Quectel Wireless Solutions, a global end-to-end IoT provider, today launched the FCM665D — a compact Wi-Fi 6 and Bluetooth module with on-board edge AI processing, built for the next generation of smart home, smart building, and industrial IoT devices. The announcement, made from the company's Belgrade, Serbia office on July 29, 2026, signals how deeply edge AI has penetrated the module-level IoT market.
The module's design philosophy: intelligence without the cloud tax
The FCM665D is designed for products that need to do more locally. Think smart lighting that adjusts based on voice commands without phoning home, video doorbells that recognize faces on-device, and central control hubs that process sensor data in real time. The module's aim is to cut reliance on constant cloud connectivity while delivering faster, more responsive user experiences.
"This module has the AI capabilities and processing power to compete at the highest level," said Delbert Sun, Vice General Manager of the Product Department at Quectel Wireless Solutions. "Smart home and IoT devices are required to do more locally at the edge, from recognizing voice commands to processing images in real time, and that requires genuine computing performance. The FCM665D gives our customers that power in a compact module that is easy to integrate and deploy."
Why edge AI matters more in 2026
The launch comes at a time when the smart building and smart home markets are accelerating at a pace few predicted. Juniper Research forecasts that industrial smart building deployments will climb from 26 million in 2025 to more than 165 million by 2030, pushed by the push to cut energy waste and automate building operations via sensor networks. The broader edge AI market is on a similar trajectory: Fortune Business Insights projects it will jump from $46.96 billion in 2026 to $445.75 billion by 2034, a compound annual growth rate of 32.5 percent.
That growth is pulling intelligence away from centralized data centers and into the devices themselves. The FCM665D is a direct response to that shift — it crams a triple-core ARMv8-M Star MCU running at up to 480 MHz, up to 16 MB of PSRAM, and 16 MB of flash into a package measuring just 31.4 mm × 18.0 mm × 2.15 mm. That is small enough to fit into a smart light switch, a video door lock, or a photovoltaic inverter.
Two SDKs, one module: voice and video on the same die
A distinguishing feature of the FCM665D is its dual-software development kit approach on a single hardware platform. The AI Development Kit (AIDK) handles voice processing — echo cancellation, noise suppression, voice activity detection, keyword spotting, and image recognition. The Audio & Video Development Kit (AVDK) handles multimedia tasks: audio and video processing, display output, and cloud integration.
This dual-SDK architecture matters because it lets manufacturers build a single product that can listen, see, and respond without needing separate voice and video chips. A video doorbell, for example, can run face recognition through the AIDK while streaming live video through the AVDK, all on one module.
The module's processor performance and large storage capacity also support multiple network connections and the local caching of long-term operating data for energy storage batteries and photovoltaic inverters. Its edge AI capabilities make it well suited to smart home applications such as intelligent lighting and central control systems.
Wi-Fi 6, Bluetooth 5.4, and Ethernet on a single module
On the connectivity side, the FCM665D supports single-band 2.4 GHz Wi-Fi 6 based on IEEE 802.11 b/g/n/ax, Classic Bluetooth 2.1, and Bluetooth Low Energy 5.4. It also supports Wi-Fi pairing via Bluetooth, which simplifies the setup process — a user taps their phone to the device, and the credentials transfer without fumbling through a Wi-Fi selection menu.
Beyond wireless, the module includes a high-performance Ethernet Media Access Controller (MAC) via a Reduced Media-Independent Interface (RMII), compliant with IEEE 802.3-2015. That makes it usable in wired network interface cards, data center bridges, and network nodes where reliable Ethernet is preferred over Wi-Fi.
Security is addressed through WPA-PSK, WPA2-PSK, and WPA3-SAE support, hardware encryption, and ARM TrustZone protection — covering the baseline requirements for IoT device security as regulators worldwide tighten standards.
Industrial-grade toughness in a consumer-sized footprint
Despite its small size, the module is built for harsh conditions: it operates from -40°C to +85°C, putting it squarely in the industrial IoT category. Quectel's QuecOpen framework provides up to 54 GPIO pins depending on memory configuration, giving hardware designers the flexibility to connect sensors, actuators, and displays without needing a separate microcontroller.
The FCM665D also supports local caching of long-term operating data, which Quectel specifically highlights for energy storage batteries and photovoltaic inverters — applications where devices must keep working even when the cloud connection drops.
What this means for the IoT market
The FCM665D enters a crowded but fast-growing market for edge AI modules. Companies like Qualcomm, MediaTek, and NXP have all launched competing products in the last year, but Quectel's strength is its end-to-end IoT portfolio — the company sells not just modules but antennas, GNSS receivers, satellite connectivity, and full ODM and system integration services. That breadth means a manufacturer can source nearly all its IoT components from one supplier, reducing integration risk.
For the broader IoT market, the arrival of sub-$20 modules with triple-core ARM processing, Wi-Fi 6, Bluetooth 5.4, and on-device AI marks a turning point. As recently as 2023, edge AI at this capability level required a separate AI accelerator chip or a high-end applications processor costing three to four times as much. The FCM665D collapses that cost and complexity into a single module.
The smart home market will likely feel the impact first. Voice-controlled lighting, video doorbells with on-device facial recognition, and central control hubs that coordinate Matter, Thread, and Zigbee devices are all addressable with this module. Industrial applications — predictive maintenance sensors, building management controllers, and photovoltaic system monitors — will follow as manufacturers design the FCM665D into their next-generation products.
Quectel, headquartered in Shanghai with offices in Belgrade and regional teams across the globe, employs over 8,900 professionals dedicated to IoT solutions spanning cellular, GNSS, satellite, Wi-Fi, Bluetooth modules, and antennas. The company's full portfolio approach positions it to capture a strong share of the edge AI IoT module market as demand accelerates through 2030.
The FCM665D also arrives at a moment when the edge AI software market itself is surging. A July 2026 report from market analysts pegged the Edge AI Software Market at $3.12 billion in 2026, with a compound annual growth rate of 24.63 percent projected through the end of the decade. Much of that growth is tied to the kinds of applications the FCM665D is designed to power: real-time automation, predictive maintenance, computer vision, and privacy-first analytics that keep data processing on the device rather than sending it to the cloud.
For manufacturers evaluating entry into edge AI IoT, the FCM665D lowers the barrier substantially. The module's QuecOpen environment ships with pre-integrated software stacks for both AI and audio-video development, which means a firm with limited embedded AI expertise can still ship a product with voice control or on-device image recognition. The 50-54 GPIO pins and the wide temperature range also mean the same module can be used across consumer and industrial product lines, reducing the number of SKUs a company needs to manage.
One area where the FCM665D could have particular impact is in the retrofit market for smart buildings. Many older commercial buildings lack the wired infrastructure for modern sensor networks, and running new cable is expensive. A battery-powered sensor node built around the FCM665D could use Wi-Fi 6's Target Wake Time feature to extend battery life to years, while running local AI inference to decide what data is worth transmitting. That combination — low power, local intelligence, and wireless connectivity — is exactly what the building retrofit market needs to scale from 26 million to 165 million deployments by 2030.
For more on IoT and connected devices, check out our IoT coverage and the latest on semiconductor advances that make edge computing possible. Read the original announcement on Business Wire and Quectel's product page for full specifications.