From AI to AV (autonomous vehicles): Which Chinese Stock is UOB KH Big On?

THE CONTEXT • Self-driving vehicles are emerging in Singapore, initially focusing on public housing estates and with a big push expected in areas starting with Punggol.​  • Singapore will see autonomous electric buses deployed on bus services 191 and 400 from 2026 to enhance last-mile connectivity. • In the US, Tesla is stealing the show with its robotaxis launched for commercial rides since June 2025 in Austin, Texas, with expanding services in other US cities planned. The unique strength of Tesla's self-driving technology is its slick system of cameras and high AI compute, unlike Chinese players which rely on a mix of cameras and radar and lidar devices and, of course, AI. • In China,  the leader is Baidu's Apollo Go which has racked up 14 million cumulative passenger rides worldwide as of August 2025. Currently, Apollo Go operates in 16 cities including Beijing, Wuhan, Hong Kong, and Dubai.  • A new UOB Kay Hian report has a comprehensive take on the industry in China, with a buy call on Baidu. Read excerpts below....

Excerpts from UOB KH report
Analysts: Julia Pan Meng Yao & Soong Ming San

Cost Efficiency As Performance Cornerstone And Robotaxi Reshape The Future

Analysis
• Baidu’s Apollo Go achieves breakeven across multiple Chinese cities, laying a promising foundation for overseas expansion and larger-scale robotaxi deployment.

However, it will still take some time before the overall business unit becomes profitable.

On 25 Sep 25, Apollo Go showcased Dubai’s first autonomous driving test licence, Plate No. 001 at the Dubai World Congress for Self-Driving Transport event.

Dubai’s Roads and Transport Authority (RTA) had granted Apollo Go the first batch of 50 autonomous driving test licences in July, doubling its fleet in the UAE to around 100 vehicles.

Under a strategic cooperation agreement signed in March this year, Apollo Go plans to deploy more than 1,000 fully driverless vehicles in Dubai as the first phase of its rollout.

Meanwhile, Apollo Go is now exploring market expansion into regions such as Australia and Southeast Asia.

• Promising progress in delivery and growth potential of Baidu’s robotaxis.

Baidu currently operates over 1,500 autonomous vehicles globally, counting only cars deployed and taking rides.

By end-25, the fleet will be 3x that of 1Q25.

Over the next five years, including self-owned and third-party fleets, the goal is to achieve hundreds of thousands of vehicles.

Overseas partnerships with Uber/Lyft are progressing, while Baidu also has agreements with T3, Hello Mobility, and CAR Inc in China. Apollo Go provided about 2.2m rides in 2Q25, up 148% yoy.

Currently, Apollo Go operates in 16 cities worldwide, including Hong Kong, Dubai, and Abu Dhabi, having completed over 14m autonomous rides and accumulated more than 200m kilometres of safe driving.

• Resilient Didi ride order data during Golden Week 2025. Daily ride-hailing demand rose 14% yoy.

Cross-city ride-hailing demand grew 89% compared with pre-holiday levels, reflecting both the trends of “exploring outward” and “relaxing nearby” that are reshaping consumers’ local holiday travel patterns.

Pre-booked ride-hailing demand grew 58% vs before the holidays.

• Significant cost-efficiency optimisation among leading players. We see accelerating commercialisation via the enabling of larger scale and reduction in costs. Baidu and PonyAI have already achieved domestic substitution of key components, bringing costs down to just 33-50% of their overseas peers.

For instance:
a) PonyAI’s seventh-generation hardware system cost fell 70% to Rmb270,000 from Rmb1m in its first generation, and
b) Baidu’s 6th generation autonomous vehicle cost is 40% lower than its predecessor at Rmb200,000 from Rmb480,000.

While pricing models are still being explored, robotaxi services are expected to be 10-20% cheaper than traditional taxis and ride-hailing.

• AD TAM forecasts. By 2030, we forecast a global fleet of several million commercial autonomous vehicles (AV) for ride-sharing.

While this would represent around 1% of current global car parc (~1.5b vehicles), it could still generate a market opportunity of over US$25b for robotaxi-based personal mobility.

This estimate is influenced by factors such as average selling prices (ASP), daily trip volumes, and average trip distances, with regional business mix further shaping revenue per trip.

• Robotaxi market expansion outlook. We estimate a robotaxi fleet of around 5,000 in 2025 (~0.1% of the shared mobility fleet), and this should reach 630,000 (10% of the shared mobility fleet) by 2030 and 2.5m by 2035.

We forecast China’s robotaxi market to expand rapidly from US$55m in 2025 to US$12b in 2030 and further to US$45b in 2035, representing a 757-fold increase over the decade and underscoring the significant market potential.

Revenue will primarily be generated from ride fares, with each robotaxi expected to generate US$69 per day by 2035 (up from US$36 in 2025), surpassing the US$28-56 per day typically earned by traditional ride-hailing vehicles due to longer operating hours.

See UOB KH's report here