 overseas stores look good, but the difference between "taking root globally" and "going overseas" lies not in the number of stores but in in - depth construction of local operations, after - sales systems, and brand recognition. Throwing money into these areas doesn't necessarily yield immediate results.

Priced at 399,800 yuan, the GX, the most expensive car XPeng has ever made, can be seen as the new XPeng's first report card.

XPeng has tried to move upmarket three times. The first was with the G3 in the 150,000 - 200,000 - yuan price range, and it managed to gain a foothold. The second was with the G9, aiming for the 300,000 - 350,000 - yuan range, but it failed. The third is now, with the GX in the 400,000 - yuan class.

He Xiaopeng's exact words were that as XPeng moves upmarket, the GX leads the way, with a sales target of being in the top 3 in the full - size SUV market.

However, XPeng's competitors are not to be underestimated.

The AITO M9 is priced from 469,800 yuan and has maintained its position at the top of the high - end SUV market for several consecutive months. The Li L9 (new Livis version) is priced at 559,800 yuan, with a solid brand recognition. The NIO ES9 has a pre - sale price starting from 528,000 yuan and is also very competitive. There is also a potential challenger in the distance, the Xiaomi YU9, which will likely be unveiled in the second half of the year.

The bet on the GX is even bigger than that on the G9. The G9 was just a bet on whether the car could sell well. The GX is a bet on whether the market is willing to pay for XPeng's new identity.

The price of 399,800 yuan is not just a number. To justify this price, XPeng has equipped the GX with all it can, including 3000 TOPS of computing power, L4 - level safety redundancy, steer - by - wire technology, and four self - developed chips. Its underlying architecture is shared with Robotaxis and flying cars.

But just because there is a lot of technological configuration doesn't mean people will believe in it.

XPeng has long been recognized for its high cost - effectiveness and strong intelligent driving capabilities, with its core price range being 170,000 - 280,000 yuan. 400,000 yuan is the dividing line between mainstream and high - end brands. The G9 failed to cross this threshold, and now a more expensive car is trying again.

When asked by a reporter if making cars was hard, He Xiaopeng didn't beat around the bush and frankly said that making cars had been an extremely painful experience.

If the GX succeeds, XPeng will no longer be the weakest among NIO, XPeng, and Li Auto. Instead, it will become a platform - type company that can cover the 100,000 - 400,000 - yuan price range and export AI technology. If it fails, the lesson of the G9 will be repeated, but this time the stakes are even higher.

4. The "NIO, XPeng, Li Auto" Label Has Become Obsolete

Taking a step back, what XPeng is going through is not just about the company itself.

The "NIO, XPeng, Li Auto" label, which has been used for five years, should now be discarded. NIO is making mobile phones and building a battery - swapping network. Li Auto is working on AI and self - developed chips. XPeng is involved in robotics, Robotaxis, and flying cars. The boundaries of the three companies are rapidly blurring.

Defining them by monthly delivery volume and gross profit margin is no longer sufficient.

We need to ask one more question: Who does this company really want to be?

XPeng's answer is clear. It no longer wants to be just a smart electric vehicle company but a company that uses AI to change the physical world. He Xiaopeng's exact words were that the greatest value of AI lies in changing the physical world, not just the digital world.

However, the market has not fully bought into this story yet.

By the end of 2025, XPeng's market value in the Hong Kong stock market was about HK$151.7 billion. By April 15th, 2026, it had dropped to about HK$128 billion, a decline of about 15.6%. In the first quarter of 2026, it delivered 62,682 vehicles, a year - on - year decline of 33.32%, ranking fourth among new - energy vehicle startups, behind Leapmotor (110,200 units), Li Auto (94,100 units), and NIO (83,500 units).

The business situation is indeed improving. In Q4 2025, it achieved a single - quarter profit of 380 million yuan for the first time. The annual revenue was 76.72 billion yuan, a year - on - year increase of 87.7%. The net loss narrowed from 5.79 billion yuan (not 5.8 billion yuan) in 2024 to 1.14 billion yuan. The R & D expenditure was 9.49 billion yuan, a year - on - year increase of 47%. By the end of 2025, it had 47.66 billion yuan in cash on hand, so it is not short of resources.

The capital market's attitude is simple. It acknowledges that you are