Nicolai Tangen: Hi everybody, I'm Nicolai Tangen, the CEO of the Norwegian Sovereign Wealth Fund, and I'm really excited about this podcast. XPENG is one of China's fastest growing EV makers, but it doesn't call itself a car company. It calls itself a physical AI company, betting that the same technology that can drive a car, pilot a flying car, and power a humanoid robot. My guest today is Brian Gu, vice chairman and president of XPENG. Brian, warm welcome.

Brian Gu: Thank you, Nicola. Very happy to be on the podcast.

Nicolai Tangen: Brian, what is XPENG?

Brian Gu: Well, I would say XPENG first of all is a technology company, but as we now dive into the depths of all things physical AI, I think we consider ourselves really kind of forefront of this physical AI movement. We are building EVs as you mentioned. It's one of the leading brands in China as well as globally. But increasingly we have broadened our product scope to be including more areas and directions including robots, flying cars and other modalities as well, but using the same core.

Nicolai Tangen: A lot of people think that electrical vehicle is a car with a battery, but you say it's the intelligence that's the differentiator, not the electrification. What do you mean by that?

Brian Gu: Well, that actually is what we actually have done differently over the years. Even when we were a very small startup, let's say about nine years ago when I first joined XPENG, that time I think everyone is looking at starting an EV company. The big pitch was electrified energy transition for a greener, sustainable future. But our founder, who comes from the internet and software background, always thought that if you're just making a car with a battery, the traditional automotive giants can eventually do that once they put their mind into it. So the true differentiation has to be coming from intelligence—whether it's the car's ability to drive itself without human intervention, or the car's ability to interact with human in the cockpit, or the car's ability to evolve over time. So we, from day one, put almost majority of our R&D into developing our own proprietary autonomous driving algorithms, software. We are the first one to commercialize the ADAS using high-definition maps in China, and we're also the first one to roll out end-to-end models in China.

Nicolai Tangen: So what does this mean in practice?

Brian Gu: Well, first of all I think when you get into a car, you have to really understand whether the car can understand you. You know the smart cockpit, for example, using natural language processing, you can speak to the car just like talking to a person next to you. You don't have to look for buttons, you don't have to use specific command words, you can just say: "Hey, I feel a bit tired today, can you adjust the seat and play some relaxing music?" And the car will understand and do that. It can also act as an assistant to schedule your day, find information, write emails for you. On the driving side, today the car can basically take you from your home parking spot to your office parking spot door-to-door. You just sit there and supervise, and the car will navigate through traffic, make turns, handle traffic lights, change lanes, and park itself into the underground garage.

Nicolai Tangen: What is this going to look like in 10 years time, you think?

Brian Gu: Well, I think with advancement in technology, I would say the Level 4 capability will be just a matter of time. It's going to happen, and it's probably going to happen faster than people think. In 10 years time, most of the cars probably will not have steering wheels or pedals. The interior space will be completely redesigned—it becomes a mobile living room, a mobile office, or an entertainment space. You can get in, go to sleep, watch a movie, or have a meeting, and the car will take you to your destination safely.

Nicolai Tangen: I mean, I'm sorry, just to inject here, Level 4 capability, what does that actually mean?

Brian Gu: Level 4 means the vehicle can fully drive itself under most operating conditions without requiring human intervention. In Level 3, you still need a human driver as a fallback who must take over when requested. But in Level 4, the liability shifts to the system. The vehicle has full redundancy and robustness so that even in edge cases or if something fails, the car can safely navigate or come to a safe stop on its own.

Nicolai Tangen: So what is what's happening with range and charging times going forward?

Brian Gu: On range, today most mainstream EVs already provide 600 to 800 kilometers of real-world range, which is more than enough for daily commute and intercity travel. The real challenge is not just putting a bigger battery to get 1,000 kilometers because that adds weight and hurts efficiency. The real breakthrough is charging speed. With 800V high-voltage Silicon Carbide platforms and 5C ultra-fast charging batteries, today we can add 400 to 500 kilometers of range in just 10 to 12 minutes. That is getting very close to the time it takes to fill up a tank of gas at a gas station. Once that charging infrastructure is widely deployed, range anxiety will be virtually eliminated.

