Key Takeaways

• Cortex‑M42 joins Arm’s Flexible Access and Total Access portfolios, bringing on‑chip ML and DSP to low‑power MCUs.
• Built on Armv8‑1‑M, it delivers up to 3× vector compute and 2× vector energy efficiency versus Cortex‑M33.
• Ideal for wearables, industrial sensors, and edge AI tasks such as keyword spotting, anomaly detection, and predictive maintenance.
• Arm’s licensing programs now support 319 Flexible‑Access members (500+ tape‑outs) and 60 Total‑Access partners (600+ projects).

Arm Expands Flexible & Total Access with Cortex‑M42

Arm has officially added the Cortex‑M42 microcontroller core to its two flagship licensing schemes—Arm Flexible Access and Arm Total Access. The move gives silicon partners a ready‑made pathway to embed machine‑learning (ML) and digital‑signal‑processing (DSP) functions directly inside ultra‑low‑power MCUs, without the cost and area penalties of a separate accelerator.

Why Cortex‑M42 Matters for Embedded AI

Cortex‑M42 marries Arm Helium (M‑Profile Vector Extension) with a compact AHB‑centric bus architecture. This combination enables a seamless upgrade path from legacy Cortex‑M4 or Cortex‑M33 designs while keeping silicon footprints under 1 mm² in a 28 nm process and power draw below 15 mW/MHz at typical workloads.

Key target use‑cases include:

• Audio front‑ends – keyword spotting and voice‑wake‑up in earbuds or smart watches.
• Sensor‑fusion – real‑time preprocessing for inertial, environmental, or vision sensors.
• Industrial edge – condition monitoring, anomaly detection, and predictive maintenance on motor drives or robotic joints (processing stays local, safety‑critical loops remain separate).

The core also supports distributed embedded processing, allowing each node in a sensor network to run lightweight inference without relying on a central server.

Performance & Efficiency Compared with Existing Arm Cores

Numbers are based on Arm’s reference silicon (28 nm) and represent typical operating points.

The custom instruction framework lets partners embed workload‑specific ops (e.g., fixed‑point MACs for audio codecs) directly into the pipeline, squeezing additional performance out of the same silicon budget.

Access Programs: Flexible vs. Total

Both programs now list Cortex‑M42 alongside other high‑end MCU cores (Cortex‑M52, Cortex‑M55, Cortex‑M85), the Ethos‑U85 NPU, the Corstone‑320 reference platform, and the Armv9 Edge AI software stack.

Real‑World Scenarios

• Wearable health monitor – The MCU can run a 10‑class keyword‑spotting model (≈8 KB) on‑device, waking the main radio only when a wake‑word is detected, extending battery life by >30 %.
• Smart motor controller – Local vibration analysis runs at 2 kS/s, flagging bearing wear before a failure, while the safety‑critical torque loop remains on a separate deterministic core.

Bottom Line

Arm’s inclusion of Cortex‑M42 in its Flexible and Total Access portfolios gives developers a low‑cost, low‑power MCU that can execute genuine AI and DSP workloads on the edge. With up to 3× the vector throughput and twice the energy efficiency of the widely deployed Cortex‑M33, the core fills a critical gap between traditional microcontrollers and heavyweight AI accelerators. The expanded licensing ecosystem—now supporting hundreds of tape‑outs and projects—ensures that partners can prototype quickly, scale confidently, and bring smarter, more responsive embedded products to market faster.