As the year draws to a close, a moment that typically signifies tranquility, Li Xiang, the CEO of Li Auto, stirred the waters of the automotive and tech industries with a bold announcement during the AI Talk 2024. The company is venturing into the realm of artificial intelligence (AI), a move that resonates like a stone thrown into a calm pond, sending ripples across the sectorThis decision not only stirs the automotive industry but also makes waves in the technology and investment fields.

Entering the AI arena means competing with heavyweights like ByteDance, Tencent, and Baidu in a battle that has already intensifiedThe market for foundational models and personal AI assistants has turned into a fierce battlefieldAs a company primarily focused on electric vehicles, the question arises: Why has Li Auto chosen to embark on this challenging yet opportunity-rich journey into AI?

This decision reflects a keen understanding of market trends, a strategic alignment with the company’s growth objectives, and a commitment to addressing evolving customer needs

In 2024 alone, Li Auto successfully delivered over half a million new vehicles, achieving a remarkable landmark as the first new force car manufacturer to reach an annual sales milestone of 500,000 unitsCumulatively, the company has delivered 1.133 million vehicles, setting itself apart from other contenders in the new energy vehicle sector and demonstrating formidable resilience.

The strategic transformation marks the dawn of a new competitive era, generating excitement and curiosity as the new year approachesWith advancements in 5G, cloud computing, and AI, automobiles are evolving into sophisticated mobile devices that integrate travel, entertainment, and work functionsLi Xiang emphasized that cars are transitioning from being mere modes of transport in the industrial age to becoming “space robots” in the AI era.

Market research indicates that the demand for smart vehicles is soaring

• Battery Investment Soars, But Output Growth Slows
• Sustained Growth of Spicy Snack Foods
• How to Profit in the A-Share Market?
• Gree's "Core" Strategy: A Breakthrough or Bust?
• Tuda Pursues Hong Kong Reverse Takeover

Over the past five years, the penetration rate of connected vehicles globally surged from less than 10% to nearly 40%. Projections suggest that by 2030, this figure could exceed 70%. In China, the world’s largest automotive market, the transition towards intelligent vehicles is acceleratingThis upsurge in smart technology is reshaping not only the competitive landscape of automotive products but also redefining consumer purchasing criteria.

Today, features such as intelligent driving assistance, voice interaction systems, and connected vehicle services have become critical factors for consumers when purchasing vehiclesLuxury automotive brands in Europe have also ramped up their investments in intelligent driving and cabin technology, while players like Huawei, Xpeng, and BYD are vigorously developing competitive smart vehicle products.

At Li Auto, the core value has always been about exceeding customer expectations by creating products and services that the company can be proud of

Through extensive research and analysis of customer desires, Li Auto has discovered that consumers have high expectations regarding the accuracy of voice assistants, the safety and reliability of intelligent driving assistance systems, and personalized smart cabin features.

This indicates a significant shift on the horizon where smart vehicles, once considered luxury items, are becoming essential commoditiesHowever, as competition intensifies, Li Auto realizes that to distinguish itself, it must embrace the current trend of smart technologyIncorporating AI could offer the company fresh avenues for growth.

Industry analysts point out that once Level 4 autonomous driving becomes a reality, the demand for personal car ownership might declineYet, Li Xiang sees this differentlyHe argues that the costs associated with family vehicles would drop, prompting more people to opt for personal cars

"Why do we buy houses? We seek high-quality companionshipWe create a stable, safe, and comfortable environment for our familiesA more efficient and enjoyable space makes ownership even more appealing,” he remarks.

In September 2022, Li Auto committed to prioritizing AI technology, particularly autonomous driving, marking it as a critical focal point for the futureAt the beginning of 2023, the company transitioned its AI integration strategy from a hidden agenda to an open one, emphasizing transparency in its objectives.

"This open strategy is essential for attracting the right talentThe advent of OpenAI has played a crucial role in ushering us into a new era of AIWe no longer need to conceal our ambitions for AI; we should present them openly," Li Xiang stated.

Beyond the market dynamics, the evolution of technology plays a pivotal role in this transformation

The Fourth Industrial Revolution’s core is AI"People often say that electrification is the first half, and intelligence is the second halfHowever, I believe that the conversation around intelligence isn't about traditional software but genuine AI," Li Xiang explains.

According to OpenAI's framework, the development of artificial general intelligence (AGI) is divided into five stages, ranging from chatbots capable of basic dialogue (L1) to innovative agents capable of advancing technology (L4), ultimately leading to an organization that autonomously manages complex human endeavors (L5). AGI should not just replicate human knowledge and skills but must encompass wisdom—a reservoir of high-quality human insight.

From a technological evolution standpoint, this AI era marks a time defined by capabilitiesThe mantra "model equals application" rings true, echoing sentiments from industry leaders like Yang Zhiling, founder of Dark Moon

"Good products emerge from the evolution of model capabilities," he asserts.

Li Auto has established four AI teams focused on addressing challenges in intelligent driving, smart assistants, intelligent commerce, and industrial intelligence, yielding impressive resultsNotably, the efforts in intelligent commerce and industrial applications are examples of the company harnessing model capabilities effectively.

For instance, Li Auto’s integration of large model technology with business sector data has produced AI assistants equipped for retail, marketing, customer service, financial fraud prevention, and financial analysisIn retail scenarios, these assistants empower many partners by learning communication skills from top retail experts and alleviating pressure on customer service teams by addressing various user inquiries efficiently.