Nicolai Tangen: When it comes to the mechanical structure of the car, how has the complexity and manufacturing process changed with the transition to EVs?

Brian Gu: An electric vehicle has about 80% to 90% fewer moving parts compared to an internal combustion engine vehicle. Traditional engine and multi-speed transmission systems are extremely complex with thousands of precision components. In an EV, the core is the powertrain: battery, electric motor, and electronic control. More importantly, manufacturing has been revolutionized. For example, we introduced integrated front and rear integrated die-casting. What used to be hundreds of stamped sheet metal parts welded together can now be cast as a single piece in dozens of seconds with giant die-casting machines. That dramatically reduces weight, increases structural rigidity and crash safety, and significantly cuts down factory footprint and manufacturing costs.

Nicolai Tangen: Does that mean software is now overtaking traditional hardware in terms of the value proposition of a car?

Brian Gu: Absolutely. That's the essence of "software-defined vehicles." In the past, when you bought an ICE car, the moment you drove off the lot, it started depreciating and the features were frozen forever. Today's smart EV is a living, evolving device. Through full-vehicle OTA updates, we can push new firmware and features every month or couple of weeks—giving users upgraded cockpit UI, new applications, smarter voice assistants, and better driving assistance capabilities. The hardware acts as the computing foundation on which software and algorithms continuously iterate.

Nicolai Tangen: How much computing power is actually required inside a car?

Brian Gu: The compute requirement is enormous. A modern smart car is essentially a high-performance supercomputer on wheels. In the era of end-to-end AI models, we process high-throughput multimodal data streams from dozens of high-res cameras, radars, and sensors in real time, requiring onboard computing platforms capable of hundreds or thousands of TOPS. Furthermore, the cloud computing cluster for training is even bigger. We need tens of thousands of GPUs in our training clusters to continuously train foundation autonomous driving models and cockpit LLMs on tens of millions of kilometers of real and simulated driving data every single day.

Nicolai Tangen: Coming back to the earlier question, how far ahead of Europe and the US do you think Chinese producers are now?

Brian Gu: Well, I think we are leading in a few areas. In electrification, Chinese players obviously have the advantage of strong battery technology, supply chain capacity, and cost advantages because China is the largest automotive market in the world. But what's really driving the competitiveness in China is also the fast evolution of technology. Chinese consumers are young, tech-savvy, and very demanding. This massive home market and intense competition force companies to iterate both hardware and software at an unprecedented pace. In terms of smart cockpit, autonomous driving, and software engineering, Chinese EV makers are definitely at the forefront globally.

Nicolai Tangen: How long do you think it would take XPENG to roll out globally? Because you've started in the Nordic countries, how do you compare against global peers?

Brian Gu: We look at a number of dimensions when competing with global peers. In terms of hardware build quality, safety standards, and driving dynamics, our vehicles are engineered to meet and exceed the highest standards, including Euro NCAP five-star ratings. In terms of software interaction, charging speed, and smart ADAS capabilities, we offer a truly differentiated and next-generation user experience. However, global expansion takes time. Building brand awareness, localized distribution, service networks, and regulatory compliance requires steady execution. We are taking a focused approach starting from high-EV-adoption European markets like Norway, Denmark, Sweden, the Netherlands, Germany, and then expanding into the Middle East, Asia-Pacific, and Latin America. Superior products and technology will ultimately speak for themselves.

Nicolai Tangen: What explains the difference in speed in innovation and generally in speed? I mean what explains what we call the "Chinese speed"?

Brian Gu: The Chinese speed is driven by multiple factors. First is the unparalleled local supply chain ecosystem. In regions like the Yangtze River Delta or Greater Bay Area, hundreds of tier-1 and tier-2 suppliers are within hours of drive. If you need a custom part, a new mold, or prototype sensor, suppliers can turn around samples in days rather than months. Second is the enormous talent pool of high-caliber software and hardware engineers graduated every year with strong work ethic. Third is agile organizational culture. Unlike traditional car makers with siloed, linear waterfall processes, we adopt agile software development methods where multidisciplinary teams work concurrently on hardware, software, and validation.

