AI Pin, image source: Humane official website

It’s fair to say that before its launch, the AI Pin generated a great deal of anticipation with its “high-concept” product design. However, shortly after shipments began in March 2024, the AI Pin was met with a wave of criticism, and the market quickly delivered Humane the “reality check” it deserved.

According to the latest data, shipments of the AI Pin are around 10,000 units—only one-tenth of Humane’s original target. And according to The Verge, excluding pre-sales and the initial launch month of April, the number of AI Pin returns from May to August 2024 was about 1,000 units. This means that roughly one in ten users chose to return their AI Pin, with the number of returns even surpassing sales during the same period.

Compared to the ambitious, sky-high expectations for the AI Pin, the story of the Rabbit R1 seems more grounded and realistic. Unlike the AI Pin, which uses projection to realize imaginative ideas, the Rabbit R1 takes a more straightforward approach, presenting itself as a practical AI device.

On January 10, 2024, Lu Cheng, former General Manager of Baidu’s smart home hardware division and founder of Raven Tech, launched the Rabbit R1. The Rabbit R1 is about the size of a music player. After linking accounts like Uber and Spotify, users can use voice commands to play music, hail a ride, order food, or plan trips.

Rabbit R1, image source: official website

Powered by large language models, the Rabbit R1 also incorporates visual recognition, enabling it to identify items in the fridge and suggest low-calorie dinner recipes. Leveraging its proprietary Large Action Model (LAM), it can learn and replicate user actions on a PC—simply issue a voice command, and the Rabbit R1 can perform previously executed operations like a robot. Yet, much like the AI Pin, the Rabbit R1 launched with fanfare but quickly lost momentum, ultimately becoming another fleeting experiment in the evolution of AI hardware.

Whether it’s the AI Pin or the Rabbit R1, their first-generation products can largely be considered failures in the AI hardware space. User experience fell short, their capabilities failed to surpass existing smartphones, and there were even suspicions of merely “repackaging” existing technology. Clearly, they did not meet the lofty expectations of “replacing the smartphone.”

IDC Vice President Francisco Jeronimo previously noted that whether consumers need entirely new hardware to use AI remains a contentious question. “Although these ideas have their merits, the reality is that consumers don’t need these devices—they need smartphones.”

Recent leaks suggest OpenAI is attempting its first truly “AI-native” hardware product. If successful, it could disrupt the longstanding dominance of Apple and Samsung in consumer electronics. However, as a challenger, OpenAI faces far steeper hurdles than simply enhancing existing devices.

The fundamental question is whether a new form of AI device can realistically challenge the smartphone’s status as the ultimate “super device.” Even the most promising AI glasses are still 3–5 years from mainstream adoption. Moreover, success in hardware is not just about individual products; it requires cross-device ecosystem integration. On this front, Apple and Google hold a commanding lead—something OpenAI cannot simply overcome with talent acquisition or financial investment.

From another perspective, the consumer electronics market has been relatively stagnant for some time. Regardless of whether OpenAI can overcome the so-called “hardware curse,” its efforts are likely to inject new energy into the industry and serve as a wake-up call for established players, urging them to move beyond incremental improvements.

Reports indicate that OpenAI’s first batch of AI devices could enter trial production as early as the first quarter of 2026, with an official launch expected by late 2026 or early 2027. Whether this venture will emerge as a breakthrough or another misstep by a pioneering company remains to be seen.