I t wasn’t that long ago that GM ran commercials advertising that its Oldsmobile division didn’t just produce cars for your grandfather, but also for everyone else. It was an attempt to reinvent the brand’s staid image—and it didn’t work.

Now, the Oldsmobile division and its iconic vehicles are gone. GM is starting, albeit slowly, to upend other parts of its business model to thrive and grow in the age of digital platforms and virtual networks. The question is this: Are they moving fast enough?

At the beginning of 2016, GM leaders decided that they wanted to be part of the changing business model landscape in the auto industry by innovating what they did and how they did it. In an interview with The New York Times, GM President Dan Ammann said, “We think there’s going to be more change in the world of mobility in the next five years than there has been in the last 50.” The result: The automaker announced that it was going to invest $500 million in Lyft, a peer-to-peer, ride-sharing service for riders and drivers that competes with Uber, as part of a $1 billion venture financing round that valued it at approximately $5.5 billion, post-financing. As part of GM’s investment, Ammann will join Lyft’s board of directors.

What is so interesting about GM’s investment is this revelation from Lyft President John Zimmer. “We strongly believe that autonomous vehicle go-to-market strategy is through a network, not through individual car ownership,” he told The New York Times in the same article.

With Uber’s recent $3.5 billion financing that valued the tech company at a reported $62.5 billion—or about 1.5 times GM’s market cap—it is clear that traditional automakers like GM need to do something dramatic to stay relevant in the coming years. According to the Times, executives at both GM and Lyft will start co-developing an on-demand network of self-driving cars, an area of research that has seen enormous investments from companies such as Google, Tesla, and Uber in recent years. GM and Lyft also plan to develop a series of short-term car rental hubs across the United States that let people who don’t own cars pick up a vehicle and drive for Lyft to earn money—similar to other car-sharing services (known as virtual networks).

But even that may not be enough at a time when Google co-founder Larry Page is already thinking even further ahead. He reportedly is personally funding two flying car companies—Zee.Aero and Kitty Hawk—at a cost that is far less than GM’s initial investment in Lyft. As such, the entire auto industry’s current business and mental models are now at risk.

New business models require new mental models

Will GM’s self-driving, network-based, car-sharing future create value for the company and its shareholders? Will its investment in Lyft make it a digital leader in 10 years? Will its seat at the table in Silicon Valley make a difference in terms of what GM does and how its leaders think and act? It’s too early to say. But, given its initial investment, one might argue that the openness exhibited by its leaders to today’s digital and network realities is proof that they are ready to do what it takes to survive and grow.

So what are the lessons you can learn from Ammann and GM?

Our work with boards and leaders suggest that the biggest challenge to organizations is not new technology, but rather the mental models of leaders. A mental model comprises the assumptions, behaviors, and beliefs of an organization and its leaders within an industry about how value is created. These assumptions drive how the organization spends and makes money. But too often, existing mental models have become outdated, enabling disruptive business models like Uber and Lyft to emerge.

So what can a leader do to ensure the firm’s future success? Here are five steps that will help you change your mental model:
1. Write down your current belief system. (For example, we make, market, and sell cars.)
2. Question your beliefs. (Is making and selling cars the best way to create value?)
3. List opposing beliefs. (Use our knowledge of “cars” and our many relationships to create a digital platform for our customers to share their vehicles.)
4. Test your new beliefs with customers, investors, suppliers, and employees. (Build a test platform for the exchange of sharing cars.)
5. Roll out your new, updated beliefs as a scalable business model. (Think of Apple’s developer community.)

Network effects are scalable and valuable

Perhaps you’re thinking that what’s happening to GM is an isolated event, or that Larry Page’s flying car investments are off the wall. Further, you might think that one or both situations don’t apply to your part of the automotive industry. We disagree.