Nicolai Tangen: How much faster would you say?

Brian Gu: In traditional automotive industry, developing a new vehicle platform typically takes 4 to 5 years. In China, we have shortened that development cycle to about 24 to 30 months, or even less. And on the software and algorithm layer, we iterate on a weekly or bi-weekly cadence via continuous integration and OTA. That level of speed was previously unheard of in the traditional auto world.

Nicolai Tangen: How much harder do you think you work?

Brian Gu: Well, I would say our team works extremely hard. It's not just about long working hours; it's about the intense focus, the sense of urgency, mission-driven dedication, and efficient teamwork. Everyone is racing against time to push technical boundaries, and that entrepreneurial spirit permeates every level of the organization.

Nicolai Tangen: Flying cars—when will cars fly?

Brian Gu: It's already flying! But we are approaching flying cars in a very practical and innovative way. We have an affiliated company called AEROHT. We developed what we call the "Land Aircraft Carrier"—a modular flying car. It consists of a ground mothership vehicle, which is a six-wheel-drive extended-range electric carrier, and an air module, which is an eVTOL (electric vertical take-off and landing) aircraft seamlessly docked inside the trunk. The mothership drives on ordinary roads, parks in standard spots, and charges the aircraft on the go. When you arrive at a scenic spot, with a push of a button, the aircraft automatically unfolds and deploys. It features automated single-key takeoff, landing, and flight path planning, allowing ordinary users without extensive pilot training to enjoy personal flight safely. We plan to deliver this product to customers within the next year or two.

Nicolai Tangen: But when do you think there'll be a mass market?

Brian Gu: For a true urban air mobility mass market, in my opinion, it will take a longer adoption cycle. Airspace regulation, flight corridor management, airworthiness certifications, and safety standards are heavily regulated and complex. That's why our go-to-market strategy starts from outdoor leisure, recreational flying, and search-and-rescue in designated airspace before moving into dense urban commuting. By accumulating millions of safe flight hours in recreational settings, we can pave the way for broader regulatory approval and infrastructure maturity.

Nicolai Tangen: Brian, let's move on to another incredible area—humanoid robots. Why are you going into robots?

Brian Gu: The reason we went into robots is actually very natural. As I mentioned earlier, XPENG's core DNA is physical AI. An autonomous smart EV is essentially a robot on wheels: it uses visual perception to understand 3D physical world, AI neural networks as the brain to make decisions, and actuators to execute motion. Humanoid robots (like our latest 'Iron' humanoid robot) share the exact same underlying technology stack: end-to-end foundation models, automotive-grade chips, sensors, power electronics, and motor control algorithms. Applying this embodied AI capability to humanoid robots allows them to work in our manufacturing plants, retail showrooms, and eventually provide household assistance.

Nicolai Tangen: What does the competitive landscape look like? There seem to be a lot of very serious players here.

Brian Gu: In the last two to three years, humanoid robotics has seen a huge boom. Globally, you have Tesla with Optimus and top AI robotics startups in the US. In China, there is also a vibrant ecosystem of robotic companies. The unique advantage of an EV maker like us compared to pure software startups is our deep capability in automotive-grade hardware manufacturing, massive supply chain scale, and access to internal factory environments where robots can perform actual assembly tasks, collecting high-value real-world physical training data.

Nicolai Tangen: The fact that you have so many players in the EV space, what does that do to the Chinese abilities in the robotic space?

Brian Gu: It has an enormous impact. The high-density batteries, lightweight structural alloys, precision motors, harmonic and planetary gear reducers, sensors, and electronic control units required by humanoid robots directly leverage the existing supply chains of electric vehicles and precision consumer electronics. The manufacturing scalability and cost-down capabilities developed by Chinese industry will drastically accelerate the commercial viability of humanoid robots.

Nicolai Tangen: Have you got access to all the chips you want? Nvidia chips and all these kind of things? When do you think they'll be on par?

