Innodisk, a leading global AI solution provider, today announced the Innodisk GMSL2 Camera Starter Kit, an all-in-one vision solution that bundles four 3MP GMSL2 camera modules, a capture card, FAKRA cables, and other necessary accessories.

A milestone in Innodisk's edge AI vision strategy, the kit lets customers set up a multi-camera system and begin validation right out of the box, shortening the path from proof of concept (PoC) to deployment. Full camera source code is included, giving customers flexibility throughout development and maintenance.

As AI vision moves toward multi-camera perception in robotics and intelligent vehicles, integrating cameras, capture cards, drivers, and host platforms has become increasingly complex, often taking six months or more before validation. The starter kit's pre-integrated hardware and software driver enable system integrators to speed up time-to-market.

Synaptics Incorporated (Nasdaq: SYNA) today introduced the xMCU, a new class of microcontrollers that runs multimodal Edge AI, with the launch of the Synaptics Astra SR200 Series. Unlike conventional MCUs constrained by limited on-chip memory and compute, the xMCU combines heterogeneous, AI-native compute and acceleration with integrated high-bandwidth custom DRAM scalable to hundreds of megabytes, enabling larger AI models and richer multimodal workloads to run entirely on-device. The Astra SR200 Series of xMCUs is designed for the next generation of AI wearables, battery-powered sensing devices, connected cameras, streamers, and industrial vision and voice systems.

The SR200 Series is powered by the Synaptics Torq AI engine, which brings a standards-based hardware architecture and unified developer experience featuring the Coral NPU from Google Research. It also comes with the industry's first pairing of the Arm Cortex -M85 and Cortex -M52. Together with integrated vision and audio capabilities and hardware-rooted security, the SR200 Series brings unmatched system design efficiency and cost benefits to Edge and Physical AI designs.

NVIDIA today announced NVIDIA Jetson Orin Nano 2, a new robotics computer set to redefine entry-level edge AI—putting frontier-class generative AI performance in the hands of millions of developers.

As AI models grow smaller and more efficient, more edge devices can become autonomous systems that understand context, interpret language and images, and act in real time. To bring these capabilities into physical machines like robots, delivery and inspection drones, and vision AI systems, developers need compact, energy-efficient robotics computers built for edge AI.

MSI today announced the launch of the PRO MAX EDGE AI+, a next-generation AI desktop purpose-built for on-device AI computing. Designed to accelerate AI innovation, the PRO MAX EDGE AI+ delivers exceptional AI performance while keeping sensitive data local for enhanced privacy and security. Built to handle demanding AI inference workloads, it combines high-performance computing with enterprise-grade privacy and security, empowering users to unlock the full potential of local AI.

The Heart of Edge AI: Powered by the AMD Ryzen AI Max+ Platform
The PRO MAX EDGE AI+ is powered by the AMD Ryzen AI Max+ 300 Series Processors, with configurations available up to the AMD Ryzen AI Max+ 395. Delivering up to 126 TOPS of total AI performance, the platform integrates up to 96 GB of Variable Graphics Memory, integrated AMD RDNA 3.5 graphics with up to 40 Compute Units, and an AMD XDNA 2 NPU capable of delivering up to 50 TOPS of dedicated AI performance.

Acrab, a technology company building agentic AI compute infrastructure for the next generation of intelligent systems, today unveiled GELIX 1, its first-generation edge AI system-on-chip (SoC), together with Agent Box, a personal edge AI system powered by the company's full-stack computing platform. As AI moves from generating answers to completing tasks, agents increasingly need to understand context, remember preferences and coordinate tools and devices in real time. Running these capabilities locally can produce faster responses, keep sensitive information under the user's control and maintain core functions when cloud connectivity is limited.

For years, models in the 100 billion parameter class have required cloud infrastructure. GELIX 1 is designed to bring state-of-the-art AI models at this scale into locally operated edge systems. Powered by GELIX, Acrab's Agent Box is a high-performance personal edge AI center designed to put AI agents into action in a more personal and customized way, with local large model inference, persistent memory, multimodal interactions and agent orchestration capabilities.

Back in January, Congatec briefly outlined a "launch of five new modules featuring the new Intel Core Ultra Series 3" processor series. Over a month later, their "Panther Lake-H"-powered range has appeared online in COM-HPC and COM Express module forms. An introductory YouTube video, with a publication date of January 5, advertises a significant new generation leap for embedded computing and accelerated edge AI workloads. The best of these modules can deliver up to 180 TOPS of AI performance, presumably enabled by Team Blue's flagship Core Ultra X9 388H APU. The credit card-sized (55 x 84 mm/COM Express Type 10) MC1000 model can be configured with a maximum of 32 GB LPDDR5X (up to) 8533 MT/s "soldered down" system memory. A soldered LPDDR5X arrangement is detailed in the HPC/mPTL's spec sheet, up to 96 GB/8533 MT/s, on a SWaP-C optimized framework (95 x 70 mm/COM-HPC mini).

