The Real Cost of Google’s Seven-Year Phone – and Why It Wants the Bill to Be High
For most of the past decade the smartphone business ran on a quiet contradiction: sell a customer an £800 computer, then treat it as obsolete within three years. Google is now betting against its own industry. Since the Pixel 8 in 2023, its phones carry seven years of Android OS upgrades, security patches, and Feature Drops – support that runs, as Google’s own launch announcement put it, “all the way into 2030”. The company has promised spare parts for the same seven years, with iFixit as its US partner, so that the hardware can physically last as long as the software is maintained.
For consumers this is easy to applaud. The more interesting story, and the one that should concern anyone who runs a hardware or platform business, is what a seven-year promise does to a company’s cost base. Supporting a seven-year-old handset is not a matter of occasionally posting a security patch. Every material change to Android has to coexist with ageing chipsets, modems, radios, cameras, and drivers without introducing regressions, and that test matrix widens with each supported cohort. Google’s VP of engineering, Seang Chau, has been candid about where the expense concentrates: the system-on-chip, he told the Made by Google podcast, “generates probably the most complexity when we’re talking about software updates”, which is why the company had to build labs equipped for continuous testing before it could make the commitment at all.
Google has attacked this problem architecturally rather than simply throwing engineers at it. The Generic Kernel Image separates hardware-agnostic kernel code from vendor-specific modules so the two can be updated independently, and Project Mainline delivers dozens of operating-system components straight through Google Play, bypassing the slow OEM-and-carrier update chain – the count of updatable modules grew from 12 in Android 10 to 37 in Android 14. That matters because, as Google explained when it launched Mainline in 2019, media components alone accounted for nearly 40 per cent of recently patched vulnerabilities. Abstraction shrinks the long tail of support cost. It does not abolish it: a severe vulnerability in a five-year-old chipset is no less serious because most owners have already upgraded, and the revenue from that cohort was booked long ago.
Here is the counterintuitive part, and the reason executives should read this as strategy rather than corporate social responsibility. The expense is not a flaw in the plan; it is the plan. A durability promise becomes a competitive moat only when rivals find it genuinely hard and costly to match. This is why the commitment is spreading in lockstep. Samsung matched seven generations of OS upgrades and seven years of security updates with the Galaxy S24 in January 2024; Qualcomm announced eight years of software and Android-version support for Snapdragon 8 Elite devices in February 2025, the longest available in the market; and the EU’s Ecodesign Regulation now mandates at least five years of OS updates – timed from the end of a model’s sales – and seven years of spare-parts availability. Longevity has moved from differentiator to baseline. The competitive question is shifting from who has the best camera sensor to who can operate a device as durable infrastructure rather than seasonal fashion, and do so profitably.
For hardware and platform leaders, three moves follow. First, price the tail in: treat multi-year support as a committed liability in the unit economics, not a marketing line, and negotiate matching support windows from chipset and component suppliers up front, as Qualcomm’s eight-year pledge now makes possible. Second, invest in the architecture that lowers the marginal cost of each additional supported year – modular kernels and store-delivered OS components are cheaper than perpetual bespoke testing. Third, decide deliberately what “support” means in years six and seven – full OS upgrades, security-only, or feature-frozen – and say so plainly, because once seven years becomes a brand promise, retreat is reputationally expensive. The firms that win the next phase will not be those that promise longevity loudest, but those that have engineered its cost down far enough to keep the promise without regret.
Sources
Personal analysis. Views are mine, not my employer's.