

AI-powered autonomous driving: global expansion and regulatory support
Autonomous driving offers enhanced safety and convenience as well as greater efficiency and productivity for both consumers and companies. Moreover, regulation and falling technology costs are paving the way for growth, driving smart mobility investments.
概要
- Autonomous driving is already a reality in major cities worldwide
- Costs continue to fall, catalyzing mass-market adoption
- Legislative momentum in US, EU and Asia supports rollout
As of the summer of 2024, anyone with a smartphone in several US cities can download an app and hail a robotaxi at any time of day, in any kind of weather, to travel more safely than a human driver. By the end of 2024, people were doing so more than 150,000 times per week (see Figure 1).
Figure 1 – Robotaxi ridership is taking off

Source: Waymo, 2024. Paid, fully driverless rides per week by Waymo in San Francisco, Los Angeles, and Phoenix. Dashed area represents ‘over 150,000’ rides per week, extrapolating toward new market openings in 2025.
Under the proverbial hood, autonomy is enabled by sensors such as cameras, radars, and lidars, whose data is processed by AI systems-on-chip. The algorithms have often been trained on billions of kilometers of data collected from vehicles already on public roads.
This scale is easier for some companies. Tesla, for example, has seven million (and counting) vehicles on the road designed to continually contribute to improving its ‘Full Self-Driving’ system (see Figure 2). Mobileye’s ‘Road Experience Management’ system crowdsources maps from vehicles of different manufacturers.
The technology underpinning autonomous vehicles (AVs) affords them capabilities and experience beyond those of any human driver; it also gives the ability to be much safer, not least because they are never drunk or distracted, and never need a rest. A study by Swiss Re of over 40 million fully autonomous kilometers driven by Waymo vehicles found that they resulted in around 90% fewer insurance claims compared to even the most advanced human-driven vehicles. 1
Figure 2 – Tesla is racking up billions of autonomous miles

Source: Tesla, 2024. Full Self-Driving (FSD, Version 12) is Tesla's most advanced autonomous driving system, currently offering advanced level 2 autonomy ('supervised') and designed to support level 3 and 4 autonomy ('unsupervised'). The AI models which underpin it are trained on the billions of miles driven by Tesla vehicles and designed to scale across the entire fleet.
Scale brings better economics
AVs are a reality; a key part of scaling is reducing the cost of autonomous driving technology. The vehicles currently providing Waymo’s service in San Francisco, Los Angeles, Phoenix, and Austin are estimated to cost around USD 150,000 to produce. The next generation of hardware will have increased performance at a significantly reduced cost. 2
Tesla, Wayve, and others have taken the approach of directly developing their systems based on low-cost hardware, using only cameras, imitating humans. Tesla has announced it will run a limited robotaxi service from the summer of 2025 in Austin. Once fully confident, it can roll the system out to most of its vehicles already on the road.
The technology underpinning AVs affords them capabilities and experiences beyond those of any human driver
Giving people back their time
In time, robotaxis will offer cheaper mobility than ride-hailing and taxis and will likely also be cheaper than owning and driving a car oneself. AVs will be utilized much more efficiently than the average private car, which spends 95% of the time parked.3
Beyond passenger vehicles, autonomous trucks are also on the cusp of commercial service on public roads in the US and China. Other than saving on the driver’s salary, working vehicles benefit from the fact that an autonomous driving system doesn’t go on strike, doesn’t fall ill, doesn’t need a break, and is easily replicable, addressing labor shortages.
In the US alone, over five trillion kilometers are driven annually, with the average driver spending an hour a day in their car. Whether commuting, shopping, or ferrying children around, driving is usually an activity which human drivers would prefer to forego in favor of a more productive or necessary activity or simply resting. For those unable to drive because of age or infirmity, AVs offer liberation. The billions of hours of (mostly unpaid) labor performed by humans driving thus suggests that the market for autonomous driving could be in the trillions of dollars.
AVs will be utilized much more efficiently than the average private car, which spends 95% of the time parked
Regulation – an enabler, not a barrier
As with human-driven vehicles, questions of liability, insurance, and vehicle approval must be addressed. As of 2024, a majority of US states have enacted AV deployment statutes, primarily in the Sun Belt. The possibility of legislation being passed at the federal level could further accelerate development.
European countries have taken a more cautious approach. Nonetheless, the first vehicle sold to the public with ‘hands-off, eyes-off’ level 3 autonomy (see Figure 3 for autonomy definitions) anywhere in the world was in Germany in 2022. This was enabled by the country’s adoption of UNECE R157, a regulation now implemented in more than 50 countries.4 Initially allowing ‘traffic jam pilots’ up to 60 km/h, the regulation has since expanded to allow up to 130 km/h, covering motorway speeds.
As of March 2025, legislation will be in force in Switzerland permitting autonomous driving on motorways as well as completely driverless robotaxis under certain conditions. 5
The Chinese government views autonomous driving as a strategic priority, and the country is home to a vibrant technology and hardware ecosystem. Local drivers are fond of ADAS6 features, with 15% of new cars sold featuring the most advanced forms of ADAS by the end of 2024 (see Figure 3).
While it regulates the commercial operation of AVs at a national level, implementation is at the municipal level. Baidu’s Apollo Go offers a fully driverless paid service 24/7 in Wuhan, and alongside peers such as Pony.ai and WeRide operates with some restrictions in parts of cities such as Beijing, Guangzhou, Shenzhen, and Chongqing.
Figure 3 – Varying degrees of autonomy

