NEW PAPER: How Android Auto reshapes the law of refusal to deal (and what it means in practice)
I have uploaded on ssrn a new paper (see here) a paper that discusses the judgment of the Court of Justice in Android Auto as well as its substantive and institutional implications. I am pleased that the paper received yesterday the AdC Competition Policy Award, organised every year by the Portuguese Competition Authority.
The way in which Android Auto has changed the law of refusal to deal, it seems to me, may not have been fully appreciated. To make this point, it is sufficient to apply the Court’s reasoning in Android Auto to the facts at stake in Bronner. The latter would have been decided differently; evidence of indispensability would not have been required to establish an abuse.
As I explain in the paper, this difference is a function of the way in which Android Auto (re)interprets the indispensability condition. In Bronner and Magill, whether or not the dominant firm had kept the assets ‘for the needs of its own business‘ was assessed by reference to the relevant market concerned by the refusal. It was therefore irrelevant that the TV channels in Magill were licensing their copyright to newspapers (but not weekly magazines) or that Mediaprint was printing and distributing another publication in Bronner at the time of the facts.
The analytical approach followed in Android Auto construes the indispensability condition differently. According to the new doctrine, where the dominant firm is dealing with a third party in market A, it can no longer invoke indispensability in relation to a refusal concerning market B. For the same reason, the judgment has significantly reduced the scope of the refusal to deal case law.
This transformation has obvious implications for digital markets. Dominant players in these markets often run systems that are partially open and partially closed. This said, the criteria followed by the Court suggest that the ruling is likely to have an impact in other industries and markets.
In any event, the implications of the judgment go beyond the shrinking of the refusal to deal doctrines. Android Auto allows for intervention that is not confined to a mere duty to deal. A dominant firm may indeed be required not just to share an input or infrastructure with third parties, but to redesign its assets by taking into account the demands of the said third parties. In this sense, the operation of the infrastructure becomes a cooperative venture.
I look forward to your comments on the paper. And thanks again to the AdC!
Chillin'Competition 
Thank you Pablo for analysing the case. I am an in-house lawyer at Google and I’ve been involved with the Android Auto case since the very beginning.
Many of us have apps – navigation app being a typical example – running on our smartphone screen whilst we drive a car. We keep our phone in a car phone holder. This is the way most people used – and still use – Enel’s app for electric car (EV) charging. In 2018 Enel wanted us to develop a template so that people could “cast” their Enel app from their smartphone screen to their car’s in-vehicle screen through the Android Auto app. It sounds simple but it’s not. At the time we had developed Android Auto templates for media and messaging only. These are apps that pretty much every car driver can use.
Android Auto could not technically support the features that Enel’s app required. And when we develop a template, we must carefully test it to reduce driver distraction (user experience, number of features visible in any given time, shape and colour of symbols, brightness, audio and so on). And at the time of Enel’s request, only 0.04% of cars in Italy were electric and could even theoretically use Enel’s EV app. The team wanted to use the limited resources to develop templates for app categories, such as navigation, that every car driver could benefit from, and not only 0.04%.
Following the authority’s decision, the team had to put everything else aside and develop a template to accommodate the features that Enel’s CEO listed in a letter. The team had to redesign Android Auto’s architecture and run extensive user tests. It took almost a year and was closely monitored by a monitoring trustee. Needless to say, the team developing a template for 0.04% of car drivers (many EV drivers in Italy have an iPhone and don’t use Android Auto so the theoretical max demand was closer to 0.02%), could not at the same time develop a template for an app category that every car driver uses. How’s that for a good allocation of resources? Or a platform owner’s right to decide how it develops its platform to make it attractive to as many users as possible?
Finally, say that you Pablo develop an ornithology app. Whenever a user of your app spots a rare bird, she can pin its location on a map. And users of the app get a notice and navigation directions to that spot. You send me an email and say that Google must develop a template for your app so that its 200 users (I’m sure your app would be more popular 🙂 can use it through Android Auto on their in-vehicle screens and not only on their smartphone screens.
How should I respond to you?
Or if a Google product team comes to me and says that they have this cool new thing they want to build. It will let people do amazing and innovative things with their apps. Should I advise them to restrict it to Google’s own apps only? And tell them that letting 3rd party apps benefit from it risks the team losing control over their resources and design of the platform?
Tero Louko
24 October 2025 at 5:52 pm
Are you serious comparing an app that will probably help in the development of electric vehicles to an ornithology app? Is that your best argument? Clearly, reading you comment, it is clear that EV drivers and manufacturers should not count on Google to help, just in case Trump gets mad. The money to develop a template is apparently better invested in Trump’s new ballroom.
EV driver
2 November 2025 at 11:26 am
I have nothing against EVs, I drive one myself 🙂
At the time of Enel’s request in 2018, EVs accounted for 0.04% of all cars in Italy. And this includes all EV cars, including those whose owners have an iPhone which does not work on Android Auto. We had to develop a template to accommodate 0.04% of the potential Android Auto users. We could have used those resources to develop a template for, as an example, navigation apps that all car drivers can benefit from. Wouldn’t that have been better use of our resources?
The goal of Android Auto is to reduce driver distraction. It’s probably better to reduce driver distraction for millions of drivers using a navigation app than for at the time thousands of EV drivers in Italy.
The question is whether a company can decide how it uses its own resources or whether that decision is in the hands of third parties. All who have their own narrow interest in mind, not the overall interest of a platform.
Tero Louko
6 November 2025 at 12:59 pm
the Court said there was no evidence in the file of other “competing” requests (paragraph 65). So the argument that Google was just giving priority to apps used by a larger number of people does not fly.
Another EV driver
23 November 2025 at 9:00 am
GSMNeo FRP makes Android FRP bypass fast and reliable. It is built for technicians who require a safe and effective way to remove Google locks and restore access to FRP-restricted phones.
saidianumia
15 December 2025 at 12:27 pm