OpenAI might be making a phone, and it could kill the app store as we know it

There’s been chatter about OpenAI’s hardware ambitions for a while now — most of it centered around a pair of earbuds. But a fresh note from reliable industry analyst Ming-Chi Kuo suggests something much bigger: OpenAI might be building a full-blown smartphone.

According to Kuo, OpenAI is partnering with MediaTek and Qualcomm on a custom smartphone chip, with Luxshare handling co-design and manufacturing. That’s a heavyweight lineup, and it signals this isn’t some half-baked experiment.

The really interesting part is what Kuo says about the software side. Instead of a traditional app grid, this phone would rely entirely on AI agents to handle tasks. No App Store, no Google Play, no middleman controlling what the phone can and can’t do. Right now, Apple and Google hold a tight grip over app distribution and system-level access, which limits how deeply AI can integrate into core phone functions. By building its own hardware and software stack from scratch, OpenAI could give its agents unfettered access to everything — notifications, sensors, messaging, camera, you name it.

This isn’t just Kuo’s fantasy. The idea that apps as we know them are dying has been gaining traction. Nothing CEO Carl Pei said at SXSW that apps will eventually disappear. And the vibe coding crowd has been predicting a future where you just talk to your device and it figures out what you need. OpenAI building a phone around that philosophy would be the most concrete step yet.

Kuo also says the phone would be designed to continuously understand user context — your habits, routines, preferences — and adapt accordingly. That kind of deep, persistent data collection is hard to pull off when you’re just an app running inside someone else’s OS. Owning the whole device changes the game. OpenAI would get a firehose of behavioral data that no third-party app could ever access.

To handle all that processing without draining battery or lagging, Kuo expects a mix of on-device small models for quick, local tasks and cloud-based models for heavier lifting. That’s the same hybrid approach we’ve seen in other AI-first hardware attempts, but OpenAI’s model quality and scale give it a real edge here.

Timeline-wise, Kuo says component suppliers and final specs should be locked down by late 2026 or early 2027, with mass production kicking off in 2028. That lines up with what OpenAI’s Chief Global Affairs Officer Chris Lehane said earlier this year — the company expects to announce its first hardware product in the second half of 2026. Multiple reports at the time pointed to those being earbuds, so maybe the phone comes later, or maybe the earbuds are just the first piece of a larger ecosystem.

OpenAI didn’t comment on the story when asked. But with ChatGPT nearing a billion weekly active users, building a dedicated device that puts AI front and center makes a lot of sense. Whether users are ready to ditch their app collections for an agent-driven interface is another question entirely.