The future of embedded IDEs is being redefined.

If we were to go back more than a decade, we could never have imagined that today's embedded MCUs/MPUs have evolved to be so complex. They are no longer just single ICs but complex ecosystems that include heterogeneous cores, accelerators, DSPs, and dedicated domains that must work in harmony. With the rise of embedded AI, integrating NPUs has become the norm.

As chips become more complex and embedded AI rises, embedded IDEs (Integrated Development Environments) are facing significant challenges: the tools that developers rely on are lagging behind the complexity of modern hardware.

Vendors have recognized this issue. To enable developers to fully unleash the performance potential of hardware and to allow every developer to embrace the trend of embedded AI, vendors are increasing their investment in software and IDEs.

The Dilemma of Embedded IDEs

In fact, the current embedded development process is undergoing a significant transformation:

First, as more and more functions are integrated on - chip, developers need to coordinate across boundaries, debug across architectures, and at the same time ensure the deterministic performance of the system, and adapt to the different instruction sets, memory spaces, and toolchains of multiple cores.

Second, the development of edge AI/ML requires developers to connect the entire process of model training (such as PyTorch, TensorFlow), embedded deployment (quantization, optimization, hardware mapping), and code generation, bridging the gap between data science and embedded development.

Finally, security has become a hard requirement. Designs must comply with standards such as the IEC 62443 and the EU Cyber Resilience Act from the very beginning. Functions such as a trusted execution environment (TEE), secure boot, and a cryptographic root of trust must be deeply integrated rather than added later.

These changes have put traditional IDEs in a difficult situation.

First, the current IDE market is fragmented. Developers often need to operate multiple vendor - specific environments simultaneously to start a system. Debugging complex issues with DSPs and MCUs can take weeks. When an AI model cannot be adapted to resource - constrained hardware, the relevant workflow comes to a halt.

Second, traditional IDEs struggle to meet the innovation needs of edge AI. Few IDEs are natively compatible with AI workflows, forcing developers to manually integrate scripts and frameworks.

Third, security integration is weak. Security, reproducibility, and automation are now essential requirements rather than options. Secure boot and over - the - air (OTA) updates are regarded as independent software development kits (SDKs) rather than part of the core workflow. If developers continue to rely on traditional IDEs designed for simple systems, the pace of innovation in the edge field will slow down.

Fourth, there is a fragmentation problem with embedded IDEs. Toolchains are tied to a single vendor, and multi - core systems usually require multiple IDEs.

Fifth, there are issues with the development experience. Some IDE interfaces are inconsistent and outdated in design, leading to errors during development. This ultimately results in wasted time, redundant work, and hinders the innovation process.

Vendors' Continuous Layout of IDEs

Specialized IDEs are a major category of embedded IDEs. They are IDEs launched by vendors for their own products. Most MCU vendors launch their own matching IDEs and continuously expand their toolchains. These IDEs are fully optimized for their own MCUs/MPUs, capable of maximizing hardware performance. They also cooperate with general - purpose IDEs such as IAR and Segger to integrate all functions.

These tools are generally free, and developers usually don't need to apply for a license when using the vendors' MCUs/MPUs. It can be said that this is both a promotion of the vendors' products and a form of after - sales service.

AI is undoubtedly a key area for MCU vendors. These vendors are continuously launching their own IDE products to simplify and accelerate the development of embedded AI.

Recently, ADI launched CodeFusion Studio 2.0 to embrace the era of physical AI. This tool introduces advanced hardware abstraction, seamless AI integration, and powerful automation tools to simplify the process from concept to deployment for ADI's various processors and microcontrollers.

What can CodeFusion Studio 2.0 offer developers?

First, it provides developers with an end - to - end AI workflow. The latest platform is based on Microsoft's Visual Studio Code and has a built - in model compatibility checker, performance analysis tools, and optimization functions. Developers can import models from TensorFlow or PyTorch and generate inference - ready code within minutes. With the Zephyr AI Profiler, they can monitor latency and memory without accessing the hardware. In addition to inference, the platform also supports AutoML for Embedded, enabling dataset training and optimization within the same workflow.

Second, it unifies the development experience. The updated CodeFusion Studio System Planner now supports multi - core applications and extended device compatibility. The unified configuration tool reduces the complexity of the ADI hardware ecosystem. Developers benefit from integrated debugging functions, including core dump analysis and GDB (GNU Debugger) support, making troubleshooting faster and more intuitive.

Finally, it ensures the security of the digital boundary. With ADI's Trusted Edge Security Architecture (TESA), developers can incorporate secure boot, TrustZone partitioning, and cryptographic protocols as part of the standard workflow. This is important because PI agents are reasoning and controlling physical systems. This control must be secure, deterministic, and auditable. CodeFusion Studio ensures that every step from model deployment to firmware update is protected.

EEWORLD also interviewed Jason Griffin, the head of software products and tools at ADI's Software and Security Division, about Code Fusion Studio. When asked about "how AI can meet the requirements of embedded applications in resource - constrained hardware," he said that Code Fusion Studio verifies whether an AI model meets hardware constraints (flash memory, RAM, operator support) by providing compatibility and performance analysis reports. It also integrates model optimization tools such as quantization and pruning suggestions and supports lightweight AI frameworks (such as TensorFlow Lite for microcontrollers). These functions help developers adapt AI models to the limited resource environment of embedded systems.

