Limitations and Benefits of Big Tech Business Models in Automotive

Big Tech is influencing many areas of industry and society today. There is no doubt about it. They include companies such as Apple, Microsoft, Amazon, and Alphabet (Google). It’s therefore quite appropriate for AutoTech Detroit to discuss ‘What to Make of Big Tech’s Automotive Investment’ this coming June. This discussion is part of the Software-defined Vehicle – Premium Conference Track and it will examine how big data will impact on the automotive cockpit, as well as where the line is drawn between protecting privacy and accommodating intrusive platforms. There will also be a conversation about how architectural changes to the vehicle will inform how automakers design and deliver new features to the customer over time – particularly given the influence of Big Tech on automakers.

Lee Street, director and head of technology services at AECOM Transportation agrees that the influence of Big Tech is huge: “As vehicles move towards to higher levels of autonomy and connectivity, automakers also want to provide a better experience for drivers. Vehicle manufacturers have historically produced software primarily for vehicle management systems, but there is now more pressure and demand to produce more complex software solutions that support automated functions, connected vehicle services, and infotainment.

“Although automakers have their own continually developing capability to provide the technical solutions, they will partner with Big Tech as they can benefit from integration with wider ecosystems and the efficiencies that Big Tech brings. The likes of Google, Microsoft, Apple, and many smaller tech companies are seeing the automotive industry as a huge opportunity for them to partner with automotive manufacturers to expand their reach and services.”

Infotainment and connectivity

Martin Kellner, associate partner and part of the McKinsey Center for Future Mobility, says Big Tech is particularly strong in areas such as infotainment and connectivity. Google, for example, offers Google Automotive Services, an infotainment system that includes Google Maps. He says the Big Tech are often content partners for maps and location services, providing consumer data for automotive solutions. He considers Google Maps to be strong at suppling accurate estimated times of arrival because Google has the data on traffic flow. Then there are the cloud players such as Amazon Web Services and Microsoft who are building the backbone to big data applications.

He explains why these are important to the development of the automotive industry: “Autonomous driving needs these backbones because you need to collect data in the real world, and you need it to test the real-life data against the simulated data. If you develop a new software version you trigger thousands of miles of virtual driving, and this is provided by the Big Tech players.”

Varying degrees of success

Nitin Kumar, co-founder of zblocks and an autonomous driving thought leader, nevertheless comments that Apple’s autonomous driving is non-public with many speculations around them. He adds that Amazon, which acquired Rivian, has not done well and its last earning reflects the losses accrued with this sector. However, it is possible for Big Tech to find success in the sector. For example, he says that Google with Waymo and Tesla have been successful in the US and Baidu has been in Asia. He adds that they all have slightly different business models of varying success.

Autonomous driving stacks

Kumar adds that the autonomous driving stacks are front of center of the business models, while also explaining that the technology stacks spans five layers: the sensor layer, the network layer, the infrastructure layer, the interface layer comprising of hardware and software, and the application layer.

He says the greater the number of sensors, the more likely it will reduce “the miles capability” and he points out that lidar is more expensive and not so appealing to the extent that many incumbents are gravitating towards a fleet business model. Camera-based cars offer an alternative approach because, he claims, they offer more degrees of freedom with design, costs and longer range capability – making them more fitting for consumers.

Kumar adds: “The sensor package, within an autonomous vehicle, is the foundational stack element to see, sense and understand the surrounding context. The greater number of sensors capture the more environmental data and increase component level redundancy.  However, they also add more hardware, fusion, interfaces and compute costs. Fewer sensors may decrease costs and can potentially even boost driving range given less consumption of power attributed to sensor operation.”

For example, Tesla has opted for the cheaper camera approach. While Waymo is devoted to LiDAR. Within the industry, there is much debate about whether Tesla’s camera only strategy will be sufficient.

