Since news of GM’s $500M investment in Lyft broke, there has been significant speculation about whether this is an “undeniably clever” strategy or a sophisticated way to pour money down the drain. Almost all of this attention has focused on the two companies’ long term promise to build out a “network of on-demand self-driving vehicles.” I think the real story here isn’t about their ability to create autonomous technology, but rather what the on-demand present can already start teaching GM about building cars for an autonomous world.
To understand why, the first question to ask is whether GM has the technical ability to build fully autonomous vehicles. Not only do they not have the technical capabilities now, but I think they are quite unlikely to ever have it. GM has suffered a series of embarrassing corporate raids on its autonomous vehicle development program. First, after the 2007 DARPA Urban Challenge, Google hired away all the top staff on the most successful teams, including the Carnegie Mellon University (CMU) team that won the event using GM’s platform. More recently, Uber hired away most of the key members of CMU’s National Robotics Engineering Center (NREC) facility, which was crucial in developing autonomous vehicle technology for GM. Not only has GM’s key research hub at CMU been eviscerated, the company also lacks the Deep Learning expertise—the software technique critical for powering autonomous functionality—and is unlikely to be able to hire away the required talent from the likes of Google, Apple, Facebook and Microsoft, who are all competing over the same limited pool of talent. As much as GM talks a big game about developing autonomous vehicles, it seems more likely that Facebook (which has thus far shown very little interest in transportation) has the required expertise to develop autonomous vehicles than GM. In the long run, it seems probable that Google or Mobileye (or another tech company focused on this problem) will license autonomous software to GM, rather than GM or Lyft developing this technology independently.
GM has suffered a series of embarrassing corporate raids on its autonomous vehicle development program.
The second question is what is the value that Lyft brings to the table? Uber received a rude awakening in early 2015 when news broke that Google, one of its largest investors, was working on a ride-hailing network of its own. Google, once ready with its autonomous vehicles, seems to be quite happy to bypass Uber and build the digital infrastructure for customers to summon a significantly cheaper ride. Uber’s aforementioned move on NREC is all about defending itself against this threat. Thus, it seems that Lyft, Uber’s little cousin, faces the same problem: As a company, it’s pretty useless that your main asset is a large network of drivers if your competitors are well on their way to building driverless cars.
So given these two problems, why invest in Lyft? Daimler’s Chairman, Dieter Zetsche, memorably declared last year at CES that his company does not intend to become the Foxconn of the automotive world—capturing the sentiments of many automotive executives that they want their companies to be more than manufacturers. While tight margins don’t look very good from the C-suite, it is clearly still far better to be Foxconn than bankrupt. And, given the current tsunami of new technologies, business models and market entrants, this is a threat that every car company must now confront.
The core competency of a carmaker like GM is making cars. But the cars in a world of autonomous vehicles are likely to be quite different from the cars GM makes today. Firstly, these vehicles are mostly going to be shared. If you don’t need to drive the car, you don’t need it to be parked, waiting nearby (today’s average vehicle is parked 95 percent of its useful life). Which means you don’t need to own it. Autonomous vehicles will be like Uber or Lyft—always available on demand, but without the driver. Second, in this new business model, these cars are also likely to be electric. Most people are reticent to buy electric vehicles because they worry about the high cost of the battery and the problem of having sufficient range to get to where they need to go. But in a world of shared autonomous vehicles, the high battery cost is shared between many users and the very low per-mile cost of EVs becomes much more important. Furthermore, passengers don’t need to care about range, because it is the fleet operator’s job to ensure that the vehicle is charged enough to complete the trip.
GM can now start building cars for the high utilization, passenger-centric demands of our autonomous future.
So the future is shared, electric, and autonomous. But in a certain sense, the future is already here. All the factors that will shape autonomous vehicle design in the future already apply to ride-sharing vehicles today (in a certain sense, Lyft and Uber are simply platforms to hire people to give cars autonomous functionality). Which means that GM can now start building cars for the high utilization, passenger-centric demands of our autonomous future. And indeed, besides the splashy goal to build a network of autonomous on-demand vehicles, it was also announced that GM is immediately going to start leasing cars to Lyft drivers through what are being dubbed “rental hubs.” It is worth noting that Lyft is in the midst of forming a global alliance with other ride-hailing platforms such as Didi in China, Ola in India and GrabTaxi in Southeast Asia. Though Lyft’s portion of the market might be small in the US compared to Uber, the deal gives GM potential access to a massive, global ride-hailing market, which is searching for ways to cut future costs and offer new customer experiences.
Given this insight, it seems no co-incidence that the Chevy Bolt, the first mass-market electric vehicle, was announced by GM in the same month as the Lyft investment. This vehicle seems to fit perfectly into the plan. Not only will the Bolt be a way for GM to learn how to optimize an electric vehicle for high utilization through drivers on the Lyft network, but the two companies will also be able to collaborate on in-car features tailored to passengers. This begins with easier access to basic controls such as climate control and infotainment systems, but extends into a whole host of potential products and services. In a recent interview, Lyft’s co-founder and President, John Zimmer, suggested that you might soon be able to order a Lyft that’s showing a football game with other fans sharing the ride. Uber has already started offering the ability to play your choice of songs on Spotify when entering a vehicle, but this kind of feature will become much more compelling and reliable when the vehicle is designed around it. Building out these kinds of features where vehicle design and functionality are tightly integrated could allow GM to maintain a critical advantage when entering a world of fully autonomous vehicles.
Over time, this learning strategy might not only allow GM to survive as an automotive Foxconn, but also to be able to capture the significantly higher margin value of an ecosystem built on top of its vehicle platform.