What Is ANKER Thus™? The AI Chip Behind Liberty 5 Pro

The match has gone into stoppage time, and you’re following every moment through your earbuds on your way home. Around you, other fans are doing exactly the same. Station announcements echo overhead, trains rumble in and out of the platform, and the noise of a busy city competes for your attention from every direction. Your earbuds are expected to handle all of it, but most audio chips were never designed for this level of real-time intelligence. They often rely on shared processing resources, cloud-based support and power-hungry workloads that can shorten battery life. ANKER Thus™ is built to change that. It’s a custom AI chip designed specifically for wearable devices, bringing faster on-device AI processing, greater efficiency and smarter audio experiences to soundcore earbuds launched in May 2026. Here’s what it is, how it works and why it could represent the next step in wearable AI audio.

What Is ANKER Thus™?

According to Anker, Thus is the world’s first compute-in-memory (CIM) AI audio chip. CIM is a chip architecture in which calculations take place directly within memory rather than in a separate processor. It is designed to run neural network processing on the device itself, without sending data to external servers or relying on conventional chip architectures that were not built specifically for AI workloads.

Traditional earbuds run AI features on general-purpose chips. That hardware was never purpose-built for this type of task. Thus is designed from the ground up to do one thing: run sophisticated audio intelligence inside a device small enough to sit in your ear.

As Anker CEO Steven Yang told The Verge at the chip’s announcement: “Thus puts the computation where the model already lives. The model never has to move again.”

“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.” — The Verge, April 2026

Why Did Anker Build Its Own AI Chip?

Because no existing chip could do what we needed it to do.

Consumer earbuds have always operated under a three-way tension: performance, power consumption and physical size. Push AI processing harder and battery life falls. Add more processing capacity and the chip runs hotter while demanding more space. Most earbuds address this by running limited on-device AI and offloading more demanding tasks to cloud servers.

Cloud processing introduces its own challenges. There is latency between sending audio data and receiving a processed result. It requires a stable network connection. And your voice data leaves the device entirely, which is an increasing concern for users who care about privacy.

Our answer was not to optimise around the bottleneck. It was to remove it architecturally. Thus is the result.

How Does the ANKER Thus™ Chip Work?

Thus™ achieves its efficiency by closing the gap between where AI models are stored and where computation takes place. It sounds like a small architectural shift, but the implications for what earbuds can actually do are significant.

How Traditional Chips Waste Power Moving Data

In a conventional chip, memory and compute are physically separated. Processing audio means continuously moving data between storage and the processing unit: filtering noise, analysing voice frequencies and isolating speech from ambient sound.

That movement is expensive. In conventional chip design, over 90% of energy consumption goes into transporting data between memory and processing units, according to research published by the IEEE Computer Society. Only a fraction is used for the actual computing. This constant back-and-forth also limits how complex an AI model can be: most earbud chips can only handle models with hundreds of thousands of parameters.

How Thus Brings Computation to the Data

Thus reverses this entirely. Instead of moving data to a compute unit, computation happens inside the memory itself. The data stays where it is.

Two things follow from this shift. First, the energy previously consumed by data transport becomes available for actual AI processing. Second, without that overhead, Thus can support AI models with several million parameters, compared with the few hundred thousand parameters that previous chip designs could handle, as reported by The Verge, representing roughly a 10× increase in model scale. The cumulative effect is significant: based on internal lab tests, Thus delivers up to 150× the peak AI computing power of our previous flagship earbuds.

What this increase in scale means in practice is that a larger model can recognise more types of sound, adapt to more complex and layered noise environments, and perform better in situations where a smaller model would struggle. The difference is not especially noticeable in a quiet room. It becomes apparent on a construction site, at a crowded event, or anywhere the acoustic environment is unpredictable and constantly changing.

Up to 150× peak AI computing power vs. the chip in our previous flagship earphones — NOR Flash CIM architecture, based on internal lab tests.

What Can ANKER Thus™ Actually Do?

Thus enables key AI audio capabilities on soundcore earbuds. These include advanced call processing with a 10-sensor fusion system for demanding environments, and on-device processing that reduces dependence on network connections for core features.