Nevertheless, Kumar says sensor type and numbers create or constrain degrees of freedom within business models: “Tesla uses the autonomous driving technology stack with cheaper vision cameras for higher data throughput to enable Autopilot, while other companies use LiDAR technology, adding costs and potentially creating reliability issues. The upshot of this higher cost makes expensive autonomous vehicles only suitable for fleet business models. However, the lower cost of sensors gives Tesla the option of selling the cars to consumers as personal vehicles, while keeping the option to also provide them to fleet providers such as Uber and Lyft.” [And yet the list of deaths attributed to Autopilot malfunction keeps growing – Ed].

Centralising compute

Kellner finds that the big architectural change is about the move from a decentralized compute, several control units with smart cameras, to a more a centralized approach involving 2-3 main computers with several smaller computers. “The big benefit is if you want to update the algorithms, you just have to update the main computers rather than thousands of smaller computers with the cars,” he explains while underlining that there are often 100 computers within the cars, and so a centralized approach would reduce the number of computers that will need updating.

He elaborates: “If you have a centralized compute, you have a high-performance computer, and compared to several smaller ones, you have more powerful chips allowing you to add more complex features such as ADAS 2 + Level 3. The centralized architecture and big data will change the experience of customers in the future.”

Replicating subscription models 

Beyond this, automakers want to replicate subscription models, just as you see in the cellphone market. He says this includes subscriptions to allow over-the-air software updates, or to enable connected and autonomous vehicle users to gain access to new services. Center page is the infotainment system because Big Tech is often driven by advertising, and this will often include location-based services to enable someone to find, for example, a nearby restaurant.

He explains: “If you talk about subscriptions, paid per month or per year, automakers are going in this direction for additional services – including ADAS, or for additional horsepower on the engine or heated seating. For this you need dedicated hardware. On the smartphone it’s just software but, in cars, it’s a matter of needing the hardware in the car too. This is one of the limitations of the Big Tech business models.”

Conflicting data                      

Street points out one of the limitations of Big Tech business models. Invaluable data and information can flow directly to connected and autonomous vehicles as well as from them.  The trouble is that the information may be different to what the transport network operator is providing and receiving to and from vehicles. “This may result in a disorganized network whereby driver behavior is less predictable and therefore leads to an ineffective and a less safe network,” he claims.

He adds, however, that Big Tech companies can get things to market quickly and integrate services into their wider ecosystems.  Despite this, the public sector in contrast faces several hurdles, including procurement issues and there has been a huge change in how services are delivered. So, public sector organizations often need to provide their data to Big Tech companies and to vehicle manufacturers “to see what they can do with it,” he says.

However, he stresses that this requires a level of governance so that information and data to drivers is both verified and trusted. “Conversely, network operators can provide trusted services and work with third party data providers to improve the operation of the network, making it safer and more efficient,” he comments.

Big Data’s impact

With regards to the role of Big Data and its impact the cockpit of the future, near-term ADAS, long-term autonomy and the retail experience, Kumar points out that it is not the all the same in the automotive and autonomous driving ecosystem. He says there are several kinds of data with each one driving a different use case and monetization models. “There is fast data e.g., real-time sensor data used for safety and slow data which is more static and used for reporting,” he says.

Street says there is much more data available today than there used to be a decade ago. Once the data has been verified, he says it can be used to provide effective connected vehicles services, and to support advanced driver assistance systems (ADAS), and higher levels of automated driving. He adds: “The challenge with newer vehicles, is how much data or information is too much for the driver to handle in a safe way. The HMI (human machine interface) will be a big differentiator between vehicles, as will trust related to the data and how this may be used for information or control of the vehicle.”

Protecting privacy

As for data privacy, Kumar warns that open-source systems require a privacy layer to make sure users are protected. Proprietary systems such as Tesla’s and Apple’s must also be regulated through data protection legislation and regulations such as GDPR. Kellner concludes by suggesting that automakers should only use data such as vehicle license plates and faces whenever they really need to do so.

He says they should otherwise avoid using personalized data for the development of ADAS and autonomous driving. To inspire trust, they also need to encourage data usage transparency to show people why the data is needed, how it’s going to be used and how it benefits the consumer. In many cases Big Tech will be needed to help with this, for Big Data analysis and for other purposes.