Qualcomm Splits One IoT Quad-Core Die Across Consumer and Industrial Lines
The Q-2390 and IQ-2390 share a Cortex-A78/A55 complex, an Adreno 704 GPU, and an optional RISC-V real-time core, with the industrial part introducing the IQ2 Series.
Qualcomm Technologies recently announced the Dragonwing Q-2390 and IQ-2390 processors, ahead of IFA 2026 running Sept 4-8 in Berlin, adding two quad-core Arm parts to its IoT portfolio for consumer, commercial, and industrial edge devices.

The Dragonwing Q-2390 and IQ-2390 IoT processors are targeted for consumer, commercial, and industrial edge designs.
The IQ-2390 is the first product in a new Dragonwing IQ2 Series aimed at the industrial edge, and underneath the two product names sits a single silicon design, binned and packaged for two very different customer sets. System-on-modules from four partners are already in an early access program, and evaluation kits for both platforms are expected in the first quarter of 2027.
One Die, Two Product Families
Both processors run one Arm Cortex-A78 core alongside three Cortex-A55 cores, clocked at up to 1.5 GHz or 1.9 GHz depending on the SKU, with 512 KB of L3 cache. They differ in packaging, temperature, and options, though.
The Q-2390 comes in a 12 mm x 12 mm x 0.92 mm BGA on a 0.4mm ball pitch, which suits the dense, cost-sensitive boards found in retail terminals and appliances. The IQ-2390 uses a 21.3 mm x 18.4 mm x 1.82 mm package on a 0.8 mm pitch, a coarser geometry that tolerates cheaper PCB stackups and survives the thermal cycling and vibration of industrial deployment.

Q-2390 block diagram
Q-2390 AA parts are rated from -30°C to 95°C, while Q-2390 AB parts and the IQ-2390 stretch to 115°C. Only AB parts get the Hexagon NPU, the optional SiFive E61 RISC-V real-time core running at up to 600 MHz with 768 KB of memory, and the Hexagon V66 audio DSP, while the S and M suffixes on Q-2390 part numbers separate parts without cellular from parts with 4G LTE Cat 4 and GNSS. The IQ-2390 comes as a single AB variant with neither modem nor GNSS, on the assumption that a factory controller is wired.

IQ-2390 block diagram
Qualcomm has committed to supplying both parts until 2036, matching the decade-plus service lives typical of industrial equipment rather than consumer product cycles. More information can be found in the product briefs for the Q-2390 and IQ-2390.
Compute, Vision, and Connectivity
The same Adreno 704 GPU runs at up to 1.1 GHz in both parts, an Adreno 921 display processing unit drives a 4-lane MIPI DSI output at up to 1080p60, and the video engine handles 1080p30 encode and decode for H.264 and H.265, with VP9 decode.
Camera support covers up to two 4-lane MIPI CSI interfaces, feeding either a 13 MP dual-camera or a 25 MP single-camera configuration at 30 fps. The NPU is rated at up to 1.1 TOPS, which places these parts well below the 50 TOPS class of edge accelerators and squarely in the territory of classification, detection, and keyword spotting rather than generative workloads.
Memory is dual-channel 16-bit LPDDR4x at up to 1.8 GHz, with eMMC 5.1 and SD 3.0 for storage. Both parts carry two Gigabit Ethernet controllers with Time-Sensitive Networking, alongside Wi-Fi 5 and Bluetooth 5.0 or 5.1 through companion WCN3950 and WCN398x devices.
TSN lets a controller share a network segment with motion and safety traffic without the jitter ordinary Ethernet introduces, and it's the feature that separates this class of part from a general-purpose applications processor. Expansion runs through PCIe 2.0 x1, USB 3.1 and USB 2.0 Type-C, 10 QUP instances, and 155 GPIO.
Qualcomm supports Android, Yocto Linux, and Ubuntu on the application cores, and Zephyr RTOS on the RISC-V real-time core where present.
Modules, Partners, and Availability
Jeff Arnold, VP and GM of auto telematics and industrial and embedded IoT consumer and connectivity at Qualcomm Technologies, named retail systems, smart appliances, consumer robots, industrial controllers, machine vision systems, and critical infrastructure as targets, and said cost, complexity, and integration barriers are what's kept AI out of many of those categories so far.
Qualcomm listed retail point-of-sale systems, kiosks, access control, smart appliances, smart agriculture, home robots, fitness equipment, and enterprise terminals for the Q-2390, and industrial automation, oil and gas, utilities, and energy systems for the IQ-2390.
All images used courtesy of Qualcomm.