Clear Calls and What It Delivers in Real Life

The most tangible application of Thus is Clear Calls, our AI-powered call noise cancellation feature built on the Thus™ AI engine.

Clear Calls uses a 10-sensor array: 8 MEMS (micro-electromechanical system) microphones and 2 bone-conduction sensors. The MEMS microphones capture the full acoustic environment around you. The bone-conduction sensors detect vibrations through your skull. That signal is almost entirely your voice, with very little surrounding noise mixed in.

Thus processes all 10 streams simultaneously, in real time. Think back to the busy train platform from the opening. The train pulling in. The crowd moving through the station. With Clear Calls active, the person on the other end of the call hears your voice — not the platform announcements, not the trains, and not the noise of people around you. The gap between what the microphones pick up and what your caller actually hears is exactly where Thus™ does its work.

On-Device AI Without the Cloud Delay

Because Thus processes key AI features on the device, these capabilities work without requiring continuous cloud connectivity. There’s no audio sent to a server, no wait for a response, and no performance drop when your connection is weak.

That means noise cancellation adjusts as your environment changes, rather than waiting for a network round-trip.

It also means that, for these core functions, audio processing stays on your device. When noise cancellation and audio analysis run locally, there’s no audio stream being sent to a server for analysis and no reliance on how an external service stores or manages that data. Your voice still travels over your mobile network when you make a call — what stays on-device is the AI layer that cleans it up beforehand. For that layer, on-device processing is not only faster. It is also more private in ways that matter beyond audio.

What soundcore Earbuds Will Feature ANKER Thus™?

Thus has made its debut in our new soundcore earbuds, launched at Anker Day on 21 May 2026.

We’ve built Thus into them to bring advanced AI audio capabilities — including Clear Calls with a ten-sensor fusion system — to everyday use without compromising battery life.

We’re also officially attempting the Guinness World Records™ title for “Highest speech quality score (G-MOS) for TWS earbuds (objective test)”. G-MOS is a standardised benchmark that measures how clearly a speaker’s voice comes through on a call, and it reflects real call performance rather than just controlled test conditions. We shared the results at Anker Day on 21 May.

The first models to feature this chip include the soundcore Liberty 5 Pro and soundcore Liberty 5 Pro Max, which combine advanced AI technology with flagship-grade audio performance to make every call as clear as a face-to-face conversation.

Specific model names, pricing and full specifications were confirmed at Anker Day.

Conclusion

Thus is the result of building something from scratch.

Calls that hold up in challenging environments, AI that responds with lower latency, and processing that keeps your audio data local. That’s what on-device AI was always supposed to feel like.

The first soundcore earbuds powered by Thus launched on 21 May at Anker Day. Get the full details and sign up for launch updates directly on our site.

FAQ

What does “compute-in-memory” mean in simple terms?

Compute-in-memory means calculations happen directly inside the memory chip rather than in a separate processor. Think of it like doing maths on a sticky note rather than copying the numbers to a whiteboard, working them out there, and then copying the result back. The data does not need to move in order to be processed, which saves both time and energy. In earbuds, this means more powerful AI that uses less battery power.

When willdid the first Thus-powered soundcore earbuds be available?

The first soundcore earbuds featuring ANKER Thus™ launched at Anker Day on 21 May 2026. Registration for launch updates is available on our site.

Does ANKER Thus™ require an internet connection to work?

No. Thus runs key AI audio features directly on the device, so core capabilities such as call noise cancellation work without requiring continuous cloud connectivity. There’s no server dependency for the AI layer and no cloud round-trip. Your call audio still travels over your mobile network, as with any phone call — what Thus keeps on-device is the noise cancellation and audio analysis that happens before and after.

How does Thus differ from chips in conventional earbuds?

Most earbuds on the market today use processors where memory and compute are physically separate. That separation limits both the size of the AI model that can run on-device and how efficiently the chip uses power for AI tasks. Thus uses a compute-in-memory architecture, which removes that separation entirely. The result is a larger, more capable AI model that uses less energy to run. That difference becomes most noticeable in demanding real-world conditions, such as noisy call environments.