Synaptics Blurs the Lines of Edge AI with the New 'xMCU' Astra SR200 Series

SAN JOSE, Calif. – October 20, 2026 — The artificial intelligence revolution is rapidly migrating from massive, centralized cloud servers to the deep sensor edge. But as device manufacturers attempt to cram increasingly complex multimodal AI into battery-powered wearables and smart home devices, they have hit a silicon wall. Traditional microcontrollers (MCUs) lack the memory and compute muscle for advanced neural networks, while application processors (MPUs) drain batteries and drive up costs. Today, Synaptics Incorporated aims to demolish that wall with the launch of the Astra SR200 Series, introducing a new category of silicon it calls the "xMCU."

Designed for the next generation of AI wearables, connected cameras, and industrial vision systems, the SR200 series integrates heterogeneous compute, a transformer-ready neural processing unit (NPU), and scalable high-bandwidth custom DRAM into a single package. It is a strategic play that allows complex AI models to run entirely on-device, privately and in real-time, without the overhead of a Linux operating system.

Bridging the Microcontroller and Application Processor Divide

The fundamental bottleneck in edge AI has historically been memory. Running multimodal workloads—experiences that simultaneously process voice, vision, audio, and motion—requires rapidly shuffling vast amounts of data. Standard MCUs, constrained by a few megabytes of on-chip SRAM, simply cannot hold these models.

Synaptics' solution with the Astra SR200 series is an advanced multi-die System-in-Package (SiP) architecture. The design pairs up to 8MB of SRAM with extensible, high-bandwidth custom DRAM that can scale up to hundreds of megabytes. For instance, the newly available Coralboard SR development kit supports up to 256MB of this integrated memory. Operating across a 64-bit parallel bus, this custom DRAM pipeline delivers throughput up to 4 Gigabits per second directly inside the package. Crucially, engineering benchmarks indicate that this architecture runs workloads at one-third the power of typical system DRAM, eliminating the need for costly and power-hungry external memory routing.

Processing this data is an industry-first dual combination of Arm processor cores. The xMCU features an Arm Cortex-M85 scaling up to 800MHz to tackle high-performance control and burst-compute workloads, paired with a highly efficient Cortex-M52 running up to 200MHz for always-on sensor fusion and wake-up functions.

"AI is quickly becoming the most important use case for Edge devices, with workloads increasingly spanning always-on sensing and inference through to more demanding voice, vision, and multi-media processing," noted Chris Bergey, EVP of Edge AI at Arm. "By bringing Arm Cortex-M85 and Cortex-M52 together for the first time, Synaptics is demonstrating how the Arm compute platform gives partners the flexibility to balance performance and efficiency."

Privacy-First Physical AI Moves to the Sensor Edge

The commercial implications of this processing power are profound. The global edge AI market is projected to skyrocket, with some industry analysts forecasting it to reach $225.5 billion by 2035. This growth is largely driven by consumer and enterprise demand for "Physical AI"—devices that understand their environment and act locally without sending sensitive data to the cloud.

"The next wave of AI will be defined by devices that understand their environment and act locally, while meeting tight power, cost, and form-factor requirements," said Vikram Gupta, SVP and GM of the Edge Compute & Connectivity Division at Synaptics. "With the Astra SR200 Series, we're introducing a new class, the xMCU, that brings applications-processor capability into the MCU domain."

This localized processing drastically reduces latency and bolsters privacy, a critical selling point for smart home and security companies. Cristina Rivas, Chief Technology Officer at Verisure, emphasized this shift, noting that the future of home security relies on devices that understand rather than just detect. By running vision, voice, and audio AI models directly on the device, manufacturers can ensure faster, more responsive intervention without compromising user privacy.

Similarly, audio pioneer Sonos is leveraging the platform to advance wireless, multi-room ecosystems. Kurt Preston, Chief Supply Chain Officer at Sonos, stated that Astra's scalable compute and audio processing are exactly the capabilities needed to push the boundaries of connected audio, enabling complex device synchronization without draining battery life.

For industrial and enterprise applications, the SR200 series also targets augmented reality. Lerdah Wang, CEO of Jorjin Technologies, pointed out that next-generation AR wearables demand real-time perception and seamless interaction. By utilizing the SR200's integrated vision architecture and high-efficiency NPU, manufacturers can execute complex scene understanding and low-latency overlay rendering on a single, highly integrated chip.

Breaking Silos with Open-Source Toolchains

Hardware is only half the battle in the AI revolution; the software ecosystem is equally critical. Historically, the embedded AI landscape has been fragmented by proprietary vendor toolchains, locking developers into specific silicon ecosystems. Synaptics is actively challenging this paradigm through a deep partnership with Google Research.

At the heart of the SR200 family is the Synaptics Torq AI engine, which integrates the second generation of the Coral NPU. Google open-sourced the Coral NPU architecture in October 2025, and Synaptics has emerged as its first major commercial silicon partner. Optimized for deep learning workloads—including Convolutional Neural Networks (CNNs) and transformer models—the engine utilizes an open-source MLIR/IREE compiler.

This allows developers to use industry-standard frameworks rather than proprietary tools, extending the Coral ecosystem into ultra-low-power IoT devices running the Zephyr Real-Time Operating System (RTOS).

"Open technologies and strong partnerships are key to making Edge AI more accessible," explained Yossi Matias, Vice President and Head of Google Research. "Our ongoing work with Synaptics combines the Coral platform with Astra's scalable hardware and developer ecosystem to power next-generation, wearable-class devices."

The Road Ahead for Commercial Adoption

Synaptics is launching the Astra SR200 Series in three pin-compatible tiers, allowing OEMs to scale performance while preserving their software investments. The power-optimized SR210 delivers 200 GOPS for always-on voice and sensing. The mainstream SR230 steps up to 300 GOPS, adding vision and display capabilities. Finally, the premium SR250 pushes 400 GOPS with dual vision and display paths for demanding edge applications.

While Synaptics is carving out a unique niche, it faces stiff competition in the broader edge processing market. Rivals like STMicroelectronics have recently introduced platforms like the STM32N6, which boasts an NPU capable of 600 GOPS, and NXP continues to iterate on its i.MX RT crossover MCUs. However, Synaptics is banking on its unique multi-die memory integration, hardware-rooted security targeting PSA Level 3, and the open-source Google Coral ecosystem to differentiate the Astra line.

Currently, the SR200 series chips are sampling to partners, and developers can begin prototyping with the newly available Coralboard SR through distributors. As consumer expectations for smart, responsive, and privacy-centric devices continue to rise, the commercialization of xMCUs represents a pivotal step forward. By successfully marrying the power efficiency of microcontrollers with the computational muscle required for modern AI, the semiconductor industry is laying the groundwork for a truly intelligent, decentralized digital ecosystem.