This article was originally published on Goldman Sachs Insights, which features analysis and perspectives on the global economy and markets from across Goldman Sachs.
Though some robotaxis, which are autonomous vehicles (AV), are already on the road in cities such as San Francisco, Phoenix, Wuhan, and Beijing, the technology has yet to be broadly deployed.
While 60% of current vehicles have some level of driver assistance, only 1% - 2% of total global vehicles sales in 2026 are expected to have Level 3 features – where drivers can take their eyes off the road and their hands off the wheel in select situations, such as on highways in clear weather.
Although higher levels of driving automation haven’t been implemented as quickly as Goldman Sachs Research had expected, there are signs that partial automation and assisted driving are becoming more widespread.
“The bottom-line is that we believe improved AI technology will help the industry reach higher levels of performance, although we also believe that wide scale AV adoption is still at least a few years away as a base case,” Mark Delaney, who covers automobiles and industrial technology for Goldman Sachs Research, writes in his team’s report.
Old vs. new GS forecasts for autonomy penetration by level globally
Source: Company data, Goldman Sachs Research
By 2030, up to 10% of global new car sales could be Level 3 vehicles. (The previous forecast was around 12%.) Sales of fully autonomous cars – Levels 4 – could amount to around 2.5% of total sales in the same timeframe (compared with the previous forecast of around 3.5%).
At the same time, partially autonomous Level 2 / Level 2+ vehicles that require driver supervision are forecast to rise from about 20% of sales this year to about 30% in 2027 (the previous forecast was around 24%).
Degree of autonomy
Examples of features
Degree of autonomy |
Examples of features |
|
---|---|---|
L0 |
No automation: manual control. The human performs all driving tasks. |
No automation |
L1 |
Driver assistance: The vehicle features a single automated system. |
Automatic emergency breaking, lane centering, or adaptive cruise control |
L2 |
Partial automation: The vehicle can perform steering and acceleration. The driver still monitors all tasks and can take control at any time. |
Lane centering and adaptive cruise control (at the same time) |
L2+ |
Partial automation: The vehicle can perform steering and acceleration, with quasi auto-pilot, but the driver is always alert/responsible and hands near wheel. |
Lane centering and adaptive cruise control (at the same time); quasi auto-pilot with enhanced security features |
L3 |
Conditional automation: The vehicle performs most driving tasks, but human override is still required. OEM liable aside from when driver warned to take over (subject to grace period). |
Traffic jam chauffeur |
L4 |
High automation: The vehicle performs all driving tasks under specific circumstances. Human override is still an option. |
Local driverless taxi |
L5 |
Full automation: The vehicle performs all driving tasks under all conditions. Zero human attention or intervention is required. |
Vehicle can drive everywhere in all conditions; pedal/steering wheel may not be installed |
Source: SAE, NHTSA, Company data, Goldman Sachs Research
The forecast also implies that, by 2030, a global fleet of a few million commercial AVs will be used for ridesharing. While that’s a sliver of the total cars worldwide, it would mean a market of more than $25 billion for robotaxis.
There are signs that AI advances could accelerate the adoption of vehicles that are substantially more autonomous. “Research on AI scaling does suggest that added computers, larger training datasets, and improved model architectures should contribute to better AI model performance,” Goldman Sachs Research analysts write.
Lower costs of hardware are another potential reason AV adoption may increase. Driver assistance and fully autonomous vehicles use dozens of cameras, sensors and, in some cases, light detection and ranging (lidar) devices. For example, a certain Level 2+ car on the road today uses 8 cameras, while a particular Level 4 vehicle uses 29. As the costs of these components drop, AVs will get cheaper and more efficient.
Looking out even further, Goldman Sachs Research sees a bull case scenario in which AV sales (Level 3 automation or higher) account for about 60% of all light vehicle sales in 2040. Even in a less optimistic scenario, AVs will likely make up close to 40% of new sales.
AV adoption rates are anticipated to be the highest in China, where Level 3 or higher AV sales could account for 90% of all sales by 2040, according to Goldman Sachs Research analysts. Nearly 80% of all car sales in Europe and roughly 65% of all car sales in the US could be advanced AV vehicles by 2040.
Key regions Level 3/4/5 penetration rate
Source: Company data, Goldman Sachs Research
A large number of these vehicles are expected to be deployed by ridesharing companies. Increasingly, the cost model makes it compelling for rideshare firms to switch to self-driving cars. Goldman Sachs Research finds that vehicle driving costs are currently an estimated $3.13 per mile for robotaxis but may decrease to less than $1 a mile by 2030 and 58 cents a mile by 2040. Robotaxi costs that factor in corporate overhead and research and development are significantly higher – but that’s poised to fall from an estimated $184 per mile for a vehicle in 2024 to about $12 a mile in 2030 (and close to $1 in 2040).
As rideshare operators scale up their AV fleets, Goldman Sachs Research analysts expect a gradual shift in the supply-side industry structure - from highly fragmented (i.e. millions of individual drivers) to more consolidated (i.e. a handful of AV fleet operators). But even if AVs were to gradually be deployed in certain geographies over the next 3-5 years, it is likely they will operate as supplemental supply for specific routes as opposed to being the only option.
Goldman Sachs Research notes that the most profitable routes are often the most complex to solve from an AV technology perspective (such as airport pick-up / drop-offs, late night pick-up / drop-offs to nightlife in crowded city streets). Over the near-to-medium term, a hybrid model combining AVs and human drivers is likely to ensure widespread availability of vehicles and a better user experience for riders.
This article is for informational purposes only and is not a substitute for individualized professional advice. Articles on this website were commissioned and approved by Marcus by Goldman Sachs®, but may not reflect the institutional opinions of The Goldman Sachs Group, Inc., Goldman Sachs Bank USA, Goldman Sachs & Co. LLC or any of their affiliates, subsidiaries or divisions. Information and opinions expressed in this article are as of the date of this material only and subject to change without notice. This article is not a product of Goldman Sachs Global Investment Research. The information contained in this article does not constitute a recommendation from any Goldman Sachs entity to the recipient, and Goldman Sachs is not providing any financial, economic, legal, investment, accounting, or tax advice through this article or to its recipient. Neither Goldman Sachs nor any of its affiliates makes any representation or warranty, express or implied, as to the accuracy or completeness of the statements or any information contained in this article and any liability therefore (including in respect of direct, indirect, or consequential loss or damage) is expressly disclaimed.
Join our Marcus social media community, where we share content and inspiration to help improve your financial health. See you there!