OpenAI’s “Sweetpea” Earbuds To Use Samsung’s 2nm Exynos Chip, Titan ASIC Launching By The End Of The Year

OpenAI seems intent on building a sprawling hardware ecosystem to complement its AI-based subscription services in order to lock in customers within its gated fortress, which would soon span across the various GPT-class AI models, dedicated AI ASICs, and a range of consumer devices, including the iPod Shuffle-like "Gumdrop" and AI earbuds "Sweetpea."

OpenAI's "Sweetpea" earbuds to tap Samsung's 2nm Exynos chip

According to the latest report out of Taiwan, OpenAI is apparently working on dedicated AI-powered earbuds that bear the internal codename "Sweetpea."

While this tidbit is light on juicy details, the report does mention that the upcoming OpenAI earbuds would rely heavily on cloud-based AI processing, while featuring a 2nm Samsung Exynos chip for some on-device processing.

The report, however, remains mum on exactly which Exynos variant the Sweetpea is likely to leverage. After all, Samsung's current-gen Exynos 2600 chip, which is expected to launch with the Galaxy S26 series next month, is based on the South Korean behemoth's 2nm (SF2) GAA process.

Moreover, Samsung's next-gen Exynos 2700 is widely expected to leverage its SF2P process, replete with ARM C2 Cores, improved thermals, LPDDR6, and UFS 5.0.

Of course, OpenAI is also working on another consumer device that is reportedly shaped like a pen and sports a size similar to that of the Apple iPod Shuffle. The device bears the internal codename "Gumdrop," and is entirely bereft of a dedicated screen. Additional details include:

1. Contextual awareness via a suite of sensors, including cameras and microphones.
2. The device will be able to run OpenAI's tailored AI models locally, with cloud computational support available for more compute-intensive tasks.
3. It will be able to convert handwritten notes into text and instantly upload them to ChatGPT.
4. The device will be able to communicate with its peers, akin to what we currently do with smartphones.
5. It won't bear the form factor of a wearable, but may be carried in the pocket or worn around the neck.
6. Expected to launch in 2026 or 2027.

OpenAI continues to work with Broadcom on its Titan AI ASICs

The report goes on to note that OpenAI is working with Broadcom on designing a dedicated AI-focused ASIC that bears the internal codename "Titan." The ASIC will be manufactured on TSMC's 3nm process, and is expected to debut by late 2026.

This comes as Broadcom is already expected to provide advanced networking, optical links, and other hardware to make OpenAI’s data centers run faster and smoother.

Of course, by launching its own ASIC, OpenAI aims to reduce its reliance on NVIDIA's GPUs while also improving its negotiating position. We already know that Google is pursuing a similar strategy with its TPUs, while Amazon is banking heavily on its Trainium ASICs.

Coming back, the report goes on to note that the next generation of the Titan chip (Titan 2) is expected to leverage TSMC's A16 process, which should result in a meteoric jump in overall performance.

Follow Wccftech on Google to get more of our news coverage in your feeds.