The Automotive Industry’s Big Data Challenge (Part 1)

In the not too distant future, automakers won’t be evaluated just on the physical, safety and performance characteristics of their vehicles.  Instead incumbent and next-generation automakers will be evaluated based on the completeness of their solution along five dimensions: Electric, Autonomous, Connected, Mobility Services (EAC+MS), and Information.  We read about the progress automakers and their suppliers are making along the first four dimensions.  There is much less conversation about the fifth dimension.  In this two-part series, we will discuss the big data challenge facing the automotive industry.  The pieces are the result of my work in the industry helping corporations with their innovation and big data strategies.  The first post provides the why and makes two points:

1. Automakers must be in the information business.  To be effective in the information business, automakers must change their perspective and start thinking about an overall process for big data in and around the car.
2. Information in EAC+MS implies big data and incumbents in the automotive ecosystem must become serious about big data.  Newcomers to the automotive industry such as Google, Tesla, Faraday Future, and likely Apple, but also Uber, and Lyft, realize this imperative.

The second piece will provide the how to try to address this challenge.

To fully appreciate the importance of the information dimension, one needs to understand that offering “EAC+MS” means that, in addition to manufacturing vehicles with variations of electrification (from fully electric to hybrid-electric, to fuel cell vehicles, etc.), levels of autonomous driving and Internet connectivity, the automaker provides transportation solutions.  Such solutions include the vehicle the consumer owns, mobility services around this vehicle, e.g., parking, roadside assistance, repair, but also broader transportation services, e.g., ride-sharing, car-sharing, car rental.  At the core of these solutions is data and in order to provide such solutions automakers must accept to be in the information business.  They must be able to create information, e.g., creating models that predict engine failure, build systems that utilize big data, e.g., the navigation systems of autonomous vehicles, and utilize third-party data, e.g., weather data from services such as weather.com to better serve the vehicle’s occupants.  

Automakers must change their perspective

To be in the information business incumbent automakers must change their perspective and the way they operate.  Over the past few years, incumbent automakers became systems integrators, and in some instances they have even been acting more like contract manufacturers; think Foxconn or LG both of which are already working with the automotive industry.  For each particular vehicle they would issue to their suppliers a master specification that satisfies perceived market needs and addresses government regulations, e.g., safety, emissions.  They would expect their suppliers to provide the components and subsystems that conformed with this specification.  They would then proceed to assemble these components into a vehicle through a highly optimized and efficient process.  This approach means that much of the innovation and knowledge on how to address these requirements go to the supplier, rather than the automaker. A perfect example of this point is NVIDIA which aspires to become one of the leaders in data processing and machine learning for self-driving cars with Level 4 autonomy.  Similarly the user experience around each such subsystem, e.g., infotainment, is owned by the supplier rather than the incumbent automaker.  For example, think about the control Tesla has on its infotainment system user experience versus any other automaker that incorporates an infotainment system from Bosch or Google.

If automakers are to participate in the EAC+MS business effectively and be evaluated on the transportation solutions they provide, in addition to the E, A, C technologies, they must develop big data expertise.  And they must develop such expertise regardless of the level of autonomy they offer in their cars.  In EAC+MS, innovations that up to now were left completely in the hands of suppliers, dealers, and other parts of the automotive value chain, must now be touched by the automakers, if not owned by them, because they all involve data.  Ford’s recent moves provide an interesting example (and here) of this broadening viewpoint.  

The ramifications of not developing big data expertise are far greater than not developing, for example, fuel pump expertise.  Acquiring data expertise won’t come naturally and won’t come easy because automakers must first understand the difference between customer-centric mobility and car manufacturing, and between EAC vehicles and the cars they’ve been manufacturing for the past 100 years, as well as the disruption these differences have on their business model which is manufacturing-centric.

Automakers must become serious about big data  

Automakers are already familiar with the data created by the systems embedded in their vehicles today.  They are just not proficient with big data, in the way companies like Google, Tesla, Uber, Lyft, and a few other newcomers are.  Over the years, an increasing number of microprocessor-controlled subsystems have been embedded in cars.  These microprocessor-controlled subsystems:

1. Have been provided by various suppliers, e.g., Delphi, Denso, Magna, Bosch, Continental, etc. and have been integrated in the car on an ad hoc basis, i.e., without a well-defined system architecture that can be updated appropriately as more components with microprocessors are added to new car designs.  We are now to the point where certain cars may have more than 100 microprocessors on board.  Interestingly, because of this approach, the Tier 1 and Tier 2 suppliers have better appreciation and knowledge of data than the automakers themselves.
2. Generate data only a small percentage of which is accessible, typically via the ODB, and utilized.  Most frequently this data is used to provide variants of a vehicle’s health report.  It ranges from the familiar “check engine” light seen in low-end vehicles, to the graphical reports provided in the infotainment system of higher-end vehicles.  

Now imagine the time when you are driving an autonomous, connected car.  Such a car will have orders of magnitude more sensors (engine sensors, its electrical system sensors, tires sensors, suspension, steering, radar/lidar, cameras, etc).  All of these sensors will generate data constantly.  The vehicle’s infotainment system (mapping, messaging, entertainment content, etc.) is another big data generator.  It is expected that such a car would be generating in excess of 1GB/sec of regular operation.  To this data one has to add the data generated by the passengers in the course of a trip, the data exchanged between each vehicle and other autonomous vehicles as they try to coordinate with one another to ensure their passengers’ safety, as well as the data exchanged between each autonomous car and the smart infrastructure it relies on, e.g., roads, bridges, toll stations, etc.  Therefore, EAC vehicles generate really big data and this is why the incumbent automotive ecosystem must become proficient with big data in a hurry.  Companies in the ecosystem have been taking some steps to gain data expertise primarily through acquisitions of (here and here), investments in, and partnerships with relevant companies, and by the hiring of data scientists (and here) and starting to set up big data infrastructures.  

There should be no question that with the adoption of EAC+MS incumbent automotive OEMs and parts suppliers must be in the information and big data business, a point which the new entrants to the automotive ecosystem have realized early on.  Developing such a competency will not come easily to the incumbents, as the information business is very different from the manufacturing business they’ve been to date.  In the next piece of this series we will discuss the new processes and value chains automakers will need to establish in the process of developing such a big data competency.

You can find Part 2 here.

© 2016 Evangelos Simoudis