Source: Robeco, 2024.
A vast new market is emerging
Fully autonomous vehicles are a reality today and in 2025 will continue to go mainstream. The improvement in the cost of the technology which allows this industrialization also means that the advent of privately owned, mass-market vehicles capable of autonomous driving is just around the bend. The application of AI to automate a universal task and give people their time back will transform the way we move, creating a vast new market in the process. This is creating new avenues for growth across the existing smart mobility supply chain, including for the makers of sensors, processors and other semiconductors; as well as for the developers of self-driving technology and AV fleet operators.
Important note: The companies cited in this article are for illustrative purposes only to demonstrate the investment strategy on the date stated. The companies are not necessarily held by the strategy. Moreover, no inference can be made on the future development of the company.
Footnotes
1 Di Lillo et al, 2024. “Do Autonomous Vehicles Outperform Latest-Generation Human-Driven Vehicles? A Comparison to Waymo's Auto Liability Insurance Claims at 25 Million Miles.”
2 Waymo, company post, August 2024. ‘Meet the 6th generation Waymo driver.’
3 US Department of Energy, August 2024. ‘Household vehicles were parked 95% on a typical day in 2022.’
4 The United Nations Economic Commission for Europe (UNECE) is a regional UN commission focused on improving economic cooperation between member countries. Regulation 157 focuses on ensuring safety and efficiency in transportation.
5 Swiss Association of Autonomous Vehicles. February 2025. Autonomous driving will be allowed on motorways when a human driver is available to take back control (level 3 autonomy). Completely driverless robotaxis will also be permitted on selected roads, as long as a teleoperator is available to suggest a manoeuvre if they get stuck.
6 Advanced Driver Assistance Systems: features such as adaptive cruise control and automatic lane-keeping, which in their most advanced forms can fully control the vehicle but require the driver to keep their eyes on the road at all times.
重要資料
本網站僅供《證券及期貨條例》(香港法例第571章)及其附屬法例所界定之專業投資者瀏覽及使用。 投資涉及風險。過往表現並不代表未來表現。本網站所載資料僅供參考之用,並不構成任何投資建議,亦非作出買賣任何證券或採納任何投資策略之要約或招攬。投資者不應僅憑本網站提供之資料作出投資決定,在作出任何投資決定前,應徵詢獨立意見(包括有關稅務影響之意見)。投資者應確保完全理解投資產品的相關風險,亦應考量自身投資目標及風險承受水平。投資乃閣下之個人決定。除非銷售投資產品的中介人已向閣下告知該投資產品適合閣下,並已解釋其符合閣下投資目標之原因,否則閣下不應投資。請參閱相關發售文件或其他法律文件,以獲取包括風險因素在內的進一步詳情。 本網站由荷寶投資管理香港有限公司發布,該公司受香港證券及期貨事務監察委員會(「證監會」)規管(中央編號:APU851)。本網站未經證監會審閱。 無法保證任何投資產品可實現其投資目標。概不就任何投資產品之表現或投資回報作任何聲明或承諾。投資的價值或會波動。本網站所載過往表現、推算或預測,均不應視作未來表現之保證或指標,且概不提供任何明示或暗示之保證。本網站內容建基於相信為可靠之來源,惟因應資料傳遞技術特性及須採用多項數據來源(包括第三方內容),故概不保證其準確性。所述觀點僅乃截至上述日期,或會隨市況變化而改變,可予更改而毋須另行通知。該等意見可能有別於其他荷寶投資專業人士之意見。因使用本材料或當中所載任何評論、意見或估算而引致之直接、間接或相應損失,荷寶概不承擔法律責任。荷寶並無責任更新本網站或任何網站內容。未經荷寶事先書面許可,不得複製、分發或刊發本網站任何材料。 除非另有說明,資料來源:荷寶。























