Powering the Future with Renesas Next-Gen GaN and AI-Ready MCUs
When it comes to driving next-generation data centres, AI engines, and energy infrastructure, power conversion needs to be fast, compact and outrageously efficient. Renesas Electronics is stepping up to the plate with a powerful new suite of high-voltage GaN FETs built on its SuperGaN platform, backed by cutting-edge AI-ready microcontrollers (MCUs). The result? A seamless fusion of ultra-efficient energy handling and intelligent processing that sets a new benchmark for high-performance design.
At the heart of this leap forward is Renesas’ newly launched 650V Gen IV Plus GaN FET lineup, designed specifically for multi-kilowatt applications across AI data centres, UPS systems, electric vehicle charging, and renewable energy infrastructure. These devices are no minor tweak. They’re a ground-up rework of GaN switching tech, delivering better thermal performance, reduced switching losses, and dramatically improved power density.
According to Primit Parikh, Vice President of the GaN Business Division at Renesas: “The rollout of Gen IV Plus GaN devices marks the first major new product milestone since Renesas’ acquisition of Transphorm last year.”
These FETs, including the TP65H030G4PRS, TP65H030G4PWS, and TP65H030G4PQS, feature a silicon-compatible gate drive input, allowing developers to integrate them into systems without needing exotic new driver circuits. This is a boon for engineers chasing performance without compromising on design simplicity.
Efficiency, Scalability, Flexibility
Built on a die that’s 14% smaller than its predecessor, the Gen IV Plus devices achieve a lower RDS(on) of just 30 milliohms and offer a 20% improvement in the figure-of-merit (FOM) for output capacitance and on-resistance. This translates to lower power losses, smaller footprints, and, ultimately, cooler systems.
Available in TO-247, TOLL, and TOLT packages, the new FETs offer scalable options for systems ranging from 1kW to 10kW or more. With bottom- and top-side cooling paths, these packages enable straightforward thermal optimisation and device paralleling for high-current designs. What’s more, the silicon MOSFET input stage means they’re easily driven using existing gate driver ICs.
These aren’t just theoretical improvements. Renesas has already shipped over 20 million GaN devices, with more than 300 billion hours of collective field usage. It’s not just about new tech—it’s about proven performance.
AI Meets Edge Efficiency
While the GaN FETs redefine power switching, Renesas is doubling down on edge intelligence with the launch of the RA8P1 MCU family. Featuring Arm’s 1GHz Cortex-M85 and 250MHz Cortex-M33 cores, and paired with the Ethos-U55 NPU, the RA8P1 pushes the boundaries of what a microcontroller can do.
With performance hitting over 7300 CoreMarks and AI acceleration of 256 GOPS at 500 MHz, the RA8P1 is built for serious edge and endpoint AI work. That means running convolutional and recurrent neural networks for applications like smart surveillance, machine vision, and industrial automation—without the need for cloud compute.
Daryl Khoo, Vice President of Embedded Processing Marketing at Renesas, explained: “There is explosive growth in demand for high-performance edge AIoT applications. We are thrilled to introduce what we believe are the best MCUs to address this trend.”
Built for Tomorrow’s AI Workloads
Renesas built the RA8P1 on TSMC’s advanced 22ULL process, offering ultra-low leakage and enabling the use of embedded MRAM. Compared to traditional Flash, MRAM gives faster writes, better endurance, and longer retention—perfect for AI model weights and high-speed code execution.
The microcontrollers also feature:
- Dual on-chip memory banks: 2MB SRAM and 1MB MRAM
- SIP options with up to 8MB external Flash
- Integrated graphics and camera interfaces for vision AI
- Advanced cryptographic security including TrustZone and Root-of-Trust features
Chien-Hsin Lee from TSMC added: “It is gratifying to see Renesas harness the performance and reliability of TSMC 22ULL embedded MRAM technology to deliver outstanding results for its RA8P1 devices.”
All-in-One Development Ecosystem
Renesas has gone the extra mile to streamline development with its RUHMI (Renesas Unified Heterogenous Model Integration) AI framework. Compatible with TensorFlow Lite, PyTorch, and ONNX, RUHMI handles everything from model quantisation to graph compilation, offering APIs and ready-to-use application examples tailored to the RA8P1 architecture.
With seamless integration into e2 Studio IDE, developers can work within a unified environment for both MCUs and MPUs. Pair that with out-of-the-box software support for FreeRTOS, Azure RTOS, and Zephyr, and you’ve got a robust, flexible toolkit ready for real-world deployment.
Real-World Solutions and Winning Combinations
Renesas isn’t just pushing silicon—it’s delivering full system solutions. From AI-enabled surveillance cameras to autonomous robotic arms and smart conferencing equipment, Renesas’ Winning Combinations give engineers validated reference designs that help slash time to market.
More than 400 of these solutions are already available, offering pre-integrated combinations of Renesas’ power, analogue, MCU, and connectivity products. This design ecosystem ensures customers aren’t reinventing the wheel with every product iteration.
Designed for Scale and Security
Security is paramount in today’s connected systems. The RA8P1 MCUs come equipped with the latest RSIP-E50D cryptographic engine, providing everything from AES and RSA to Ed25519 and SHA3. With support for decryption-on-the-fly via XSPI interfaces, the devices can securely load encrypted firmware directly from external Flash.
Add in tamper detection, secure boot, and device lifecycle management, and you’ve got a fortified platform that meets the toughest IoT and industrial security requirements.
High Voltage. Higher Intelligence.
Renesas has clearly thrown its weight behind two of the biggest growth engines in modern tech: efficient power conversion and edge AI. With its Gen IV Plus SuperGaN FETs, developers now have access to higher power density, broader thermal options, and backward-compatible silicon-level integration. Meanwhile, the RA8P1 microcontrollers bring blistering AI performance to the edge, wrapped in a secure, development-friendly ecosystem.
Whether it’s managing megawatts in an AI server rack or orchestrating intelligent automation in a factory, Renesas has positioned itself as the enabler of smarter, smaller, and more efficient systems.