The ruggedized TC1000 and TC1000r models share the same overall dimensions (95 x 95 mm), that conforms to Congatec's COM Express compact format. The TC1000 board design sports dual SODIMM sockets, that can accept up to 64 GB DDR5/7200 MT/s memory modules, with total system capacity reaching 128 GB. The TC1000r's dram spec is described as: "96 GB LPCAMM2 memory based on LPDDR5x with up to 8533 MT/s" transfer rates. The largest "Panther Lake"-based option—HPC/cPTL (95 x 120 mm/COM-HPC Client Size A)—features an LPCAMM2 socket that accepts up to 96 GB of LPDDR5X memory, with max. transfer rates of 7466 MT/s. At the time of writing, Congatec's PR department has not issued a comprehensive PR piece. In the meantime, curious customers can peruse the manufacturer's detailed product pages, select various memory configs, and request quotes.

Chinese firm M5Stack has released its new AI Pyramid series, a small-form-factor edge AI PC aimed at developers and makers working with local inference workloads that stands out with its unique pyramid shape design. The lineup consists of two variants: AI Pyramid and AI Pyramid-Pro. Both models are built around the Axera AX8850 SoC, which combines an octa-core Arm Cortex-A55 CPU running at up to 1.7 GHz with a 24 TOPS INT8 NPU. The platform targets fully local AI processing, covering use cases such as computer vision, multimodal input, and on-device model inference, without depending on cloud acceleration.

Memory configurations differ between the two versions, with the standard AI Pyramid shipping with 4 GB of LPDDR4x, while the Pro model offers 8 GB. Storage is provided by 32 GB of onboard eMMC. The AX8850 supports hardware-accelerated H.264 and H.265 encoding and decoding at resolutions up to 8K at 30 FPS, as well as up to 16 concurrent 1080p decode streams. The specs are completed with dual Gigabit Ethernet and USB 3.0 ports, HDMI output, with the AI Pyramid-Pro adding HDMI input support. The device also integrates onboard audio hardware, an OLED status display, RGB lighting, and an active cooling solution. Pricing starts at $199 for the 4 GB AI Pyramid Computing Box, with the 8 GB Pro variant listed at $249 on the M5Stack website.

Quadric, the inference engine that powers on-device AI chips, today announced an oversubscribed $30 million Series C funding round, bringing total capital raised to $72 million. ACCELERATE Fund, managed by BEENEXT Capital Management, led the round. Uncork Capital returned with one of the largest insider commitments through its opportunity fund, joined by insider Pear VC. New investors include Volta, Gentree, Wanxiang America, Pivotal, and Silicon Catalyst Ventures. The funding comes as Quadric hits a revenue inflection: product revenues more than tripled in 2025 vs. 2024. Quadric is entering 2026 with accelerating design-win momentum, driven by growing adoption of the General Purpose NPU (GPNPU) processor IP across edge LLM, automotive, and enterprise vision applications.

"We've been deeply impressed by Quadric's innovative architecture, its disruptive approach to AI inference at the edge, and their strong market traction particularly in Asian markets," said Hero Choudhary, Managing Partner at BEENEXT. "Those attributes indicate a very clear path for further growth with a strong potential to be a generational business. We believe Quadric is poised to revolutionize the edge AI hardware sector, and we look forward to supporting their journey as they continue to push the boundaries of what is possible."

SAPPHIRE Technology is rolling out the EDGE AI Series, a new line of ultra-compact AI mini PCs designed to bring next-level performance and real-time intelligence to everything from content creation and office productivity to edge analytics and smart automation. Following a soft launch earlier this year, the SAPPHIRE EDGE AI Series is now preparing for full market availability.

At the heart of each system is the powerful AMD Ryzen AI 300 Series processor, combining a high-efficiency multicore CPU, AMD Radeon 800M graphics, and a built-in Neural Processing Unit (NPU) delivering up to 50 TOPS of AI acceleration-all in a sleek, highly compact footprint. From intelligent automation in offices and factories to local AI inference in healthcare and education, this platform is designed for professionals who need real-time AI performance with ultimate flexibility.