Anker made its own AI chip (3 minute read)
Anker developed a custom compute-in-memory AI chip to enable local AI processing in small devices like wireless earbuds.
Deep dive
- Anker is entering the custom silicon market with Thus, a compute-in-memory AI chip designed for edge devices
- Traditional AI chips store the model in one location and perform computation in another, requiring constant data transfer between the two for every inference
- The Thus chip uses a compute-in-memory architecture where processing happens directly where the model is stored, eliminating this data movement bottleneck
- This design is smaller and more power-efficient, making it suitable for battery-powered devices with tight size constraints
- Anker is starting with wireless earbuds because they represent the most challenging use case: minimal space, limited power, and the chip needs to run continuously while worn
- Previous earbud AI chips could only run neural networks with a few hundred thousand parameters due to power limitations
- The Thus chip can handle several million parameters, a 10x+ increase in model complexity
- The first earbuds will combine the chip with 8 MEMS microphones and 2 bone conduction sensors for improved voice isolation
- The target use case is call noise cancellation that can cleanly separate voice from complex environmental noise where current solutions struggle
- Expected first products are the Liberty 5 Pro Max ($229.99) and Liberty 5 Pro ($169.99), leaked in March
- Full product details and additional AI features will be announced at Anker Day on May 21, 2026
- This positions Anker to compete directly with Apple AirPods Pro 3 and Sony WF-1000XM6 on AI-powered features
Decoder
- Compute-in-memory: An architecture where data processing happens in the same physical location as data storage, eliminating the need to move data back and forth between memory and processor
- MEMS microphones: Micro-electromechanical systems microphones, tiny microphones built using semiconductor manufacturing processes
- Bone conduction sensors: Sensors that detect voice vibrations transmitted through the skull, used to distinguish the wearer's voice from ambient noise
- Edge AI: Running AI models locally on a device rather than sending data to the cloud for processing
- Neural network parameters: The learned weights and values that define an AI model's behavior; more parameters generally means more capability but also more computational requirements
Original article
Anker made its own chip to bring AI to all its products
The Thus chip will come to earbuds first before being rolled out to other Anker products.
Image: Anker
Anker has announced its own custom silicon that the company says will bring local AI to audio devices, mobile accessories, and IoT devices. The Thus processor is the world's first neural-net compute-in-memory AI audio chip, which is smaller than traditional chips, and requires less power to run complex computations. That makes it an attractive solution for smaller devices.
When comparing Thus to existing chips, Anker CEO Steven Yang said, "Every AI chip built until now stores the model on one side and does the computation on the other. To think, the device has to carry all those parameters across, many times per second, every single inference. Thus puts the computation where the model already lives. The model never has to move again."
The first Thus chip will integrate into Soundcore's upcoming flagship earbuds. The company says it's starting with earbuds because they're the most challenging devices to put AI chips in due to size constraints. The small space limits the amount of power available, and because the chip is always active while you're wearing the earbuds, previous designs had to rely on small neural networks capable of handling a few hundred thousand parameters. But Anker says that with the more energy-efficient compute-in-memory design, the Thus chip is capable of handling several million parameters, significantly increasing the computing power to handle things like complex world noise.
Traditional call noise canceling relies on those small onboard neural networks and can have difficulty isolating your voice in very noisy environments, which results in ambient noise leaking through or voices getting highly compressed, making it difficult to hear. Anker says the larger neural network available on the Thus chip, plus eight MEMS (micro-electromechanical systems) microphones and two bone conduction sensors to focus in on your voice, in its yet-to-be-announced earbuds will have significantly cleaner call audio, regardless of the environment.
It sounds intriguing, but we'll have to see how the compute-in-memory Thus chip performs in the real world against the competition — including the Apple AirPods Pro 3 and Sony WF-1000XM6. Based on a leak in March, the first earbuds to include the Thus chip are likely the Liberty 5 Pro Max and Liberty 5 Pro, expected to be priced at $229.99 and $169.99, respectively. Anker will release full earbuds product details, as well as additional AI-powered features, at Anker Day on May 21.