New platforms, such as the ones that Lyft, Uber, and Larry Page are building, can become economic juggernauts under the right circumstances. As proof, virtual networks can be found in almost every industry today and span social (B2C), business (B2B), and financial (B2F) realms. And the reason they are springing up everywhere is because they are special. In short, virtual networks:
• Provide exponential growth and value creation
• Erect barriers to entry that thwart incumbent industry competitors
• Create “Winner Take All” market outcomes

Indeed, these network effects are like a flywheel—the faster you spin them the more momentum they generate. Further, network effects are not the same as economies of scale where bigger is better. (GM has more than $100 billion of fixed assets, whereas Uber and Lyft have almost no fixed assets—but plenty of riders and drivers.)

To understand the differences between the two in terms of value creation, growth, and profits, note that “economies of scale” organizations achieve competitive advantage and defensibility by getting really big to realize the reduced cost of production, marketing, and sales. For example, GM and Ford both benefit from having massive assets (plant, property and equipment) that enable them to manufacture and sell cars worldwide using armies of employees who design, develop, and assemble their cars for retail sales by dealers. Lyft and Uber, on the other hand, are “network-effect” firms. They act as a virtual orchestra leader connecting riders with drivers of their own cars worldwide. The second scales faster and less expensively than the first.

Metcalfe’s Law captures the difference in these two models—economies of scale and network economics—as a simple equation where the value of a network = N², in which N is the number of nodes. There are three main types of network-effect organizations:
• Physical-only networks, in which the value of the network increases as it expands. The U.S. highway system, local power, and water utilities and waterways used by shippers to move goods domestically and internationally are examples.
• Digital-only networks, in which the value of a product or service increases exponentially with the number of users. Communications networks like instant messaging, texting, email, and video chats are all examples. Companies that are benefiting from digital-only networks include WhatsApp, WeChat, and Skype.
• Hybrid networks or platforms, where network orchestrators aggregate buyers and sellers to facilitate transactions in a marketplace (platform companies). The presence of many sellers on the platform means there is variety, competition, and price pressures in the market—which attract more buyers. And as more buyers use the platform, more sellers are enticed to participate in the marketplace. Uber is today’s best known example of this effect: The more cars, mobile app users, and data that are generated, the better is their traffic algorithm. And the better the algorithm, the more pricing power it enjoys.

Indeed, the flywheel of people, data, interactions, and algorithms is extremely potent in today’s digital world. The more people (riders and drivers of Uber, Didi, and Lyft) participate, the more likely it is that others will join the network, and thus more data will be generated. In turn, the more that each participant interacts (cars, feedback, and recommendations), the more likely it is that someone else will do so, too. This is very different from economies of scale organizations (like GM and Ford), where the cost advantages are obtained from the size and scale of operations (number of employees or factories) as well as the output. Here, the cost per unit of output (be it products or services) generally goes down as costs (buildings, training, education) are distributed across more units of output.

A whole new world of automobile innovation

It is easy to think of Uber as a foregone conclusion now that it is scaling the world. It’s clear that it is fulfilling a need in the market—an opportunity that the auto industry missed due to the outdated mental models of its leaders. The good news is that GM may be able to adapt to the new digital world with Lyft’s help. But it is unclear whether GM has come too late or too early to this automotive reinvention party.

What is clear, however, is that business model innovation can happen in any industry, and networks as well as enabling technologies—flying cars, self-driving automobiles and other still-yet-to-be-imagined digital offerings—are going to reshape the automotive industry.

The bottom line: This is no longer the same industry that brought you the Oldsmobile. That car brand is long gone. However, the automotive industry newcomers in the likes of Uber, Lyft, Google, and Apple aren’t car companies in the traditional sense, either. They don’t think or act like automakers. If the industry that brought us the Oldsmobile wants to be here for much longer, it will need to put the pedal to digital metal.

Knowledge@Wharton contributors Barry Libert, Megan Beck, and Jerry (Yoram) Wind would like to thank LiquidHub for sponsoring the research for this article. Libert is CEO of OpenMatters, and Beck is the chief insights officer. Wind is a Wharton marketing professor and director of Wharton’s SEI Center for Advanced Studies in Management. They also wrote the book, The Network Imperative: How to Survive and Grow in the Age of Digital Business Models (Harvard Business Review Press, 2016).

First published June 17, 2016, on the Knowledge@Wharton blog.