Google Tells Android Developers to Cut Their Memory Use, or Else

Google will start policing memory-hungry Android apps, a direct response to the industry-wide shortage of RAM.

The company published a memo setting out the Play Store's role in enforcing new memory-usage restrictions, emphasising the importance of meeting the limits "to help developers navigate industry-wide hardware constraints and Android's broader memory limits." Failing to meet them will carry consequences.

Where the limits came from

The restrictions were introduced with Android 17, appearing first on Pixel phones. It is probably not a coincidence that the base configurations of the Pixel 11 Pro and 11 Pro XL ship with less RAM than last year's models.

That detail is the whole story compressed into a spec sheet. Memory prices have risen sharply as AI data centre construction consumes global supply, and a component that phone makers previously added freely to differentiate a flagship has become a meaningful line in the bill of materials.

Per-app memory limits arrived first on Pixel devices with Android 17 and will extend across manufacturers over the coming year.

What developers are being told

As Android 17 rolls out across the wider ecosystem, Google is warning developers that they will need to work across a much broader range of devices.

"Over the coming year, an increasing number of manufacturers will leverage the Android per-app memory limits across their portfolio of device RAM configurations from 4GB to 16GB+ devices," the company said.

That range is the difficult part. An app that behaves acceptably on a 16GB flagship may be unusable under a hard cap on a 4GB device, and developers now have to target both from a single build.

Why enforcement through the store matters

Google has previously issued memory guidance and largely been ignored, because guidance without a penalty is a suggestion.

Routing enforcement through the Play Store changes the incentive completely. Distribution is the one thing Google controls absolutely on Android, and an app that cannot be listed or updated has no users regardless of how well it performs.

This is the same lever Google has used to force adoption of target API levels, 64-bit builds and privacy disclosures. It works, and it is deeply unpopular each time, because it converts an engineering recommendation into a condition of continued existence.

The wider pattern

What makes this notable is the direction of causation. Software has spent two decades expanding to fill whatever hardware shipped, on the assumption that memory would keep getting cheaper.

That assumption has broken, and the constraint is arriving from an unrelated industry: the same components that make a phone responsive are being bought in enormous volume for AI infrastructure.

The result is a rare instance of a platform telling its developers that the resource budget is shrinking rather than growing. For a generation of engineers who have never had to optimise for memory, that will be an unfamiliar exercise — and the Play Store has been designated as the thing that makes them learn it.

Who absorbs the cost

The burden falls unevenly, and not on the companies best able to carry it.

Large developers maintain dedicated performance teams and test across device matrices as a matter of routine. A small studio shipping a single app supports whatever hardware it can afford to buy, and now has to certify behaviour across a four-fold range of memory configurations it cannot test on.

The likely outcome is consolidation at the margins: fewer independent apps, and more that quietly drop support for the cheapest devices rather than engineer for them.

That is the opposite of what a memory constraint driven by affordability should produce. The phones with 4GB of RAM are the phones bought by people with the least money, in the markets where Android's share is highest — and they are the devices most likely to be abandoned when meeting the limit becomes the developer's problem rather than the manufacturer's.