Why XPeng may be more than just China’s answer to Tesla

For the past few years, XPeng has largely been viewed as one of several emerging Chinese BEV brands aiming to challenge Tesla’s dominance in the global electric vehicle space.

That comparison is understandable given the company’s strong emphasis on software, intelligent driving systems, and AI-driven mobility technologies.

However, after a recent visit to XPeng’s headquarters in Guangzhou, China, it became increasingly clear that the company’s ambitions may extend far beyond merely becoming another Tesla rival.

While Tesla remains the benchmark in many areas of electrification, software integration, and charging infrastructure, XPeng appears to be positioning itself differently.

Rather than focusing purely on BEVs, the company is aggressively investing in a much broader ecosystem involving autonomous driving, artificial intelligence, in-house semiconductor development, robotics, and even aviation.

And perhaps more importantly, many of these technologies no longer feel like distant concepts.

The car almost became secondary

One of the biggest surprises during the Guangzhou visit was how quickly attention shifted away from the car itself, but of course, we had to cover that first.

Members of the Malaysian media were allowed to experience XPeng’s latest intelligent driving technology through the new XPeng P7 and G7 models, which may or may not enter the Malaysian market anytime soon.

Instead, local expansion is rumoured to focus on other future products, such as the upcoming GX flagship SUV, while rumours of a possible REEV variant have also begun to surface as XPeng continues to expand its regional plans.

Yet despite the sleek design and impressive performance of the P7 (we managed to briefly test this on the open road), the real highlight was XPeng’s latest VLA-based intelligent driving system paired with its Navigation Guided Pilot (NGP) technology.

To put it simply, once a destination was entered into the navigation system and NGP was activated, the vehicle could handle nearly the entire driving process on its own.

Acceleration, braking, steering inputs, lane changes, junction navigation, signalling, and traffic flow management were all executed with minimal driver intervention.

Under current Malaysian regulations, systems such as these are still classified under Level 2 autonomy, meaning drivers by law must remain attentive at all times with hands on the steering wheel and be prepared to take over immediately if required.

Malaysia has yet to legalise Level 3 autonomous driving, although the Ministry of Investment, Trade and Industry (MITI) has previously indicated ambitions toward enabling such technologies by 2030.

To put things into perspective, Level 2 systems still require full driver responsibility despite offering steering and speed assistance. Level 3 allows the vehicle to take over certain driving functions under specific conditions.

Level 4 represents a far more advanced stage where the system is capable of handling most driving situations independently within defined operational environments. And based on what was demonstrated in Guangzhou, XPeng’s latest VLA system is already within that upper tier of capability.

Central to that experience is XPeng’s latest Vision Language Action (VLA) architecture, which essentially combines visual perception, AI reasoning, and real-time decision-making into one unified system.

Unlike traditional assisted driving technologies that rely heavily on predefined rules or high-definition mapping, VLA functions more like an AI-driven interpretation model.

Using cameras and sensor inputs, the system continuously observes its surroundings, analyses traffic behaviour, predicts potential movements from nearby vehicles, and determines how the car should respond almost instantaneously.

In simpler terms, the vehicle is not merely following programmed instructions. Instead, it is actively interpreting the driving environment in a way that attempts to mimic human judgment and reaction patterns.

This is also where XPeng’s approach begins to differ from many conventional Level 2 systems currently available on the market. Rather than operating purely as a lane-centring or adaptive cruise control assistant, VLA appears designed to behave more like an AI co-driver capable of handling increasingly complex real-world traffic situations.

Impressive, slightly unsettling, but undeniably convincing

What made the experience particularly fascinating was not simply how capable the system appeared, but how naturally it behaved in real-world traffic scenarios.

Many assisted driving systems can feel abrupt or overly cautious, but XPeng’s VLA-powered setup handled urban roads with a level of confidence that often felt, dare we say, surprisingly 'human-like'.

