Aiming to replace AirPods, OpenAI's new hardware is officially exposed.

After much anticipation, there is finally the latest news about OpenAI's self-developed hardware.

According to multiple sources, OpenAI is developing an AI headset with an internal code name "Sweetpea". The appearance is not the rumored "pen", but a rather mature in-ear hardware that can meet complex AI computing requirements.

This headset is designed by Jony Ive, the former chief design officer of Apple. Its goal is to replace AirPods and redefine the portable interaction paradigm in the AI era.

Different from the traditional in-ear wearing style of TWS headsets, "Sweetpea" uses a metal material to create an oval-shaped main unit, with two capsule-shaped units inside for wearing behind the ears. This design seems quite "unconventional", but it reserves sufficient space for high-performance chips and AI computing.

According to reports, "Sweetpea" will be equipped with a 2nm chip and plans to run most AI inference tasks locally. With a customized chip, it can even replace iPhone operations through instructions, truly breaking the interaction barriers between devices.

According to the plan, "Sweetpea" will be released in September 2028, with a first-year shipment target of up to 40 - 50 million units. This figure is approaching the single - product level of the iPhone, demonstrating OpenAI's absolute confidence in its market potential.

However, in terms of the time span, technology giants have accelerated the launch process of AI headsets, and the release time of "Sweetpea" is significantly behind.

As early as May 2024, OpenAI acquired the hardware company io founded by Jony Ive for $6.5 billion, completing a key reinforcement from AI algorithms to hardware design capabilities.

The io team was originally committed to developing screen - less interactive devices "born for the AI era", and "Sweetpea" is a continuation of this concept. This product abandons the dependence on screens and realizes environmental perception through built - in microphones and cameras, replacing passive command responses with proactive AI services.

If everything goes smoothly, this product is not only a highly topical piece of hardware but also an entry point for OpenAI to build an "AI companion" ecosystem. Its potential model of "subscription service + hardware terminal" is very likely to change the profit logic of traditional consumer electronics.

OpenAI's entry coincides with the critical node when the AI headset market is moving from the concept stage to the high - growth stage.

Market data shows that in the first quarter of 2025, the online sales of AI headsets in China increased by 960.4% year - on - year, and the single - quarter sales exceeded the whole year of 2024. The global market size is expected to grow from $217 million in 2025 to $680 million in 2031, with a compound annual growth rate of 20.99%.

Currently, three major camps have formed in this market: First, there are the innovators who innovate in appearance design, exploring multi - modal interaction by equipping headsets with cameras to achieve binocular vision perception; Second, there are terminal manufacturers such as Huawei and Apple. These manufacturers hope to use their established hardware ecosystems to turn headsets into smart entrances across devices; Finally, there are traditional acoustic manufacturers who hope to leverage their hardware foundation and add AI capabilities to gain a share of the AI headset market.

From another perspective, the core competitiveness of current AI headsets actually lies in the "proactive intelligence" ability.

Currently, most AI hardware remains in the "passive intelligence" stage of "waiting for commands", while OpenAI aims to develop proactive services. By continuously perceiving the environment and user habits, it can provide accurate suggestions without explicit commands.

Currently, large - scale AI models are already capable of supporting complex scene understanding, the cost of edge - side inference has decreased to enable local computing, and the maturity of multi - modal sensor technology has given traditional hardware "audio - visual capabilities".

However, while the prospects are broad, the AI headset market still faces multiple challenges.

Privacy disputes come first. The continuous environmental perception function inevitably involves the collection of user data. How to balance user experience and privacy protection has become a difficult problem that all players must solve. At the technical level, the balance between battery life and performance, the smoothness of multi - device collaboration, and the accuracy of interaction in noisy environments are still bottlenecks to be broken through. At the market level, although the "low - cost hardware + value - added service" model is accelerating its popularization, the user learning cost remains high.

How to make consumers accept the concept that "headsets are smart terminals" tests the product definition and marketing capabilities of enterprises.

For OpenAI, a greater challenge lies in supply chain integration. Lacking hardware mass - production experience, although it has joined hands with Foxconn to ensure production capacity, achieving the shipment target of 40 million units still requires crossing multiple hurdles such as yield rate and cost control.

In contrast, although Apple lags behind in AI capabilities, its excellent supply chain and cost advantages are enough to capture a share of the AI headset market this year.

This article is from the WeChat official account "Meku.com". The author is Meku.com. It is published by 36Kr with authorization.