What's on my Phone 2026: the home screen has become an AI dashboard

01The annual ritual, updated for the agent era

Every year Marques Brownlee — the reviewer better known as MKBHD — publishes “What's on my Phone,” and every year it doubles as a snapshot of what a power user's smartphone has become. The 2026 edition is different in kind, not just degree. Where past videos cataloged launchers, icon packs, and camera apps, the 2026 home screen is organized around AI assistants and agents: chat apps that plan, summarizers that read, photo tools that edit by instruction, and agents that chain tasks across services. The phone has quietly become the primary interface to the model market.

The ritual matters beyond one reviewer's preferences. Wikipedia describes the smartphone as the dominant computing device of the era — iOS “a mobile operating system created and developed by Apple,” Android “an operating system developed by Google... based on a modified version of the Linux kernel.” When the most-watched reviewer's daily driver is reorganized around AI, that is field evidence of a platform shift, not a keynote promise.

02What changed since the last edition

Three shifts separate the 2026 home screen from its 2025 predecessor. First, AI assistants consolidated: instead of a dozen single-purpose tools, a few multimodal assistants — Gemini, ChatGPT, Claude — absorb the tasks that used to belong to separate apps. Second, agents arrived on mobile: apps now execute multi-step work, booking, researching, and filing, rather than only answering. Third, on-device features went quiet: photo cleanup, call screening, and notification summaries happen locally on the NPU without anyone calling them AI.

The consolidation is the interesting part. The long-predicted “super-app” arrived not as one app but as one model embedded in many. A reviewer's home screen in 2024 was a portfolio of specialized utilities; in 2026 it is a portfolio of model front-ends plus the utilities that models have not yet absorbed.

03The categories that survived

Not everything was absorbed. The video's durable categories are instructive: messaging remains native — no one wants an AI intermediary in a conversation with a friend. Camera and photo apps persist because hardware capture is still app-bound, even as editing migrates to model-driven tools. Media, reading, and podcasts survive on ownership and offline habits. And the utility tier — password managers, banking, transit — persists precisely because their jobs are transactional and trust-bound rather than generative.

The pattern generalizes: AI absorbed the word-shaped tasks first (summarizing, drafting, searching), is absorbing the workflow-shaped tasks now (agents), and has made least progress on the trust-shaped tasks where a wrong output costs money or relationships. That ordering predicts the next two years of app-store evolution better than any vendor roadmap.

04On-device AI and the NPU dividend

Beneath the app layer, the hardware story has matured. Every 2026 flagship — Apple's A-series, Qualcomm's Snapdragon 8 Elite Gen 5, MediaTek's Dimensity 9500 — ships a neural processing unit measured in tens of TOPS, the compute that makes local summarization, transcription, and photo editing free and private. Wikipedia's system-on-a-chip entry captures the trend: modern SoCs integrate “most or all key components of a computer” onto one die, and the NPU is now one of those key components rather than a bullet-point.

The dividend is behavioral. When inference is free and offline, usage stops being rationed. The 2026 home screen's AI-heavy composition is only possible because the marginal cost of a summary or a photo rewrite hit zero — the same economics that drove open-weight models on laptops, mapped onto the phone.

05What the home screen says about the platforms

Platform strategy is visible in the app grid. Google's Android ships with Gemini woven through the system surface; Apple's iOS integrates its model stack into Siri, Photos, and Notifications; both now expose on-device models to third-party apps. The reviewer's choices therefore reflect what the platforms enable: deep-system assistants on one side, polished selective integration on the other. Wikipedia's entries on iOS and Android describe two decade-old platforms converging on the same answer from different starting points — the assistant is becoming the OS.

There is a caveat worth keeping. Home-screen composition is a power-user signal, not a mass-market measurement. Most users' phones still look like 2021's: grids of single-purpose apps. The direction is set by the reviewers and developers; the median follows a few years behind.

06The takeaway: the phone is now the AI delivery vehicle

The 2026 edition of “What's on my Phone” documents a completed transition: the smartphone is no longer just a computer in your pocket — it is the primary delivery vehicle for artificial intelligence, through cloud models, subscription front-ends, and on-device NPUs. The apps that thrived are the ones that either are model interfaces or do things models cannot be trusted with.

For N43 readers the practical reading is direct. When choosing a phone in late 2026, the spec sheet that matters is no longer just screen and battery — it is NPU throughput, on-device model support, and how deeply the platform's assistant is woven into the system. The home screen has already voted; the checkout page is catching up.