Three step change: Autonomous vehicles, emerging business models and new data partnerships

Highly networked, data-driven autonomous mobility business models are emerging, led by a variety of industry participants, aggressive investment, and regulator and government support. These will shape the transportation of people and goods, as well as additional services made possible by autonomous systems. With the commercialisation of autonomous mobility becoming a reality, (re)insurers need to consider how they best engage with these new business models.

Figure 1: Demand side of autonomous vehicles

Business models from a customer perspective

A. Transportation of people

The use of autonomous vehicles (AVs) to transport people is gaining traction. A number of driverless projects such as Waymo One, are being tested in preparation for commercial rollout. For example, drivers in some jurisdictions, including Germany, will soon be legally allowed to take their eyes off the road in certain stop-and-go traffic situations up to certain speed limits. Robotaxi services are also becoming a reality. Intel's Mobileye will launch robotaxi services in Munich in 2022. In China, tech giant Baidu's commercial driverless taxi service Apollo Go is targeting scalable fully autonomous ride hailing operations by 2025 and aims to take 50% market share in ride hailing by 2030. Apollo's fifth generation robotaxi has achieved a 60% drop in cost per mile, and is on track to be cheaper than drivers for ride hailing by 2025.

Over time these business models could evolve in multiple dimensions tailored to different customer needs related to comfort, ownership and pricing models. (see Figure 2).

Figure 2: Design options for people transport business models

Source: Swiss Re Institute

Note: The shared owned mode is when the vehicle is owned by one party but made available as a shared service to other parties. The owned used mode is when the vehicle is both owned and used by the same party.

  • Ownership modes: AVs can be owned by individuals; but a high unit price suggests fleet structures of hire AVs are initially more likely. Shared use carries a different risk profile compared to single use vehicles as a result of risk accumulation arising from multiple journeys across the fleet.
  • Pricing models: Customers could make one-time payments for vehicles; rent or subscribe to an AV for a fixed time period; or order robo-taxis for specific journeys. Robo-taxi services on demand could operate one-to-one door-to-door; or on semi-permanent routes with multiple passengers. Each scenario carries different risk implications.
  • Degrees of comfort: Services will differ through the features they offer (interiors, entertainment, etc.) as well as the quality of the ride (affordable pooled versus exclusive premium).

B. Transportation of goods

Vehicle manufacturers will also focus on goods transport and logistics in order to monetize R&D and sustain needed funding for the next decade. In the United States, 26 states have approved autonomous trucking for commercial use (with a further 18 states permitting testing). Spurred by events such as the TuSimple IPO, large scale use of delivery AVs may soon become commercial reality, particularly on repeatable long-haul routes.

Multiple business models may evolve, depending on routes, delivery models and type of vehicle ownership (see Figure 3).

Figure 3: Design options for goods delivery business models

Source: Swiss Re Institute

Near term AV business models for goods transport will involve fixed routes in non-complex driving environments. Over time, as AV technology improves, the possibility of greater route flexibility will enable a wider range of business models, including last mile delivery.

  • Fixed Route (B2B) – The majority of the distance in longer distance delivery is on highways or motorways, less-complex environments with fixed routes. AVs will enable faster fixed route trips without the need for mandated driver breaks. Currently, semi-truck drivers are legally limited to 11 hour shifts in the US; L4 autonomous trucks could, by contrast, operate 22 hours. The impact of AVs on the economics of long-haul freight and trucking business models will be significant. Insurers will have to understand the risk profiles of these changes and consider incorporating them into their pricing and conditions accordingly.
  • Flexible Routes (B2B and/or B2C) – In the longer term, AVs adoption could expand from fixed to flexible routes. The last mile of a truck journey is typically the costliest and most complex.
  • From a fixed point (A) to multiple destinations (N): This could be used in cases of multiple delivery, such as takeaway food.  
  • From point (N) to point (N) – Routes could originate from multiple locations to travel to multiple destinations. Retailers such as Walmart or Amazon are already opening their last mile logistics networks to deliver goods on behalf of other companies. We could see this used in AVs as well.

