Author: Jiu Lin
Traditional earphones, dead?
AI earphones are becoming mainstream in the wireless earphone market. According to estimates by Research and Markets, the global AI earphone market size was approximately USD 5.99 billion in 2025, and is expected to climb to USD 7.42 billion in 2026, reaching USD 17.34 billion by 2030.
The most frequently repeated judgment in the consumer electronics industry is: all hardware is worth being reimagined with AI.
Since the beginning of this year, AI-related features in earphone products have been increasing: voice wake-up, call noise cancellation, recording transcription, real-time translation, and smart assistants are all being incorporated into the narrative of "AI earphones." Although AI earphones have not yet become a category with clear boundaries, chipmakers have already started preparing the next-generation platforms for them.
01. AI Earphones: What Kind of New Species Are They?
Looking only at new products from the past year, AI is becoming a crucial enabler for earphones to enter the CNY 1,000 price bracket.
In September 2025, Huawei launched the open-ear FreeClip 2, priced at CNY 1,299. Compared to the previous generation, this product is equipped with a dedicated NPU AI processor for the first time, with AI computing power increased by about 10 times, focusing on tasks such as voice enhancement, calling, and interaction.
In August this year, Shokz introduced the OpenFit 2 AI, priced at CNY 1,398. Integrated with Alibaba's Qwen large language model, it focuses on long-duration recording, voice transcription, meeting summarization, and multi-language translation. Earphones have thus transformed from "playback devices" into portable voice portals.
iFlytek's approach is closer to a task-oriented terminal. Its AI translation earphones went on pre-sale in November 2025, priced at CNY 2,499, covering scenarios such as face-to-face translation, call translation, online simultaneous interpretation, and eavesdropping simultaneous interpretation. Teardowns in May 2026 showed that these products are more complex than ordinary TWS earphones in terms of the main controller, microphone array, and local storage. Similarly, in 2026, AI earphone solutions with eSIM-connected charging cases appeared in the global market. Plaud integrated earphones with AI Agents, attempting to reduce reliance on smartphones.
OpenAI acquired LoveFrom, a design company co-founded by former Apple Chief Design Officer Jony Ive, for approximately USD 6.5 billion. According to multiple supply chain sources, the collaborative product codenamed "Sweet Pea", is highly likely to be a smart earphone focusing on AI interaction, featuring a custom chip built on the 2nm process, aiming to achieve local processing of AI tasks on the earphone itself, and is planned for release in the second half of 2026.
Cameras are pushing AI earphones toward "omni-perception" devices. In the spring of 2026, Guangfan Technology launched the Lightwear camera AI earphone. Publicly reported launch prices started at CNY 1,799, with some versions priced at CNY 1,999. The product adds a camera to the earphone and utilizes the charging case's cellular network and positioning capabilities to collaboratively process images, sound, and location. It no longer just understands what the user says, but also attempts to understand what the user sees. Apple's camera-equipped AirPods remain in the rumor and supply chain preparation stage, with multiple media outlets expecting the release to be delayed to 2027.
At this stage, AI earphones can be roughly divided into three categories.
The first category is smartphone-coordinated. The earphones are responsible for capturing sound, playing results, and triggering commands, while complex computations such as speech recognition, real-time translation, and open-domain question answering are handled by the smartphone or the cloud. The real-time translation of AirPods Pro 3 is a typical example, which requires connection to an iPhone compatible with Apple Intelligence. Its advantage is that the earphone hardware modifications are limited, and features can be upgraded along with the smartphone system; the limitation is that once separated from the smartphone, the earphones have almost no independent AI capabilities.
The second category is edge-enhanced. Low-latency tasks such as voice activity detection, keyword wake-up, echo cancellation, call noise cancellation, human voice separation, and environment recognition are processed on the earphone side, while complex recognition and large model Q&A are still handed over to the smartphone or the cloud. Samsung's Galaxy Buds3 Pro and Huawei's FreeBuds Pro 5 "AI hands-free conversation" both belong to this direction. Whether the earphones can stably capture sound in noisy environments and switch to transparency mode in a timely manner depends on edge algorithms and the main control chip, not just the capabilities of cloud models.
The third category is task-oriented AI terminals. They redesign the software and hardware division of labor for earphones around recording, transcription, translation, and meeting minutes, requiring longer listening durations, larger data caching, and more stable connections. The earphones, smartphone, and cloud thus form a distributed computing system: wake-up and noise cancellation are completed in real-time at the ear, transcription can run on the earphones or smartphone, while translation, summarization, and open-domain Q&A rely more on the cloud.
Once computing tasks migrate to the earphone side, the issue shifts from "adding a software feature" to a systems engineering challenge involving the main controller, memory, storage, and power management. With a weight of just a few grams and a battery of tens of milliamp-hours, simultaneously maintaining Bluetooth connectivity, audio playback, active noise cancellation, sensor acquisition, and neural network inference, the chip has become a question that the industry chain must answer.
