He Xiaopeng: The goal for XPeng Robotics is to ship more than 10,000 units in 2027.

```

The AI labels of car companies are being strengthened.

On April 24, the 2026 Beijing International Auto Show officially opened. On the first day, XPeng not only released the first second-generation VLA intelligent driving report and launched the flagship model XPeng GX for the era of physical AI, but also showcased the new generation IRON humanoid robot and land carrier.

XPeng Group Chairman and CEO He Xiaopeng judged in an interview that if the past decade was about the shift from fuel vehicles to new energy vehicles, then the next decade will bring a fundamental transformation from new energy vehicles to AI vehicles.

At the same time, XPeng is trying to build a complete technological evolution and monetization path from smart cars to a general physical AI ecosystem.

Intelligent Driving Adoption Accelerates

The commercial penetration rate of intelligent driving is a core indicator of a car company's ability to monetize software. Order data disclosed at this press conference shows the market is reaching an important inflection point in accepting advanced intelligent driving.

He Xiaopeng revealed that in March this year, XPeng's Ultra series orders increased by 118% month-on-month. The more critical cross-verification is: the sales volume of first-time car buyers choosing Ultra and Ultra SE (advanced intelligent driving version) models increased by a whopping 129.3% month-on-month.

In the vehicle terminal market, this level of month-on-month growth indicates that advanced intelligent driving is no longer limited to early enthusiast users, but is substantively entering the core decision chain of mass consumers buying cars.

On the technology implementation side, He Xiaopeng disclosed that the second-generation VLA system will soon launch features like campus and underground parking roaming. From the perspective of industry technology evolution, the intelligent driving competition of the past two years was mostly focused on highways and urban main roads (structured roads), while campuses and underground parking lots represent complex lighting and GPS-less unstructured blind spots.

Addressing this scenario means the intelligent driving system is breaking through the “last hundred meters” from parking space to parking space, continuously expanding the boundaries of full-scenario intelligent driving. He Xiaopeng also stated in interviews that in the next few years, the software capabilities of smart vehicles will march infinitely toward L4 or even L5 level autonomous driving.

From a product perspective, the first new model unveiled at this auto show, the XPeng GX, is officially defined as a product for the era of physical AI, with its core commercial logic being the lowering of Robotaxi intelligent driving capabilities.

Previously, vehicles with Robotaxi-level perception and computing power redundancy mainly served B-end commercial trial operations and were constrained by high hardware costs to limited adoption. The launch of XPeng GX aims, through mass production for the consumer market, to dilute the marginal cost of advanced perception hardware and large model research and development.

Physical AI Accelerates Implementation

XPeng's viewing of automobiles as one carrier of AI is another core message conveyed at this auto show. The concentrated display of the new IRON humanoid robot and land carrier is a direct implementation of its "physical AI" strategy outside the automotive field.

Regarding the currently hot but early-commercialization embodied intelligence track, He Xiaopeng gave a relatively objective and calm industry assessment in interviews: "I think robots today have not really helped people yet."

But he also pointed out, "After next year, there will be changes." This judgment is based on the speed at which AI models' generalization ability improves; once the threshold of complex non-standard tasks can be crossed, the practical value of robots will undergo a qualitative change.

Based on this technical forecast, XPeng has set aggressive production targets.

He Xiaopeng revealed to media such as Wallstreetcn that in 2025, the total shipments of China's robot industry are expected to be just over ten thousand units. XPeng's robot aims to "far exceed" this number in 2027.

Its confidence comes from the high reuse of underlying technologies between cars and robots. Currently, XPeng’s accumulated capabilities in smart car vision perception, large model algorithms, and even electric-drive and battery supply chains, can all be massively transferred to the humanoid robot field.

He Xiaopeng predicts that once robots achieve mass production, their growth speed and scale will surpass the historical trajectory of new energy vehicles in China.

Summing up the information from this press conference, XPeng's ten-year transition blueprint is clear, but from the perspective of objective industry development, the leap from new energy to physical AI must still face practical tests.

In terms of laws and liability systems, although technology is rapidly advancing towards L4 level, relevant road laws, accident liability determination and the establishment of commercial insurance systems are still in pilot and exploratory stages. The gap between technological breakthroughs and regulatory relaxation will become a variable the whole industry faces.

Validating commercial closed loops in non-automotive scenarios is also a significant challenge. Although the underlying supply chain can be reused, for robots to achieve explosive shipment growth in 2027, they must prove, in real scenarios such as industrial manufacturing, warehousing, logistics, or home services, they possess cost-reduction and efficiency improvement abilities superior to traditional automation equipment—in other words, they must find true demand from buyers.

In this transformation of the auto industry from power-driven to computing-driven, XPeng has taken the lead in releasing its technology roadmap. The doubling of advanced intelligent driving orders is direct proof of its current strategy's effectiveness, while whether it can deliver on its robot mass production goals in the next two years will become the key test of its physical AI credentials.

Risk Warning and DisclaimerThe market has risks, investment requires caution. This article does not constitute individual investment advice and does not take into account the specific investment objectives, financial situation or needs of individual users. Users should consider whether any opinions, viewpoints or conclusions in this article are suitable for their particular situation. Investment based on this is at your own risk. ```