RiseLink Redefines Edge AI with Low-Power Chip for Smart Devices

NUREMBERG, Germany – March 11, 2026 – RiseLink Technologies, a semiconductor firm with a two-decade history in low-power connectivity, today made a significant move to shift artificial intelligence from the cloud to the hardware itself. At the Embedded World 2026 conference, the company unveiled its BK7259 “powerhouse” chip, the cornerstone of a new strategy to enable a generation of smarter, more secure, and truly autonomous devices that can operate for extended periods on a single battery.

The launch positions RiseLink to compete in the fierce Edge AI market, where the demand for on-device processing is surging due to privacy concerns, latency issues, and the high cost of constant cloud communication. By integrating complex AI capabilities directly onto a single, highly efficient chip, the company aims to provide developers with the core components needed to build the next wave of intelligent products.

The Power of On-Device Intelligence

At the heart of RiseLink's announcement is the BK7259, a chip the company describes as a highly integrated cognitive hub. Built on a single die, it is designed to replace the inefficient and power-hungry multi-chip stacks common in today's smart devices. The chip combines high-definition video processing, advanced audio capabilities, precise motor control, and a dedicated AI processing unit into one unified platform.

This integration is powered by a formidable set of specifications. The BK7259 features dual Arm Cortex-M55 cores alongside an Arm Ethos-U65 microNPU (Neural Processing Unit), which together deliver 0.3 TOPS (Tera Operations Per Second) of AI acceleration. While TOPS is a raw measure of computational power, the chip’s real-world performance is enhanced by its unified memory architecture. This design allows the NPU, GPU, and CPU to access memory simultaneously, eliminating bottlenecks and preventing frame drops during complex, real-time tasks like object tracking and natural language processing.

Perhaps most critically for the future of IoT, the chip achieves this performance with remarkably low power consumption. RiseLink reports a keep-alive current of less than 60 microamperes (µA) and a deep sleep current of just 2 µA. These figures are crucial for what RiseLink CEO Dr. Pengfei Zhang calls “Battery-First Embodied AI.” This design philosophy prioritizes extreme energy efficiency, making it feasible to run sophisticated AI models on devices powered by small, single batteries, a feat that was previously impractical.

Redefining Privacy in the Smart Home

A key pillar of RiseLink’s strategy is addressing the growing consumer anxiety around data privacy in an increasingly connected world. By processing data locally, the BK7259 ensures that sensitive information never has to leave the device and travel to a cloud server, where it could be vulnerable to breaches or misuse.

The most compelling demonstration of this capability is the chip's support for fully on-device 3D facial recognition, which it can execute in under 200 milliseconds. This allows a smart lock, for example, to authenticate a user by scanning their face and verifying their identity entirely on the local hardware. The biometric data—the unique map of a person's face—remains securely on the device, offering a level of privacy that cloud-based systems cannot guarantee.

This shift to edge processing directly confronts a major hurdle for smart home adoption. As devices like cameras, smart speakers, and locks become more intelligent, the volume of personal data they collect grows exponentially. RiseLink is betting that a verifiable commitment to on-device security will become a powerful differentiator in a market crowded with competitors like NXP, Renesas, and Ambiq, all of whom are also racing to deliver secure, low-power AI solutions.

Building an Ecosystem for Edge AI Development

Recognizing that hardware is only one part of the equation, RiseLink is simultaneously launching the R2 Unified AI Developer Platform. This suite of tools is designed to reduce friction and accelerate the development cycle for companies building intelligent products. The platform provides verified drivers for popular components, including GC2145 and OV2640 image sensors and specialized circular displays like the ST7701S, allowing for rapid prototyping.

To streamline the complex process of deploying AI models on resource-constrained hardware, the R2 platform is built on the Armino SDK and embraces industry-standard tools. Support for CMSIS-NN, a collection of efficient neural network kernels for Arm Cortex-M processors, and TensorFlow Lite Micro, a framework for running machine learning models on microcontrollers, enables developers to leverage a vast and familiar ecosystem. This standardization is critical for attracting a broad developer base and ensuring that skills learned elsewhere can be applied to RiseLink's hardware.

Furthering this effort, the company is fostering an interactive developer community through dedicated forums on Reddit and a channel on Discord. These platforms provide a direct line of communication between product developers and RiseLink’s own engineers, creating a collaborative environment for problem-solving and innovation.

Strategic Alliances and a Vision for the Future

To accelerate its market penetration, RiseLink is deepening its long-term collaboration with Tuya Smart (NYSE: TUYA; HKEX: 2391), a global cloud platform provider that powers a vast ecosystem of smart devices. At Embedded World 2026, the two companies are jointly showcasing a range of products that leverage the BK7259's capabilities. These include the T5 AI board, a developer kit designed for creating LLM agents and AI-powered toys, and Tuya-powered smart locks that feature the on-device 3D face recognition.

This partnership provides RiseLink with a direct route into Tuya’s massive global network, which spans over 200 countries and supports thousands of product categories. The collaboration also extends to the growing smart energy sector, where Tuya’s AI-driven monitoring solutions can use the efficiency of edge processing to improve energy savings and cost-effectiveness for consumers and businesses.

The company's leadership is actively promoting its vision for a decentralized AI future. In an Embedded World 2026 video interview, Diana Zhu, RiseLink's Head of the United States, emphasized the company's commitment to “delivering more compute with lower power consumption.” This mission is central to the emerging trend of embodied AI—intelligent systems that can perceive and act upon the physical world through a hardware “body.”

Dr. Pengfei Zhang’s talk at the Exhibitor's Forum, titled “From Hearing to Seeing: Powering Battery-First Embodied AI with the BK7259,” frames the new chip as a foundational technology for this next frontier. By providing the “body and nervous system” for intelligent devices, RiseLink is enabling developers to focus on creating the unique “soul” of their products, paving the way for a future where autonomous, private, and energy-efficient AI is seamlessly integrated into the fabric of daily life.