Horizon aims to become a "smart driving infrastructure giant"

Author | Chai Xuchen

Editor | Liu Baodan

As the "smart driving equality" competition in the automotive circle is in full swing, Horizon has thrown out a "bold statement" -- smart driving will eventually become a standard feature, and instead of frantically developing in-house solutions, "latecomer" players should invest their energy and resources into areas that can provide emotional value to users.

Recently, at the China Electric Vehicle 100 Forum, Horizon founder Richard Yu took the stage and posed a question, "Many car manufacturers and suppliers are anxious about smart driving, wondering what will happen if they don't keep up with the trend. Is smart driving the soul of the car?"

From the perspective of a smart driving supplier, Richard Yu believes that such anxiety seems unnecessary. He used the example of mobile phone industry basebands to explain this viewpoint.

"Looking back at the mobile phone industry, communication functions are very important, but most mobile phone manufacturers do not develop their own communication basebands because making phone calls is a very standardized function. In contrast, all mobile phone manufacturers will develop their own camera technology; whether a photo is beautiful or not has no standard, but it is related to the user demographic. Photography defines the brand and user group, bringing emotional value."

In Richard Yu's view, smart driving is the infrastructure of the intelligent automotive era, the most important functional value point. For everyone, smart driving is about getting from A to B safely, comfortably, and conveniently. It is like a baseband, a standardized function, but difficult to differentiate.

In other words, Richard Yu sees the smart driving territory that car manufacturers are fiercely competing for as a basic function that will be standard in future vehicles, "not providing emotional value," and will not define the brand.

Therefore, as a provider of smart driving chips and solutions, Horizon's ambition is to persuade most car manufacturers with "smart driving anxiety" to use their solutions to navigate through the turbulent cycle and keep pace with industry competition.

Richard Yu candidly stated that this may seem like a counter-consensus thought in the industry, but looking at the past 30 years of mobile phones, mobile phone manufacturers have experienced a transformation from everyone developing their own communication basebands to 80% relying on suppliers after the emergence of Texas Instruments and Qualcomm.

Richard Yu believes that when functional value undergoes a technological leap, manufacturers either develop in-house or use suppliers, but the unchanging core is to outpace the slow and outperform the low. Those who fail to keep up with the speed, like "Nokia and Motorola," will be overturned by the times. "Automotive smart driving, like mobile phone basebands, is not about whether to develop in-house, but whether it can achieve the goal and keep up with the speed of industry change."

Coincidentally, Bosch's President of Intelligent Driving Control in China, Wu Yongqiao, also believes that OEMs do not need to "reinvent the wheel," as the cost of developing smart driving in-house is very high, while the cost-effectiveness of human resources is relatively low.

He stated at the forum that OEMs need to invest a team of one to two thousand people and hundreds of millions of funds each year to develop smart driving, but they can only adapt it to a dozen of their own models. In contrast, suppliers can cover dozens of models with similar resources; more importantly, there is the issue of time, as full-stack in-house development by OEMs requires three years to streamline all processes.

Considering all factors, Wu Yongqiao believes that the cockpit will be the main battlefield for differentiated competition among car manufacturers in the future, and Richard Yu has also proposed a similar conclusion "Technology is changing rapidly. In three years, intelligent driving will achieve hands-free driving; in five years, it will achieve eyes-closed driving; and in ten years, it will achieve free driving." Richard Yu said that when intelligent driving reaches L3 and L4, it will spur in-car content consumption. "An engineer from Haidian gets off work on Friday afternoon, gets in the car to play games, watch videos, and sleep, and the next morning watches the sunrise by the sea in Qingdao. This scenario will be realized in at most ten years."

Richard Yu believes that just as the maturity of 4G infrastructure gave rise to applications like TikTok, when intelligent driving enters a higher stage, cars will no longer be traditional means of transportation, and users will spend more time consuming content in the car. For car manufacturers, the most important thing is to capture users' time and digital consumption in the car, turning it into an intelligent space and terminal, which will be the core of future intelligent strategies.

Richard Yu predicts that the endgame of intelligent driving will also follow the "80/20 rule"—20% of car manufacturers will develop in-house, while 80% will seek the strongest partners.

He revealed that since the beginning of this year, the industry has seen a wave of "intelligent driving equality," with car manufacturers like Chery, Geely, Changan, and GAC launching related products, and Horizon Robotics is an important supplier behind these launches. "We are their only or one of two suppliers. Currently, there are mainly three players at the table: Horizon, NVIDIA, and Qualcomm."

This means that "intelligent driving equality" will give rise to new industry giants, and last year, Horizon Robotics first broke through a 30% market share in the intelligent driving chip market. "I believe we will achieve the largest share this year, just like the recent nickname given to Horizon in the industry, 'the greatest common divisor in the field of intelligent driving,'" Richard Yu stated with a hint of pride.

However, when key technologies are changing at ten times the speed in the industry, whether self-developed or collaborative, it is crucial to outpace the slow and outperform the low; keeping up with the pace of technological evolution is the most important.

On stage, Richard Yu unveiled a new flagship product—the Journey 6 series flagship version—Journey 6P. He introduced that this chip has a computing power of 560 TOPS and is a high-performance intelligent driving chip aimed at urban intelligent driving scenarios, which has recently completed successful tape-out and real vehicle verification.

Horizon Robotics' "sunshine strategy" is gradually coming to light; it aims to become the NVIDIA or Qualcomm of the automotive intelligent era, playing an indispensable foundational role.

But the problem is that end-to-end has become a consensus, and the functional experiences of various intelligent driving systems are converging. What should the industry do next, and how can Horizon continue to lead and consolidate its position? Richard Yu has his own thoughts on this.

"In the AI era, everyone thinks users and traffic are important; I want to challenge this conventional wisdom." Richard Yu cited the example of the world-shocking AlphaGo, which was trained using human chess data and rated at 9 dan. Later, AlphaZero started from scratch, relying solely on virtual data and improving its chess skills through reinforcement learning in a virtual world, ultimately surpassing human levels.

Richard Yu pointed out that learning from human data will limit the upper limit of AI capabilities. "Reaching human levels is not enough; surpassing humans is the pursuit, and intelligent driving is no exception."

He bluntly stated that when Huawei launched intelligent driving two years ago, there were only a few hundred cars and not much data, but they found that 99.99% of human driving behavior was not worth learning. In the AI era, to surpass the limits of human driving, the product logic must shift from "insight into human life" to "approaching the truth of the world." With this in mind, Horizon has launched the Super Drive advanced intelligent driving system for high-level urban full-scene intelligent driving, which will be mass-produced in Q3 of this year. "With 11 cameras and 560 TOPS of computing power, it can perform a three-point turn at complex intersections in urban areas. There are probably only one or two companies in the industry with similar capabilities," said Richard Yu.

Perhaps, a new anomaly in the intelligent driving industry is quietly emerging. Driven by companies like Horizon, automobiles are stepping into the era of AI robots, and the automotive industry is about to 迎来 its most exciting moment of transformation