Is the 'AI Phone' the most expensive delusion in tech history?

A provocative Reddit post argues that OpenAI building its own smartphone would be 'the most expensive delusion in tech history.' The author cites a graveyard of failed hardware gambits: Facebook's Portal, Amazon's Fire Phone, and more recently Humane's AI Pin and Rabbit's R1. None survived because hardware success requires trust, ecosystem, and user inertia—not just a better AI. OpenAI's motivation is clear: Apple and Google will never grant full access to the device's camera, location, or payment systems, which is essential for truly useful ambient AI. So instead of playing by their rules, OpenAI wants to own the entire stack.

The critique deepens by diagnosing the underlying strategy as fear-based. GPT-5 supposedly failed to create a decisive performance gap in the model race, pushing OpenAI to seek a hardware moat before competitors like Anthropic or Google do. But building a phone is not a software problem—it's a trust and habit problem. People don't switch phones for smarter AI; they stick with iMessage, their apps, and years of muscle memory. The author concludes that the AI winner in five years will be the one that disappears into the phone already in your pocket, not one that asks you to carry a second device.