C. Mixed models: Transportation of goods and people

Integrating passenger and freight transport will provide better returns on assets, while reducing congestion, energy use and emissions. This could reduce time assets are not in use, for example a passenger AV repurposed to deliver goods at night. The potential for multiple use AVs, lowering total cost of ownership (CTO), is considerable (see Figure 4).

Figure 4: Business model flexibility – transporting people and delivering goods and/or different verticals (use case flexibility)

D. Platform business models in autonomous vehicles

The distribution of AV services will be dependent on platform delivery. If the owner of the AV has monopoly ownership of its distribution platform, it will deal directly with customers. However, if third parties begin to dominate the platform space, it will increase competition, demand intermediation fees, and shave margins. It may be that multiple third parties seek to become intermediaries; or even that a super ecosystem emerges that aggregates services from multiple platforms (see Figure 5). This will pressure profitability, with implications for all involved.

Figure 5: Customer access to services and evolution of increased AV model competition

Business models from the provider perspective

We do not know exactly how AV business models will develop, what services will be offered or how the (large amounts of) data will be used and shared. Insurers will have to consider mapping emerging ecosystems to assess how they can best offer coverage.

The commoditisation of AVs will lead to new and exciting business models beyond people and good transportation.
Evangelos Avramakis, Head Digital Ecosystems R&D, Swiss Re Institute

Business models could evolve continually

Disruption caused by AV business models could start at the core of their value chain – the automation of the transportation of people and goods. The commoditisation of individual journeys could lead to much higher competition and force transportation firms to massively scale their fleet operations. A second level of disruption could come with production facilities not tied to a fixed location. AVs could support mobile coffee shops, clinics or stores that travel autonomously to different locations in response to consumer needs. Vehicles designed by Chinese AV firm Neolix can already deliver a variety of services from fast food, medical supplies and financial services to end consumers. A further step in the evolution of AV could be differentiated services (see Figure 6). Cinemas could be transformed into city entertainment experiences untethered from fixed locations and flexibly moving to where consumers are. Autonomously transported pop-up shops could move between cities and shopping centres offering new retail experiences for products and services without a base. Providers of goods and services can use AVs to take their products to their consumers almost anywhere, anytime.
Figure 6: Disruptive AV business models beyond people transportation and goods delivery

Business models will involve potential greater interaction between stakeholders:

Firms entering the AV market will be strategic in positioning their goods and services. Original Equipment Manufacturers (OEMs – the big car players) will team up with external tech vendors to develop AVs. Interwoven relationships and partnerships will allow OEMs to access AV technology and share development costs. OEMs may reach across industry to a variety of different partners. GM, for example, is invested and collaborating with Cruise, but it is also working together with Honda (OEM), Walmart (retailer) and Microsoft (tech) to develop and test their AV programme.

As full-stack AV players like Argo AI, Aurora, AutoX, Cruise and DiDi develop their technology, we also expect to see greater collaboration between technology players and ride share/fleet operators. Ford, Argo AI and Lyft have announced joint plans to launch Ford AVs on the Lyft network by the end of 2021 in Miami, Florida; with plans to roll out in Austin, Texas in 2022. Figure 7 demonstrates how potential areas of interaction could vary between different stakeholders, depending on their focus area.  

Figure 7: Key value creation areas of potential insurance involvement

Note: The columns show different AV players, and the rows show what each of the players do. The darkness indicates which specific function might be addressed by different AV players that might lead to potential data collaborative opportunities. The collaboration approach does not guarantee any automatic access to data.

