NXP Unveils the i.MX 93W, Replacing "Up to 60 Discrete Components" in Physical AI Designs

NXP Semiconductors has announced a new entry in its i.MX family, created by combining the i.MX 93 applications processor with an IW610G triple-radio — to build the i.MX 93W, an all-in-one chip the company claims can replace "up to 60 discrete components" in on-device edge machine learning (ML) devices.

"With the i.MX 93W, we're extending the i.MX 9 family to help customers scale physical AI [Artificial Intelligence] faster," claims NXP's Charles Dachs of the new part, unveiled at Embedded World 2026 in Nuremberg today. "This new platform simplifies integration of AI and secure wireless connectivity, reducing design complexity and allowing customers to more quickly deploy AI agents at the edge."

The i.MX 93W isn't, strictly speaking, an entirely new product. Rather, it's the combination of two previously-distinct products: the i.MX 93 application-class system-on-chip and the IW610G dual-band Wi-Fi 6, Bluetooth 5.4 Low Energy (BLE), and Thread- and Matter-compatible IEEE 802.15.4 "tri-radio," plus the components required to make the latter work. As a result, NXP claims it can replace "up to 60 discrete components."

The chip itself includes either one or two Arm Cortex-A55 cores running at up to 1.7GHz, a Cortex-M33 microcontroller coprocessor running at up to 250MHz, and an Ethos U-65 "micro-NPU," a neural coprocessor offering performance equivalent to a claimed 1.8 tera-operations per second (eTOPS) and designed to accelerate artificial intelligence and machine learning applications running on-device.

While attendees at Embedded World 2026 in Nuremberg this week can see the new i.MX 93W in action at Booth 4A-222, the rest of the world will need a little patience: NXP is targeting sampling in the second half of 2026, with no word yet on pricing and general availability. More information is available on the NXP website.