Last week, The Wall Street Journal reported that Ford and Alphabet Inc. (nee Google) are in talks to partner on the production of self-driving cars. Ford CEO Mark Shields is said to be concerned that his company must not become the automotive version of Nokia, a hardware giant that failed to adapt as smartphones and software became essential to mobile consumers.

Shields’ ambition is spot on. Hardware manufacturers increasingly compete in a world of connected devices. If they fail to cultivate a software-driven ecosystem, eventually they’ll find it impossible to differentiate. Without their own software and services, they cannot own the customer experience and all the data it produces. That leaves only one path to survival, and it’s a tough one: a race to the bottom on pricing coupled with as much scale as possible.

Samsung, HP and Dell provide perfect examples of this. Their hardware serves as dumb chassis for the likes of Google (Android) and Microsoft (Windows).

Having worked at Nokia for seven years, including all the way through its acquisition at the hands of Microsoft, I have seen first-hand the conditions hardware manufacturers need to grow healthy ecosystems. If Ford is to survive as a meaningful concern in a world of self-driving vehicles, a world I do not believe to be inevitable, by the way, it must offer:

1. Millions of devices (cars in this case) that share a stable and unfragmented operating system (OS).

2. Hardware and OS access for third party developers.

3. Tools and training for those developers to build compelling third party applications on top of the OS.

4. A marketplace where developers can distribute and monetize their apps.

5. Consumer marketing to drive demand for these third party apps.

6. Essential services built on top of the OS.

7. Ownership of consumer data produced by the use of these services.

Alphabet/Google meets all these conditions. The challenge, as Samsung, LG, and every other Android OEM has discovered, is that Google conditions the use of its Android OS on also accepting Google services: Gmail, Google Maps, YouTube,  Google search, and the Google Play store. That way, Google benefits no matter who sells its Android phones and no matter what consumers do with those phones. That’s because Google monetizes via advertising. These core services are core to Google’s advertising platform, and since nearly every Android phone comes with them, they also make it really difficult for phone OEMs to differentiate themselves.

OEMs can run a version of Android without these five services. With the exception of a few Chinese handset manufacturers, though, OEMs that have attempted this, e.g. Amazon Fire tablets and Nokia X phones, all have failed.

The reason they have failed should worry Ford. Consumers consider each of these five services to be best-in-breed, so when OEMs have tried replacing them with alternatives, consumers have taken a pass.

Could Ford deliver better maps or a superior search engine, much less developer-friendly software on which to run them? Probably not. That means Ford probably will depend on Google’s services, which means its differentiation will depend on developing a core suite of services that drivers and passengers cannot do without and that Google doesn’t offer.

One intriguing possibility, at least to me, that would fit this criteria: an augmented reality system that projects onto the windshield, enabling the car to indicate the presence of gas stations, lodging and restaurants in the distance. A system such as this would obviate the need for many mapping use cases and keep a driver’s eyes on the road.

A self-driving vehicle is not inevitable. Auto manufacturers, insurance companies, dealers, governments, and consumers all need to agree on liability and who will pay in the event of the damage that will occur when a driver-less car crashes. This is a huge dilemma that must be solved, one for which there is no straightforward or obvious solution.

If we can get past the issues, though, Ford, GM, Toyota, VW, and others would need to figure out a meaningful, profitable role for themselves to play. Partnering with the likes of Alphabet/Google, Microsoft or Apple makes a lot of sense on the surface, but only if they find a way to avoid commoditization.