GM’s Self-Driving Car Strategy: Vertical Integration

Using maximum control to mitigate risk, GM can generate trillions in revenue by creating an autonomous ride-hailing ecosystem.

March 1, 2018

Gerald Conover

General Motors is vertically integrating the three layers of self-driving — vehicle hardware, the self-driving software, and the ride-hailing service. It is not using fleet learning, the crowd-sourced collection of neural network training data from its fleet of production cars.

Vertical integration makes sense for GM for two reasons. The first reason is technical: it allows for a better integration between the vehicle hardware and the self-driving software. The second reason is financial. Controlling all three layers means that a company will capture 100% of the revenue — and, hopefully, profit — from an autonomous ride-hailing service.

GM took a hardware step through the acquisition of lidar-startup Strobe. The company also has a long-term relationship with Israeli camera-maker Mobileye. More importantly for GM, it is doing its self-driving car assembly in a real production plant — Orion Assembly in Michigan. The foibles coming from installing all the electronic architecture, sensors, etc., will hopefully come to the fore early in the development process.

The software piece came through GM’s acquisition of, and huge investment in, Cruise Automation (GM Cruise LLC). GM hasn’t said how much it has plowed into Cruise, but analyst estimates run from $500 million to $1 billion. This well-funded startup appears to be coordinating most of GM’s self-driving car program development.

GM appears to be focused mostly on the vehicle integration and assembly parts of the equation.

The ride-hailing picture has several frames. GM launched its homegrown ride-sharing Maven brand and has also thrown in with ride-hailer Lyft (with an investment of $500 million). Both moves are part of a comprehensive business which is quickly expanding outside the U.S. In any event, GM’s promises suggest it will be on the street with ride-hailing — maybe autonomous, maybe not — by late 2019.

The key to all this is to get the self-driving technology to work — and work faultlessly. Currently, each trip with a ride-hailer loses money for the operating company. But remove the human from the driver’s seat and the picture changes — one can finally make money at it.

But, is self-driving bullet-proof ? Lyft executives, at the 2018 CES, said they were very unlikely to take the driver out of their branded cars “for the foreseeable future.” In other words, Lyft is not sufficiently confident in the self-driving technology to risk its reputation or its bankroll to a lawsuit in case something goes wrong.

Even with reliability equal to that of U.S. spacecraft (0.9999998%), math whizzes tell us there would still be 300 fatal crashes per year — a big improvement from 39,000 in 2017. The probability of one of them being in a ride-hailing car is extremely low, but the outcome for the ride-hailing operator is too great to risk.

It then becomes clear why GM is taking on its vertical integration strategy — maximum control, to mitigate risk, with maximum access to profit, to sustain a business model.

By the way, all of those self-driving cars, to GM’s thinking, will be electric.

This furthers “The General’s” need to be environmentally responsible and meet myriad clean-air rules in many different markets. CEO Mary Barra is pretty cagey for an engineer.

Gerald Conover is a globally recognized expert in intelligent transport systems, automotive innovation and business modeling. He is a recipient of the SAE Delco Electronics Intelligent Transportation Systems Award. Conover is editor-in-chief of the Car-Smart Automotive News Weekly.