AVs will supply new and higher frequency data which will support underwriting and pricing. Insurers should prioritise accessing such data.
Mahesh H Puttaiah, Senior Insurance Economist, Swiss Re Institute

Engaging in deeper partnerships beyond distribution to create value for both parties

More granular and higher frequency data from new business models could allow for enhanced insurance product tailoring. This could enable a more precise assessment of risk profiles for different vehicle makes and models; as well as tracking risk more frequently (when and where a vehicle travels). Insurers should go beyond initial conversations with partners to more comprehensive joint pilot projects and mobility working groups. This will allow them to gain access to data and market insights to focus on implications for risk and protection. The partnership approach should therefore involve data sharing and serve the interests of both parties. However, there is much ground to cover. (Re)insurers currently lack data and intelligence on mobility behaviour even for traditional vehicles, including simple details, such as who is driving the vehicles they insure.

Figure 8: Data access challenges for (re)insurers

Note: This figure indicates the potential variety of data-sets that could be accessed depending on the scope of the partnership model and approach, which could differ on a case to case basis. The partnership approach does not guarantee any automatic access to data.

Types of data partnerships can vary from engagement with a single entity such as a tech/start-up company or OEM to multiple entities in an ecosystem (see Figure 8). Partnerships yield different types of data:  

  • Data from niche tech companies or start-ups – These companies include specialised AV solution providers as well as high-definition mapping and localisation data.
  • Data from OEMs, full stack tech firms or fleet owners – This could provide a fuller picture on how vehicles behave. Data will be accumulated in stages, as AVs progress in a phased manner from L2 to L3 and L4 (L5 in development).
  • Data from mobility platform providers – Data from companies such as Uber and Lyft will inform about customer behaviour or trip information that may further improve risk assessment (for example, identify customers who frequently travel through high risk zones).
  • Data from ecosystems: Partnerships with major platform developers, such as Google or Baidu, could provide access to technology and consumer data. This can also provide insights on demand and supply as well as better understanding how the vehicle is being used.

Combining and curating these data sources to produce usable insights is not a trivial exercise. While many data vendors focus on extraction and distribution of data, few concentrate on data curation and refinement. We expect that this will give rise to aggregators focused on integration, curation and data analytics. However, the partnership approach does not guarantee automatic access to all data, as the scope of data transfers might be dependent on regulatory restrictions and the specific nature of the collaboration and partnership outcome.

Value chain implications for (re)insurers

AVs are initially likely to be used in the context of fleet provision and shared mobility rather than in the form of individual ownership. Access to the insurance customer will therefore be mediated by the mobility service provider. Nevertheless, insurers can remain well positioned to assess risk, price insurance and service policy holders. Accident liability will partially shift towards AV manufacturers and fleet owners; but individual vehicle users cannot be absolved from liability under current regulations. Car owners and drivers will therefore need to maintain minimal liability coverage for the foreseeable future.  

Existing motor underwriting models could see changes. The explanatory variables used to assess risk may be substantially different with additional and different types of data, as well as the gradual or abrupt shift in task execution from the driver to the vehicle. Insurers will move from assessing the risk profile of the driver to the risk profile of the autonomous system.

The breadth of AV use cases and business models could impact insurance rating structures in a more nuanced manner and to a greater extent than in the case of traditional vehicles. The use of a vehicle for goods or passenger traffic already has an impact on insurance pricing for traditional vehicles. However, this difference could be even more marked for AVs for the following reasons:

  1. the algorithms might be tweaked based on what is being transported (different settings for people, goods and types of goods);
  2. pricing could depend on the amount of risk the customer is willing to retain, which will probably be higher for goods than people.

The operational flexibility of AV fleet models, as shown in Figure 4 and 6, will also influence the risk profile of individual AVs. Even if the large scale commercialisation of AVs remains some years in the future, as regulatory and scalability issues are tackled, there has been considerable progress in technology development and business model design. Achieving the full disruptive potential of L4/5 AVs remains a journey. However, to be forewarned is to be forearmed. Pioneering (re)insurers will already be making interim assessments and developing the entire value chain including distribution and product propositions, as well as pricing and claims approaches, to cover AVs in a sustainable risk-adjusted manner.


Autonomous vehicles related content:

​Future smart mobility

The future of mobility is autonomous and sustainable.