Furthermore, the company’s manufacturing facility in Changzhou isn’t merely focused on production; it acts as an evolving machine with a complete self-developed core system, full automation, digital twins, AI empowerment, and smart logistics—all aimed at continuous improvement.

The Li-Mos smart manufacturing super brain, launched in July 2023, is an all-encompassing production management system that manages the entire manufacturing lifecycle of each Li vehicle—from production planning to logistics, stamping, welding, painting, and assembly.

With AI's rapid advancements, Li Xiang foresees a convergence between seemingly disparate domains—AGI and fully autonomous driving are likely to intersect soon, merging under a unified framework

In such circumstances, various enterprises will compete on a shared playing field, as “the foundational models act as the operating system and programming language of the AI era.” Hence, companies must vie for “AGI agents and Level 4 autonomous driving,” akin to the monumental shift seen with the advent of smartphones.

The competitive landscape is evolving as the advent of AI demands research that extends beyond just the technology itself but also includes its interplay with human cognition"A business must prioritize research before tech development, leading to product realizationNeglecting any of these steps is akin to crafting an ornate decoration on an insufficient foundation,” Li Xiang reasons.

Fundamentally, Li Auto's prowess entails recognizing its AI ambitions alongside its deeper aspirationsFrom the very beginning, the company has envisioned itself as more than just a car manufacturer

Li Xiang narrates a story of the internal debate on whether the branding should center around "Li Auto" or simply "Li." He revealed that whilst “Li Auto” was a convenient label for communication and marketing, the notion of “Auto” was not included in their logo.

"Our mission transcends the mere intelligentization of cars; it's about embedding AI into automobiles, thereby extending technological advantages to every household," Li Xiang declaredThe goal is to harmonize the physical and digital realms, realizing unlimited possibilities in a finite space.

To achieve this, Li Auto seeks to significantly amplify its research funding and clarify its strategic objectivesNearly half of the more than $15 billion spent annually on R&D is directed toward AI initiativesFour AI teams are currently tackling complex issues ranging from autonomous driving to intelligent assistants and industrial intelligence.

While their accomplishments in industrial applications are notable, it is in intelligent driving and smart assistants where Li Auto has made remarkable headway, capturing widespread attention

"We are the first automotive company globally to develop foundational modelsOur end-to-end and VLM approach has been among the earliest to transition from foundational papers to technological research to actual product delivery," Li Xiang remarked.

Specifically, the company introduced an innovative dual system solution, the industry's first, which integrates end-to-end and VLM capabilitiesSince the initiation of the "Early Bird Plan" in July 2024, updates have been implemented bi-weekly, with a comprehensive rollout planned for late October, featuring “parking space to parking space” functionality, and an advanced version slated for a January release.

By the end of November 2024, Li Auto achieved its millionth vehicle sales milestone, thanks to a booming user base, gathering a staggering 2.9 billion kilometers of intelligent driving experienceThis significant data accumulation, coupled with sophisticated training algorithms and world models, enhances the performance of autonomous driving systems.

Operating under a comprehensive strategy and substantial funding, the company has set clear objectives

For intelligent driving, they aim to achieve Level 3 supervised autonomous driving by 2025, laying the groundwork for advancing to Level 4 and Level 5 capabilitiesAttaining Level 4 autonomy necessitates building a fleet of over five million vehicles, mastering VLA foundational models, and securing sufficient capital to recruit top-tier talent and computational resources.

“We anticipate achieving handover capabilities of 500-1000 kilometers in the next two years," Li Xiang shares"My personal aim is to unveil the first car designed for Level 4 autonomy within three years.”

Moreover, in terms of foundational models and their applications, Li Auto has firm ambitionsLi Xiang asserts, “We must ensure our foundational large language model ranks among the top three in the industry within China over the next few yearsWe are prepared to invest the necessary computational resources to compete with leading enterprises.”

Building on their foundational models, Li Auto is making notable strides in expanding application scenarios

For instance, based on Mind GPT, the Li learner initiative is transitioning from in-vehicle systems to smartphones, with an app already fully launched by December 27, 2024. There are future plans for Li learners to extend beyond vehicles to various devices, ensuring a consistent AI experience.

Li Xiang articulates his aspirations for AGI, a realm he envisions with clarity and depthHe categorizes the evolution of AGI in three stagesThe first stage, termed "enhancing my capabilities," embodies AGI as a supportive tool, maintaining human decision-making authorityThe second phase, "becoming my assistant," sees AGI capable of independent task completion, shouldering responsibility for the outcomesThe aspirational third stage, which Li calls “silicon-based family,” would see AGI as an integral member of the household, potentially organizing family affairs without human directives.

Li Xiang views this progression as hopeful—he and his team are poised to achieve this milestone within their lifetime and wish to witness similar advancements by other industry leaders

To realize this vision, he posits that ensuring right decisions in the AI domain and nurturing top-tier talent is criticalHe emphasizes the necessity of an expansive and profound experience in product development, saying, "Lack of these fundamentals could undermine future efforts.”

As a technology-driven entrepreneur, Li engages in several AI meetings weekly, exploring cutting-edge papers and varying teams' best practices, which sharpen his understanding of the AI frontierThe company’s capital and tech teams also invite top industry professionals for collaborative discussions.

Reflecting on his formative experiences in secondary education, he stresses the importance of cultivating problem-solving abilities through broad reading and the practical application of knowledge"My journey embodies a learning, validating, and growing cycle akin to AI's pre-training and reinforcement learning frameworks,” Li Xiang notes.

On nurturing leadership talent, Li aims to translate his methodologies into the competencies of his team and employees