02. AI Earphones: Chips Lead the Way
The competition in AI earphones does not start with large models first, but with main control chips.
The main control SoC (System on Chip) of traditional Bluetooth earphones is mainly responsible for Bluetooth connectivity, audio codec, active noise cancellation, microphone acquisition, touch interaction, and low-power management. With the addition of AI capabilities, the chip also needs to undertake voice activity detection, keyword wake-up, human voice separation, environment recognition, and partial neural network inference. Therefore, the main control chips for AI earphones usually follow two routes: one is to integrate an NPU or AI acceleration unit inside the traditional audio SoC; the other is to add an independent AI audio chip to work in coordination with the original Bluetooth main controller.
Bestechnic BES2800 is currently one of the most representative models among AI earphone main control chips. Previously, during the launch of the Samsung Galaxy Buds3 series, Bestechnic confirmed that the series would debut with the BES2800. Built on a 6nm FinFET process, this chip integrates multi-core CPU, GPU, NPU, RAM, low-power Wi-Fi, and dual-mode Bluetooth, and is no longer just a Bluetooth audio chip in the pure sense. Compared to the previous generation BES2700, the BES2800's CPU performance has approximately doubled, and NPU performance has improved by up to four times. Features such as Galaxy AI real-time translation supported by the Samsung Galaxy Buds3 Pro still require the participation of Galaxy smartphones, but microphone acquisition, noise cancellation, voice activity detection, and audio transmission on the earphone side all need to be completed by the main control chip.
Qualcomm has merged AI, Wi-Fi, and high-end audio to launch the Snapdragon S7 and S7 Pro Sound Platforms. The Xiaomi Buds 5 Pro Wi-Fi edition adopts Qualcomm's S7 series audio platform. The S7 platform integrates multi-core DSP, sensor hub, ultra-low-power AI engine, and Bluetooth connectivity. The S7 Pro also adds ultra-low-power Wi-Fi and XPAN technology.
Google Pixel Buds Pro 2 adopts Google's self-developed Tensor A1 chip. Tensor A1 is a custom chip specifically designed for earphone audio processing, capable of processing ambient sound at a frequency of about 3 million times per second, used for active noise cancellation and voice enhancement. Pixel Buds Pro 2 is also Google's first earphone product designed around Gemini. The Pixel Buds 2a launched in 2025 also uses Tensor A1 and brings active noise cancellation to the Pixel A series earphones for the first time.
Vortech WQ7036AX is a high-performance, low-power Bluetooth audio SoC launched by Vortech Microelectronics, mainly positioned as the audio main controller for smart audio glasses and high-end wireless audio devices. Adopting a "RISC-V + high-performance DSP" dual-core architecture, teardown reports of Huawei's first-generation FreeClip show that the earphone uses the Vortech WQ7036AX audio main control chip. Teardowns of iFlytek's AI translation earphones also show that they use the same series of main controllers for wireless connectivity, audio data processing, complex audio algorithms, AI noise cancellation, and multi-microphone ENC.
Thus A1, jointly developed by Anke and Zhicun Technology over three years, is called by both parties as the first neural network computing-in-memory AI audio chip. According to disclosures, the overall computing power of traditional Bluetooth audio chips is about 30M FLOPS, while Thus A1 reaches 5G FLOPS, with theoretical computing power increased by about 150 times; at the same time, the chip does not come at the cost of significantly increased power consumption.
03. Storage Is Another Threshold for AI Earphones
The next threshold for AI earphones is not just the main control chip, but also storage. After all, ordinary TWS earphones are equipped with a single 32M or 64M NOR Flash chip per earbud, while taking the Doubao AI earphone as an example, with a price of over CNY 1,000 and equipped with two 128M large-capacity NOR Flash chips per earbud and four per pair, the ASP increase brought by AI earphones can reach 3 to 5 times.
From the product roadmaps of storage companies, the storage upgrade for AI earphones mainly has four directions.
First, capacity continues to increase. NOR Flash will expand from 32Mbit and 64Mbit to 128Mbit, 256Mbit, and 512Mbit, while SPI NAND will take on larger capacity data storage for recordings, meeting files, and voice caching.
Second, the distance between the memory and the main controller is shortened. Model inference is more sensitive to bandwidth and latency. On-chip SRAM, external PSRAM, and highly integrated ePOP can reduce data movement and improve edge processing efficiency.
Third, packaging develops towards miniaturization. BGA, WLCSP, KGD, ultra-thin packaging, and multi-chip package-in-package will become important directions for earphones and other wearable terminals. Memory cannot just look at capacity; it must also meet the earphone motherboard's requirements for thickness, area, and heat dissipation.
