Introduction
“Fragmentation” has been the criticism that defined Android since 2009. Unlike iOS, where Apple controls the hardware, software, and updates, Android runs on thousands of devices from hundreds of manufacturers, each with different screen sizes, processors, RAM configurations, and — critically — different OS versions.
The result: at any given moment, a significant portion of Android devices are running versions that are years out of date. Developers can’t use the latest APIs. Users don’t get security patches. And the experience of “Android” varies wildly depending on what phone you bought and when.
But the fragmentation narrative has changed. Google’s efforts — Project Treble, Project Mainline, Google Play Services modularization, and manufacturer pressure — have measurably improved the situation. The fragmentation of 2025 is not the fragmentation of 2015.
This is the story of Android’s fragmentation — the data, the causes, the attempted fixes, and where things stand today.
What Fragmentation Means
Android fragmentation has several dimensions:
- OS version fragmentation: Different devices running different Android versions, from the latest release to versions 5+ years old
- Screen size and density fragmentation: Thousands of different screen configurations, requiring responsive layouts and multiple asset densities
- Hardware fragmentation: Different processors, GPUs, RAM amounts, storage, sensors, and capabilities
- Manufacturer skin fragmentation: Samsung One UI, Xiaomi MIUI, OnePlus OxygenOS, and others modify Android’s behavior, appearance, and APIs
- Update fragmentation: Some devices get monthly security patches; others never receive a single update after purchase
The most visible — and most criticized — dimension is OS version fragmentation.
The Data: OS Version Distribution Over Time
Google publishes active-device OS distribution data periodically. Here’s a snapshot at key moments:
2011 (Android 2.3 Gingerbread peak)
| Version | Share |
|---|---|
| 2.3 Gingerbread | 55% |
| 2.2 Froyo | 30% |
| 2.1 Eclair | 10% |
| 1.6 Donut | 3% |
| 1.5 Cupcake | 2% |
The latest version at the time (2.3) held just over half the install base. A full 15% of devices were running versions released in 2009 or earlier.
2013 (Android 4.1-4.3 Jelly Bean peak)
| Version | Share |
|---|---|
| 4.1-4.3 Jelly Bean | 59% |
| 2.3 Gingerbread | 26% |
| 4.0 ICS | 12% |
| 4.4 KitKat | 3% |
Gingerbread — a version released in 2010 — still powered over a quarter of Android devices three years later. Developers targeting modern APIs had to support Gingerbread or lose 26% of their addressable market.
2015 (Android 5.0-5.1 Lollipop era)
| Version | Share |
|---|---|
| 4.4 KitKat | 39% |
| 5.0-5.1 Lollipop | 21% |
| 4.1-4.3 Jelly Bean | 30% |
| 4.0 ICS | 6% |
| 2.3 Gingerbread | 4% |
Lollipop — the latest version — had only 21% adoption a year after launch. KitKat, two versions behind, was the most-used version at 39%.
2019 (Android 9 Pie era, post-Treble)
| Version | Share |
|---|---|
| 8.0-8.1 Oreo | 28% |
| 9 Pie | 22% |
| 7.0-7.1 Nougat | 19% |
| 6.0 Marshmallow | 17% |
| 5.0-5.1 Lollipop | 9% |
| 4.4 KitKat | 5% |
The curve was flattening. The latest version (Pie) was growing faster than Lollipop ever did, and the long tail of ancient versions was shrinking.
2023 (Android 13 era)
| Version | Share |
|---|---|
| 13 | 20% |
| 12 | 25% |
| 11 | 22% |
| 10 | 18% |
| 9 Pie | 8% |
| 8.x Oreo | 4% |
| 7.x and older | 3% |
The distribution is much flatter and more compressed. No single version dominates the way Gingerbread or KitKat did in their eras. The long tail of pre-8.0 versions has nearly disappeared.
Why Fragmentation Happened
1. The Chipset Vendor Bottleneck
Before Project Treble (2017), every Android update required Qualcomm, MediaTek, and other chipset vendors to update their driver code for each new framework release. Vendors had no financial incentive to support old chipsets. Once a chipset was 2–3 years old, driver updates stopped. No drivers meant no Android updates.
This was the single biggest cause of fragmentation. It wasn’t laziness — it was a structural problem in Android’s architecture.
2. Manufacturer and Carrier Incentives
Phone manufacturers make money when you buy a new phone, not when they update your old one. Carriers control the update pipeline in many markets and have zero incentive to approve OTAs quickly — updates can introduce bugs that increase support calls.
The result: flagship phones might get one or two updates, if the manufacturer felt generous. Budget phones got nothing.
3. Low-End Device Economics
Phones sold for $50–100 in emerging markets had razor-thin margins. Manufacturers used the cheapest chipsets available — often years-old designs with no ongoing driver support. These devices shipped with whatever Android version was available when the chipset was manufactured, not the latest version.
The Android One and Android Go programs were designed to address this, with limited success in the early years and better results recently.