Infineon recently launched a new integrated development environment (IDE) - AURIX Configuration Studio (ACS), aiming to simplify the application development process using the AURIX TC3x series of devices, speed up product launch, and reduce development costs. ACS is built on the mature DAVE (Digital Application Virtual Engineer) technology and integrates an Eclipse - based editor, a GNU C compiler, and an open - source debugger.

The highlights of this IDE include an intuitive graphical user interface (GUI), automated resource management, and code generation. Specifically, ACS's GUI allows users to configure and customize projects in an intuitive way, significantly reducing development complexity compared to traditional development methods. The framework automates resource management through an AI - driven solver, which can automatically allocate and maintain hardware resources, freeing developers from time - consuming and repetitive tasks. The tools in ACS automatically generate high - quality, production - ready code based on the interface settings, greatly reducing manual coding errors.

As an embedded engineer, one cannot avoid STM32. STM32CubeIDE for VS code was released as a Release version in mid - October, completing a major upgrade from V2.x to V3.x.

The new VS Code extension removes the dependency on STM32CubeCLT and instead introduces the STM32Cube bundles manager for automatic plugin management. This tool can automatically download, install, and update CLI tools and STM32 device support files. Developers no longer need to separately download and install the STM32CubeCLT package, manually install the Cmake tool, or manually set the tool path. They can complete the installation with one click and use the latest compiler or get support for the latest STM32 devices.

STM32CubeIDE for VS Code is a "STM32 - specific development environment" built on a VS Code extension. It is positioned as the next - generation free IDE for the STM32Cube ecosystem and is deeply integrated into the STM32 development system from the very beginning. As a part of the STM32Cube ecosystem, STM32CubeIDE for VS Code can seamlessly connect with STM32Cube ecosystem tools.

STM32CubeIDE for VS Code supports all platforms including Windows, Linux, and macOS, and supports all STM32 MCU product series through CMSIS - PACKs.

Renesas has been closely following the trend of embedded AI. As early as 2023, it announced the establishment of an interface between its Reality AI Tools and the e2 studio integrated development environment, enabling designers to seamlessly share data, projects, and AI code modules between the two programs. The real - time data processing module has been integrated into Renesas' MCU software development tool suite to facilitate data collection from Renesas' own tool suite or customer hardware using Renesas MCUs. This integration will shorten the design cycle of Internet of Things (IoT) edge and terminal artificial intelligence (AI) and Tiny Machine Learning (Tiny ML) applications.

Since acquiring Reality AI in 2022, Renesas has been committed to researching, improving, and simplifying AI design. Reality AI Tools, as a software environment built to support the development of complete AI application products, allows users to automatically explore sensor data and generate optimized models. Regarding Renesas' IDE, e2 studio, Renesas still maintains a quarterly upgrade frequency and continuously iterates.

The Future of IDEs

In addition to helping developers with embedded AI development, vendors are also actively using AI technology to optimize the IDE development experience.

On November 5th, Microsoft published a blog post announcing the latest AI roadmap for its integrated development environment (IDE), Visual Studio. It clearly defined the current and future core of its work - to create an "AI - driven agent experience" and provide developers with more intelligent, faster, and more intuitive programming tools.

Microsoft's roadmap focuses on four main directions. First, it will launch various new agents for customization, testing, and debugging, and support concurrent operation of these agents. Second, it will improve the chat function based on community feedback, such as introducing slash commands. Third, it will fully implement the MCP specification, allowing enterprises to set a server whitelist to enhance security. Fourth, it will integrate the latest models such as GPT - 5 Codex and provide an automatic model selection function.

Actually, many embedded engineers are still used to their most familiar software environments. Some engineers said, "When AI meets embedded systems, Keil is still the hidden king." Of course, some engineers believe that it is better to directly use the IDEs provided by vendors than to spend a lot of time and effort building their own IDEs.

Everyone has their own optimal solution. But for MCU/MPU vendors, how to save customers' precious time and enable them to quickly embrace embedded AI must be the key to enhancing the competitiveness of their products.

References

[1]ADI: https://developer.analog.com/newsroom/rethinking-ides-the-future-of-embedded-and-ai-development

[2]ADI: https://developer.analog.com/newsroom/analog-devices-unveils-codefusion-studio-20-to-simplify-and-accelerate-embedded-ai-development

[3]ADI: https://developer.analog.com/newsroom/codefusion-studio-accelerating-deployment-of-physical-intelligence

[4]Infineon: https://www.infineon.com/market-news/2025/INFATV202510-013

[5]ST Chinese Forum: https://shequ.stmicroelectronics.cn/thread-868922-1-1.html

[6]Renesas: https://www.renesas.cn/zh/about/newsroom/renesas-extends-its-aiot-leadership-integration-reality-ai-tools-and-e-studio-ide

[7]Electronic Design: https://www.electronicdesign.com/technologies/embedded/media-gallery/21243757/electronic-design-the-top-11-ides-for-embedded-applications

This article is from the WeChat public account "EEworldbbs", author: Fu Bin, published by 36Kr with authorization.