Fourth, the importance of low power consumption and data security rises. AI models need to reside for a long time, and recording data also involves privacy. Memory must not only reduce standby and read/write power consumption but also support secure boot, firmware protection, data encryption, and reliable OTA upgrades.
The product layout of storage manufacturers is also changing accordingly. PUYA's P25Q series covers low-voltage NOR Flash, with the P25Q128SN having a capacity of 128Mbit and an operating voltage as low as 1.1-2.0V; the company has listed AI earphones, AI glasses, and AR/VR as key directions in consumer electronics and disclosed that related NOR products have achieved mass shipment. Hengshuo Semiconductor has launched 256Mbit NOR and is advancing 512Mbit products, with applications covering smart wearables and IoT (Internet of Things) terminals. The Wuhan Xinxin XM25Q series covers 16-256Mbit, targeting wearable devices that require larger firmware and algorithm space.
When model weights and audio caching require higher bandwidth, PSRAM becomes another route. Unigroup Guoxin launched 32Mbit, 64Mbit, and 128Mbit PSRAM products in 2025, compatible with the Xccela interface, adopting BGA24L ultra-thin packaging, and supporting KGD form. AP Memory's APS6404L provides 64Mbit QSPI PSRAM, and Winbond also provides PSRAM and HyperRAM for low-power terminals. The advantage of PSRAM is its higher read/write speed and random access efficiency, making it suitable for runtime caching, but it requires additional packaging and power supply, so designers must make trade-offs between bandwidth, standby power consumption, and motherboard area.
Biwin Storage's route leans more towards highly integrated storage. The company's products cover SPI NOR Flash, SPI NAND, eMMC, UFS, ePOP, and LPDDR, and it lists smart wearables as an application direction. Its 144-ball ePOP based on LPDDR4X packages eMMC 5.1 and LPDDR4X together, with a chip size of 8.0×9.5×0.8 mm and a maximum capacity of up to 32GB+16Gb. Biwin has also launched ePOP5X to expand the lightweight storage capabilities of AI wearable devices.
ePOP is not a standard configuration for traditional TWS earphones, but it represents another possibility: as earphone forms develop towards recording earphones, conference earphones, and task-oriented terminals with independent computing capabilities, a single NOR Flash can no longer meet the demand, and memory needs to be integrated with running memory to a higher degree. Through the combination of ePOP, LPDDR, and embedded storage, terminals can obtain larger capacity and higher bandwidth within limited space.
04. AI Earphones with Cameras Are Here Too?
Notably, some AI earphones also come with cameras.
In addition to the AI earphones from Guangfan Technology mentioned earlier, Shenzhen VibeLens previously launched the MusicCam earphones, which is the world's first AI camera earphone successfully crowdfunded on Kickstarter. This product focuses on outdoor sports scenarios, and therefore has high requirements for camera performance, equipped with a Sony IMX219 sensor, featuring 32 megapixels (32MP) and supporting 1080P video recording. Notably, the camera is rotatable, supporting a ±30° adjustable angle (left or right) and a 73° field of view width.
AI camera earphones are not an entirely new concept. As early as 2023, Apple and Huawei filed multiple related patents, describing a head-mounted device integrating a camera and microphone that can capture the user's field of vision and perform contextual awareness, suitable for scenarios such as assisted navigation.
For environmental perception earphones (camera earphones), domestic chip company Ingenic has also launched one of the industry's smallest wearable ISP solutions. The CW020 series features an ultra-small 2×5.5mm (11mm²) package and a DDR-less architecture. With an area of only 11mm² (about the size of a grain of rice), it can be easily laid out in spaces where traditional ISPs cannot enter, such as earphone stems and glasses temples, without the need for DDR/PSRAM, requiring only SPI NOR Flash.
05. Conclusion
According to IDC data, in the Chinese Bluetooth earphone market in 2025, true wireless products shipped 77.21 million units, a year-on-year increase of 6.7%, driven by entry-level products and shipment policies bundled with smartphones. Open-ear products shipped 29.96 million units, a year-on-year increase of 20.2%. Open-ear products have gradually shifted from a stage of rapid explosion to steady growth, with segmented markets showing different development structures.
The earphone track is becoming increasingly popular. With the support of large models, smart wearables are expected to change the habit of scrolling through smartphones. The AI earphone market in 2026 has not yet formed a unified product boundary, but changes in the supply chain have long been surging beneath the surface.
Earphone main controllers have expanded from Bluetooth connectivity and audio codec to NPU, sensor fusion, secure boot, and model updates; storage has expanded from a single firmware Flash to a combination of NOR, NAND, PSRAM, and ePOP; cameras, cellular communication, and positioning are pushing some products towards wearable computing terminals. In the future, with the continuous upgrading of AI, connectivity, and low-power technologies, earphones will become one of the important carriers for AI smart hardware.