4. The Custom Skin Problem
Samsung, Xiaomi, OnePlus, and others overlay Android with their own skins. Every Android release requires the manufacturer to port their skin to the new version, test it, and get carrier approval. This process takes months, even when the underlying Android code is ready.
Samsung’s One UI 6 (Android 14) started rolling out in October 2023 and took until mid-2024 to reach all eligible devices. That’s 8–9 months from Android 14’s release to full Samsung deployment.
What Google Did About It
Project Treble (2017)
The single most important anti-fragmentation measure. Treble separated the Android framework from vendor driver code with a stable Vendor Interface (VNDK). Framework updates no longer required vendor driver changes. This removed the chipset bottleneck.
Post-Treble, manufacturers could update the Android framework independently of chipset vendor support. The result: update timelines shortened from 12+ months to 3–6 months for most flagships.
Project Mainline (2019)
Mainline modularized specific Android system components — media codecs, networking, timezone data, Wi-Fi — and delivered updates through Google Play, bypassing manufacturers entirely. These modules are updated silently, like apps, with no OTA required.
Mainline covers 20+ system modules and closes the gap between full OS updates. A device might be on Android 12 but have Android 14’s media framework, networking stack, and security patches.
Google Play Services
Play Services is Google’s backdoor update mechanism. It’s a system-level app that Google updates directly, adding new APIs and features without an OS update. Many features branded as “Android 14 features” actually ship through Play Services and work on Android 10+.
This has been controversial — Play Services is closed-source and gives Google significant control — but it’s been effective at reducing API-level fragmentation.
Manufacturer Pressure
Google has used its leverage to push manufacturers toward longer update commitments:
- 2017: Google required 2 years of security updates for Google Play certification
- 2020: Google pushed Samsung and OnePlus to commit to 3 years of OS updates
- 2022: Samsung committed to 4 years of OS updates + 5 years of security patches
- 2023: Google committed to 7 years for Pixel 8
The competitive pressure from Google’s own commitments has forced other manufacturers to follow suit. Samsung’s 4+5 commitment is now the Android industry standard for flagships.
The Current State
Android fragmentation in 2025 is significantly improved from 2015:
- The pre-Oreo long tail has collapsed: Devices running Android 7.x or older represent under 3% of active devices, down from 30%+ a decade ago.
- Version distribution is flatter: No single version dominates. The spread from Android 10 to 14 is relatively even, meaning developers can target API 29+ and reach ~95% of devices.
- Security patch adoption is better: Monthly security patches reach flagship devices within 30–90 days of release. Google Play System Updates fill gaps between full OTAs.
- The budget segment is healthier: Android Go devices ship with recent Android versions and receive updates through Google’s Go program. A $50 phone in 2025 runs Android 14 Go, not KitKat.
But fragmentation isn’t solved:
- Still slower than iOS: In the weeks after a new iOS release, 50–70% of iPhones adopt it. Android 14 took over a year to reach 20%. The gap is inherent to the open ecosystem model.
- Budget devices still lag: Sub-$100 phones from obscure manufacturers often ship with Android 12 or 13, not 14. They receive few or no updates.
- Carrier bottleneck persists in the US: Verizon, AT&T, and T-Mobile still control OTA approval for branded devices, adding weeks or months to update timelines.
The Developer Perspective
Fragmentation dictates Android development strategy:
minSdkVersiondecisions: Most apps in 2025 target API 26 (Oreo) or higher, reaching ~97% of active devices. Targeting API 21 (Lollipop) reaches ~99%. The trade-off is development complexity.- Jetpack and AndroidX: Google’s Jetpack libraries abstract away version differences. Developers use
AppCompatActivity,Fragment,ViewModel, andWorkManagerwithout worrying about which API level they were introduced in. - Support Library / AndroidX: Backports of newer APIs to older versions, reducing the need to compromise on features.
- Testing fragmentation: Device farms (Firebase Test Lab, BrowserStack, AWS Device Farm) let developers test on hundreds of device configurations without owning the hardware.
The developer experience has improved dramatically. The days of supporting Gingerbread while targeting Jelly Bean — maintaining separate code paths for every API level change — are over. Jetpack and AndroidX handle the fragmentation, and developers write against a consistent API surface.
The Future
Android fragmentation will never be eliminated. The open ecosystem model — thousands of devices, hundreds of manufacturers, dozens of carriers — is fundamentally different from Apple’s vertically integrated approach. Fragmentation is a feature of the model, not a bug to be fixed.
But the trend is positive. Each Android release narrows the gap between the latest version and the install base. Each manufacturer commitment extends the update window. Each Mainline module reduces the importance of full OS updates.
The Android fragmentation of 2030 will be better than the fragmentation of 2025. It will never be as clean as iOS. It was never meant to be.
References
- Android Distribution Dashboard (developer.android.com, 2010–2025)
- Project Treble technical documentation (source.android.com)
- Project Mainline / Google Play System Updates documentation
- Samsung, OnePlus, Google update commitment announcements
- Google Play Services version history and API distribution data
- Jetpack / AndroidX documentation (developer.android.com)