Edge Ai

TAIPEI – November 06, 2026 -- Innodisk reported third-quarter 2026 revenue of NT$25.9 billion, up 579.9% year on year, as AI-related businesses grew to more than 40% of group revenue.

Net income reaches NT$9.1 billion as margins widen

Gross profit rose 925.5% year on year to NT$12.8 billion, a gross margin of 49.5%. Operating income climbed 1,482.0% to NT$11.4 billion, an operating margin of 43.9%.

Net income grew 1,310.7% to NT$9.1 billion. Earnings per share came in at NT$93.09.

AI infrastructure memory and storage draws cloud and networking customers

The Taiwan-based maker of industrial-grade memory and storage is adding customers among global cloud service providers and leading networking and communications companies. Its targets are AI infrastructure, data centers and high-performance computing.

PITTSBURGH – November 03, 2026 -- Efficient Computer's Electron E1 processor delivers up to a 100X improvement in energy efficiency over low-power CPUs, and edgeAI has selected it for its physical AI and critical infrastructure products.

edgeAI runs large AI models on-device with lower power and cost

Efficient Computer says the Electron E1 outperforms many AI accelerators on both performance and energy efficiency. According to the company, this lets edgeAI run large AI models on-device while easing power, cost and form-factor constraints.

edgeAI CEO Yang Jung Woong said that after evaluating a wide range of processors, "none matched" the Electron E1's combination of performance and efficiency. He said the chip runs edgeAI's models in real time on a fraction of the power. It also gives the products headroom to support the latest vision and compute stacks for years.

Shanghai – November 03, 2026 -- PATEO has secured a project nomination for a new-generation vehicle from a leading new energy vehicle automaker, with mass production expected in the second quarter of 2027, as it deploys NVIDIA Nemotron and ByteDance's 35-billion-parameter Doubao language model in its AI Box built on NVIDIA DRIVE AGX Thor.

At Auto China 2026 in Beijing, PATEO announced joint development of the solution with ByteDance and a leading new energy vehicle automaker. The AI Box pairs local inference with PATEO's AI Agent platform to handle multilingual interaction, read cabin context and coordinate vehicle functions.

Oslo – November 03, 2026 -- Nordic Semiconductor has introduced the nRF71 Series of ultra-low-power systems-on-chip, which cuts standby power by up to 20 times compared with the current nRF70 Series Wi-Fi products. The chips combine Wi-Fi, Bluetooth Low Energy (LE), Thread and edge AI on a single device and are now sampling with selected customers.

Axon NPU runs machine learning models 10x to 17x faster than the Cortex-M33 alone

The integrated Axon Neural Processing Unit runs inference 10x to 17x faster than the 256 MHz Arm Cortex-M33 processor alone. Nordic says keyword spotting, anomaly detection and image classification become practical on battery-powered devices, with no external AI processor required. Developers can also run ultra-small models built from their own data with Nordic's Neuton software on the processor alone. Over-the-air model updates run through nRF Cloud.

MUNICH – November 03, 2026 -- Longsys (301308.SZ/09976.HK) will unveil its first industrial-grade, wide-temperature mSSD at electronica 2026, running November 10-13 in Munich, as demand for storage built for edge AI grows in automotive and industrial applications.

SiP design removes SMT assembly and solder-joint failure risk

The mSSD is built on system-in-package (SiP) technology and integrates the controller, flash memory and passive components into a single package. That eliminates surface-mount technology (SMT) assembly, cutting material use and manufacturing process requirements. Without the solder joints of conventional PCB assembly, the design also removes the associated risk of solder-joint failures. Longsys cites water resistance, dust protection and shock resistance as the basis for its industrial-grade reliability.

ROCKVILLE – – November 01, 2026 -- Ceva (NASDAQ: CEVA) has secured the first production deployment of its NeuPro NPU family, with Seoul-based Mobilint integrating the NeuPro-Nano NPN32 NPU and SensPro AI DSP into its REGULUS edge AI system-on-chip (SoC).

REGULUS delivers 10 TOPS at under 3W with up to 8GB of memory

REGULUS is a standalone processor for on-device AI. It targets power- and space-constrained applications including robotics, drones, machine vision, smart cameras and IoT devices.

Multi-engine design splits sensing and inference workloads

SensPro handles programmable sensing and signal-processing workloads. NeuPro-Nano handles always-on neural inference. Both sit alongside Mobilint's proprietary AI processing architecture, so each workload runs on the engine best suited to it.

WATERLOO – November 02, 2026 -- Applied Brain Research (ABR) has made its speech AI software development kit generally available on Zephyr RTOS for Alif Semiconductor's Ensemble family of edge AI processors. On the Ensemble E8, its Niagara ASR engine delivers 133 ms average latency from first audio to first token while drawing 15 mW of active power above baseline.

SDK bundles streaming recognition and synthesis that run without a cloud connection

The kit combines two engines. Niagara ASR provides full-vocabulary streaming speech recognition with low latency, transcribing speech continuously as it is spoken. Nith TTS provides low-latency streaming speech synthesis.

The SDK also offers native Zephyr RTOS support for battery-powered and air-gapped products, plus ongoing model updates and expansion.

Rockville – October 29, 2026 -- Ceva (NASDAQ: CEVA) has unveiled a next-generation NeuPro NPU portfolio that delivers up to 2X higher AI throughput per mm2 and 50% better energy efficiency than the first generation, and extends the company's neural processing IP into cloud and data center silicon for the first time.

Four NPUs share one architecture and one SDK

The portfolio comprises NeuPro Nano, Prime, Max and Ultra, all built on a common core architecture and a single unified software development kit. Chip designers can move between cores as requirements change, or combine several within one SoC and assign each workload to the best-suited engine. Ceva says this preserves software investment across product generations.

Global – October 29, 2026 -- Small language models under 10 billion parameters are starting to run on GPU/NPU-powered edge hardware with as little as 4GB of RAM, but most deployed IoT devices still depend on the lightweight compute inside the cellular modem SoC, according to Telit Cinterion.

Cellular modules host inference on the same chip that manages the network

Edge AI runs the model on the device itself. The device responds with near-zero latency, consumes far less bandwidth and keeps working if the network connection drops. Cloud-dependent products send a sensor value to a server and wait for the result, which adds latency and network dependence.

In many IoT products, the cellular module is the edge. The chip that handles the radio and network connection can also run a small model. Because the module reads its own signal quality, a model can monitor network availability and optimize reliability dynamically.

Taipei – October 13, 2026 -- Gemtek Technology, Qualcomm Technologies and gFiber are targeting commercial deployment in 2027 for an AI-native Wi-Fi 8 broadband gateway, with gFiber set to be the first operator to roll it out. Gemtek is showing the device at Network X 2026.

Gateway runs on Qualcomm's Dragonwing NPro A8 Elite platform

Gemtek engineered the gateway on the Qualcomm Dragonwing NPro A8 Elite Platform. It combines Wi-Fi 8 Quad-Band connectivity with Edge AI computing. Software comes from Genix Networks, a Gemtek subsidiary.

The device is designed as a home internet hub and a local platform for agentic AI. It supports network optimization, device awareness, security and proactive management.