Brian Gu: On the vehicle side, we have been designing our own custom high-throughput computing chips optimized for physical AI and end-to-end models, offering superior energy efficiency for our neural network workloads. On the cloud training side, despite geopolitical restrictions, we leverage algorithmic optimizations, heterogeneous compute orchestration, and diverse computing infrastructure to support our training pipelines. While US players lead in foundational general LLMs, in physical AI, autonomous driving end-to-end models, and multimodal interaction with real-world physical data, Chinese innovators are moving incredibly fast and are globally competitive.

Nicolai Tangen: We talked about self-driving cars, flying cars, humanoid robots. What is a city going to look like in 20 years time?

Brian Gu: In my view, future cities will become much more decentralized and dispersed. People will no longer need to crowd into expensive, dense downtown city centers just to be close to work. When autonomous vehicles become comfortable mobile living spaces and three-dimensional low-altitude transport reduces intercity commute to minutes, people can live in suburban or satellite communities close to nature with higher quality of life. Surface parking lots and traffic congestion will be converted into green parks and pedestrian spaces, while repetitive and hazardous physical chores will be handled by robots.

Nicolai Tangen: Do you think we're going to be happier?

Brian Gu: I certainly hope so. The fundamental purpose of technology is human liberation. Today, a significant portion of our waking hours is consumed by tedious routines—sitting in traffic jams, doing household chores, and performing repetitive labor. When physical AI and robotics shoulder these burdens, humans will be freed up to pursue creative endeavors, arts, meaningful human connections, and personal growth.

Nicolai Tangen: Moving on to the personal side, what matters in your life?

Brian Gu: I joined XPENG nine years ago, joining my friend and partner He Xiaopeng, our Chairman and CEO. Before that, my background was in biochemistry where I earned my Ph.D. from the University of Oregon, followed by an MBA from Yale, and then spent many years as an investment banker leading J.P. Morgan's Asia-Pacific investment banking franchise. Being at the epicenter of an industry transformation that reshapes mobility and human lifestyle, building the future with a group of brilliant innovators—that continuous journey of learning and moving boundaries is what drives me every day.

Nicolai Tangen: How does your day look like? When do you wake up?

Brian Gu: I wake up around 7:30 AM and start my first meeting around 8:30 or 9:00. Every single day is different and dynamic: ranging from strategic discussions on global markets, supply chain coordination, meeting with global investors and partners, diving deep into R&D reviews on next-gen product architectures, to recruiting top-tier global talent. That intensity keeps me energized and curious.

Nicolai Tangen: What's the way you look into the future? Who do you speak to and what do you read?

Brian Gu: Because of my diverse background, I love conversing with the brightest minds across different fields—AI scientists, hardware engineers, macroeconomists, multinational executives, and young entrepreneurs. I read widely across scientific literature, tech reports, and books. Crossing disciplines often reveals unseen synergies and opportunities where technologies converge, which is tremendously rewarding.

Nicolai Tangen: What would you think your biggest personal failure or regret in life? What has that been?

Brian Gu: I would say I always had a bit of regret that I didn't pursue my first scientific passion further. My early background was biochemistry, working under a brilliant mentor on tubulin and protein folding mechanisms. Back then, computational tools were limited and progress was tough. About 20 years later, AI breakthroughs like AlphaFold solved the protein folding problem. But you can't live five parallel lives at once; the key is to embrace choices, focus on the path ahead, and seize the next big frontier.

Nicolai Tangen: What would be your advice to young people generally?

Brian Gu: My advice for young people is: never be afraid of change; embrace change. The world is evolving at an exponential pace with technological disruptions. Don't confine yourself to a narrow comfort zone. Cultivate intellectual curiosity, resilience, adaptability, and an open mind. Disruption and change are always where the greatest opportunities are born.

Nicolai Tangen: Brian, this has been tremendous. You combine the depth of a scientist with the insight of a business leader, offering us a fascinating glance into the future. It's been so interesting!

Brian Gu: Thank you very much, Nicolai. Very nice talking to you. It's great.

Nicolai Tangen: Thank you. Really hope you liked the episode. If you did, please like it, subscribe, and put on the notification. This really helps us to get new great guests.