Vol.07 LLMs on Wheels: How "Attuned" Could Cars Get in 2025?

AI technology continues to leap forward, and our daily lives — what we wear, eat, where we live, how we get around — are being profoundly reshaped by this technological storm.

This episode focuses on "AI + Mobility," exploring the development and application of large language model technology in automotive intelligent cockpits. We invited Li Pan, CMO of Yitu Technology, and Sang Yu, tech investor at Yunqi Capital, to jointly analyze the technical evolution, pain points, competitive landscape, and future trends of LLMs in vehicles — revealing how AI is transforming cars from "transportation tools" into "intelligent spaces that understand you."

Host

Linda, Managing Director at Yunqi Capital

Guests

Li Pan, Co-founder and CMO of Yitu Technology

Master's degree from Shanghai Jiao Tong University, with 15 years of experience in product-market-ecosystem strategy and brand management in the intelligent connected vehicle industry, as well as pioneering multiple innovative business initiatives. Previously served as head of intelligent product at a top domestic OEM and head of product ecosystem at a leading domestic connected car company. Specializes in intelligent vehicle product innovation planning and definition, as well as ecosystem resources and business model design for the internet and tech industry.

Sang Yu, Tech Investor at Yunqi Capital

Graduated from the Department of Electronic Engineering at Tsinghua University, focusing on cutting-edge technology fields including embodied intelligence and intelligent driving.

Timeline

• 01:03 Opening: Meet the guests
• 02:41 The technical evolution of LLMs in vehicles: from mechanical interaction to proactive service
• 11:33 Every family will need three robots in the future: Home Robot, Pet Robot, and Mobility Robot
• 19:28 Industry pain points: OEM architecture, data algorithms, and ecosystem dilemmas
• 29:42 Yitu Technology's breakthrough strategy: tiered products and AIOS system
• 32:41 OEM partnership models: differentiated needs of traditional automakers, new forces, and cross-sector tech giants
• 38:02 Smarter means more power-hungry? Balancing computing power and energy consumption
• 41:08 Future predictions: 2025 inflection points for intelligent cockpits and autonomous driving
• 44:10 Car-buying advice from industry insiders

Glossary

SOA (Service-Oriented Architecture)

In plain terms, "automotive SOA-ization" means turning vehicle functions into "software services" that the onboard computing platform can call up and update, much like a smartphone.

TTS (Text-to-Speech)

Siri, Tmall Genie, Baidu's Xiaodu, and voice navigation in map apps all rely on TTS technology.

"NVIDIA Thor"

NVIDIA's dedicated in-vehicle computing platform designed specifically for intelligent vehicles, primarily used for autonomous driving and intelligent cockpit tasks. It represents a significant leap in computing power over the previous NVIDIA Orin platform.

TOPS (Tera Operations Per Second)

The unit for measuring AI chip computing power. 1 TOPS = 1 trillion operations per second.

Key Takeaways

1. Core Value of LLMs in Vehicles

From "passive Q&A" to "proactive service": upgraded voice interaction, scenario-based intelligent agents (such as itinerary planning, emotional companionship).

Multimodal fusion: collaboration of voice + vision + vehicle control data to create a "human-understanding" mobile space.

2. Technical Challenges and Breakthroughs

• Technical evolution: TTS → large model end-to-end technology → agents, multimodal
• Ecosystem reconstruction: automotive underlying architecture needs to be redefined
• Data bottleneck: in-vehicle scenario data is fragmented, requiring combination of domain-specific models and reinforcement learning

Breakthrough strategy: tiered products and technical architecture

3. Major OEM Partnership Models

• Lightweight cooperation: directly providing intelligent agent modules (such as voice enhancement, coffee ordering)
• Deep customization: combining supplier model capabilities with OEM hardware capabilities
• Full-stack self-development: outsourcing engineering implementation needs

4. 2025 Predictions

• Intelligent cockpit: end-to-end voice interaction deployment, AI Agent and AIOS率先实践 in automotive scenarios
• Autonomous driving: VLA models solving complex game-theoretic challenges, end-to-end control improving the autonomous driving experience

For more insights on LLMs in vehicles and other AI industry observations, you can find more of our analysis here:

WeChat Official Account: Yunqi Capital Contact us: community@yunqi.vc

View episode transcript on Xiaoyuzhou