The vehicle merged into traffic smoothly, reacted quickly to surrounding vehicles, navigated junctions confidently, and maintained consistent composure even in more complex traffic situations.

At times, it genuinely felt as though the driver had transitioned from actively operating the car to merely supervising it.

Yet despite the impressive demonstration, the experience also highlighted the psychological adjustment that still comes with advanced autonomous technology.

There were moments where the system’s confidence while navigating narrower roads or tighter traffic gaps felt slightly less comfortable than what some human drivers may personally prefer.

In certain situations, human intervention felt necessary, not because the technology appeared incapable, but simply to create a greater sense of comfort and spatial reassurance behind the wheel.

Ironically, that may say more about how rapidly the technology is evolving than its shortcomings.

One of the more fascinating aspects of XPeng’s latest VLA architecture is its reduced reliance on heavily pre-mapped environments, instead operating primarily through vision-based AI interpretation and real-time environmental analysis.

In theory, this could allow the system to adapt more naturally across different markets and road conditions globally without depending entirely on localised mapping infrastructure.

Even with tech this advanced, and its technical capability appears increasingly mature, the human side of trust and adaptation may require slightly more time to fully catch up.

XPeng’s bigger ambition may not actually be cars

Perhaps the most interesting takeaway from the Guangzhou visit was the realisation that XPeng increasingly resembles a tech company that happens to build vehicles, rather than the other way around.

This became even more apparent after witnessing the company’s wider AI ecosystem initiatives beyond automotive production.

Alongside its intelligent driving demonstrations, XPeng also showcased its ongoing aviation efforts via its Aridge brand through its so-called Land Aircraft Carrier, a modular flying vehicle system consisting of a large ‘Mothership’ six-wheel vehicle paired with a detachable aircraft unit.

XPeng showcased how the aircraft could be deployed and stored back into the carrier vehicle within minutes, while company representatives also confirmed that the project is expected to enter production and sales before the end of 2026.

According to XPeng, the Land Aircraft Carrier is expected to carry an estimated price tag of roughly 2 million Yuan, with over 100 bookings reportedly secured during the recent Beijing Auto Show.

Whether flying vehicles eventually become mainstream remains difficult to predict, particularly given regulatory and infrastructure challenges globally.

However, the fact that XPeng is already moving toward commercialisation rather than mere conceptualisation with its Aridge brand says a great deal about the scale of its long-term ambitions.

The same philosophy can also be seen in XPeng’s continued investment in its in-house Turing AI chip development.

The company’s in-house Turing AI chip is expected to play a major role in supporting the immense computational demands required for systems such as VLA, autonomous mobility, robotics, and future aviation projects.

The company’s Iron humanoid robot project further reinforces this direction. In many ways, the vehicle itself increasingly feels like just one component within a much larger AI-driven mobility ecosystem.

Why Malaysians should pay attention

For Malaysian consumers, XPeng’s ambitions may become increasingly relevant over the next few years. Beyond the already available G6 and X9, the brand is expected to continue expanding its local presence with additional models potentially arriving before the end of 2026.

The GX flagship SUV has already been heavily linked to Malaysia, while growing regional interest in REEV technology could also open the door for future electrified variants tailored toward Southeast Asian market conditions.

Of course, XPeng still faces several major challenges before it can truly rival established global automotive and technology giants. Questions surrounding software localisation, long-term after-sales support, regulations, and public trust toward autonomous systems remain significant hurdles as the company continues its international expansion.

Tesla itself also remains enormously influential as the benchmark for software-focused automotive development globally. Yet after spending time inside XPeng’s headquarters in Guangzhou, one thing became increasingly difficult to ignore.

The company no longer appears focused solely on building competitive BEVs. Instead, XPeng seems to be preparing for a future where artificial intelligence, autonomous mobility, robotics, and aviation become increasingly interconnected parts of everyday transportation.

And if that vision eventually succeeds, XPeng may ultimately become far more than just China’s answer to Tesla.