The Car as a Smartphone: He Xiaopeng's Bet on In-House Software

Prologue: The Outsider's Perspective

When He Xiaopeng, a software veteran, began to immerse himself in the automotive world in 2017, he was struck by how little the industry had changed. For a hundred years, the car had been defined by its mechanical parts: the engine, the transmission, the chassis. Software was often an afterthought, a clunky infotainment system outsourced to a third-party supplier.

To He Xiaopeng, this was completely backward. He had just spent a decade building a company based on the principle of rapid, over-the-air software updates and a seamless user experience. He looked at the modern car and saw a product that was essentially "frozen in time" the moment it left the factory.

He posed a fundamental question to his new team at XPeng: "What is a car?"

His answer was different from that of his rivals. A car, he argued, was not a static piece of hardware. It was a smart, connected device, more akin to a smartphone than a traditional vehicle. Its primary value would no longer come from its horsepower, but from the intelligence of its software. And if software was the soul of the car, he was absolutely convinced that XPeng had to build that soul itself.

Act I: The 'Full-Stack' Gamble

This philosophy led to one of the most important strategic decisions in XPeng's history. At the time, most car companies, including many EV startups, were taking a shortcut on autonomous driving. They licensed the core technology from established suppliers like Mobileye or Bosch. It was faster, cheaper, and less risky.

He Xiaopeng rejected this path. He believed that relying on a supplier for the car's most critical software was a recipe for long-term failure. "If the core of the smart car is its brain, how can you outsource your brain?" he would ask. He argued that a third-party system would never be as deeply integrated or as rapidly upgradeable as one built in-house.

He made the hugely expensive and difficult decision to build a "full-stack" autonomous driving system from the ground up. This meant hiring hundreds of AI engineers, building a massive data collection infrastructure, and developing everything from the perception algorithms to the decision-making software in-house. It was a massive gamble that would consume a huge portion of the company's precious capital and add years to the development timeline.

Act II: The Birth of XPILOT

The internal development program was codenamed XPILOT. It was a grueling, complex endeavor. The team had to solve immense technical challenges, from training their neural networks on complex urban driving scenarios to designing the user interface that would build trust between the driver and the machine.

He Xiaopeng was deeply involved, treating the XPILOT team like the core of the company. He brought his product manager's sensibility to the process. He wasn't just focused on the technical capabilities; he was obsessed with the user experience. How did the lane-keeping feel? Was the automatic lane change smooth and confident? Was the system easy and intuitive to activate?

The first versions were rough. But the in-house approach allowed for a cycle of rapid iteration that was impossible for competitors relying on suppliers. XPeng's fleet of test vehicles was constantly collecting data, and that data was fed back to the engineering team in Guangzhou and Silicon Valley, who could then push a new software update over the air in a matter of weeks. The system was learning and improving at a pace the traditional auto industry had never seen.

Epilogue: The Soul of the Machine

The payoff for this massive bet came with the launch of the XPeng P7 sedan in 2020. The P7 was a sleek, high-performance EV, but its star feature was its advanced version of XPILOT, which included a revolutionary "Navigation Guided Pilot" (NGP) feature for highways. It was one of the most advanced driver-assist systems on the market, rivaling and in some cases surpassing Tesla's Autopilot in the Chinese market.

Crucially, it was XPeng's own technology. This gave them a powerful marketing story and a deep competitive moat. While rivals were stuck on the product cycle of their suppliers, XPeng could add new features and improve its system with every software update, just like a smartphone.

He Xiaopeng's bet had been vindicated. He had correctly identified that in the new era of mobility, software was not a feature; it was the entire product. By building the car's "brain" himself, he had given his company a soul, and a fighting chance to win the future of driving.