OpenAI chief Sam Altman said that the future of AI devices will go beyond today’s hardware capabilities. In a video posted on X, Altman called the Apple iPhone “the greatest piece of consumer hardware ever made,” but noted that it was not designed for an AI-first experience.

Altman suggested that future AI systems may require hardware capable of understanding ongoing conversations, remembering past interactions, learning user preferences, and remaining context-aware. He emphasised that today’s smartphones are too limited for the next generation of AI assistants. According to him, current devices only function when users actively interact with them.

The renewed focus on AI hardware comes amid speculation that OpenAI is planning to build an AI-focused smartphone. According to media reports and analyst Ming-Chi Kuo, the AI startup has reportedly accelerated development plans for an AI smartphone and could begin mass production as early as 2027.

OpenAI’s AI-focused smartphone: All you need to know 

The reported hardware is being described as an “AI agent phone” that could move beyond the traditional app-based interface. Instead of manually switching between apps for tasks such as messaging, shopping, or bookings, users may interact with AI agents capable of autonomously completing actions using contextual data from the device.

Must read: OpenAI, Jony Ive reportedly working on AI devices like glasses, smart speaker, AI Pin

OpenAI is also reportedly in talks with companies such as MediaTek and Qualcomm for processors designed to handle AI-centric workloads. Reports further suggest that MediaTek could emerge as the primary processor supplier for the expected AI hardware.

The AI agent phone is expected to feature LPDDR6 RAM, UFS 5.0 storage, dual NPUs, advanced image processing, and vision-related capabilities. Reports also suggest that the device could focus heavily on AI-powered camera experiences.

In addition, the smartphone is expected to include enhanced security features such as pVKM (protected visualisation) and inline hashing. However, it is also expected to rely on a mix of cloud-based and on-device AI, meaning data processing may not happen entirely locally.