News
Arm FCSA and the Journey to Standardizing Open Chiplet-Based Design
Arm has donated its Foundation Chiplet System Architecture (FCSA) to the Open Compute Project (OCP), marking a significant step towards standardizing open chiplet-based designs. FCSA is an ISA-neutral initiative focused on addressing inter-chiplet communication challenges, particularly for AI-driven automotive and infrastructure applications. This move aims to enable a more modular and scalable approach to building complex, multi-die systems by defining standard interfaces like AMBA CHI C2C, allowing chiplets from various vendors to seamlessly interact.
Launches
ChipAgents AI Revolutionizes RTL Design and Verification with Autonomous Root Cause Analysis
ChipAgents AI is pioneering an AI-native approach to Electronic Design Automation (EDA), targeting a 10x productivity improvement in RTL design and verification. A recent demonstration highlighted the system’s capability to autonomously identify and resolve design bugs in a complex PCIe Gen 3 design using natural language commands. The system completed a task that typically takes engineers dozens of hours in approximately 20 minutes, signaling a transformative shift towards AI-augmented design processes.
- AI-native EDA targets 10x productivity boost in RTL design/verification.
- Autonomous root cause analysis resolves complex bugs in minutes via natural language.
- Demonstrates a significant leap towards AI-augmented design workflows.
Charts
Comparative Analysis of CFET and NSFET Architectures Reveals Performance and Area Gains
A recent study from TU Munich and Indian Institute of Technology presents a comparative analysis of Complementary Field-Effect Transistor (CFET) and Nanosheet FET (NSFET) architectures, focusing on the impact of aging, self-heating, and parasitics on standard cell performance. The research highlights significant advantages of CFETs in advanced process nodes.
- CFETs offer approximately 50% area savings at the standard cell level compared to NSFETs.
- They demonstrate a 42% improvement in inverter propagation delay.
- CFET SRAM cells provide area gain with faster and more stable write operations, indicating strong potential for high-performance applications.
Research
Study Highlights Compute Efficiency and Density of Integrated Photonic Computing Architectures
Researchers from IBM Research – Europe, University of Heidelberg, and University of Münster have published a technical paper on the performance metrics of integrated photonic computing. The study provides a microarchitecture-level analysis of three prominent photonic computing architectures, taking into account both the optical circuits and the essential electronic peripheral circuitry required for optoelectronic and analog-to-digital conversions. It also demonstrates a heterogeneous photonic-electronic in-memory processing approach suitable for low-latency neural network inference, offering crucial insights into design aspects for enhancing compute efficiency and density.
Insight
Agentic AI Set to Transform EDA, Experts to Discuss Infrastructure and Future
SemiWiki is hosting a pivotal webinar on December 4, 2025, titled “Is Agentic AI the Future for EDA — and What Does It Mean for EDA Infrastructure?”. Industry leaders from Cadence, NetApp, and AMD will convene to discuss how Agentic AI is poised to reshape chip design and verification. The session will cover the current state and future potential of Agentic AI, as well as the necessary infrastructure changes—including compute, storage, orchestration, and data management—to support AI-driven EDA workflows. The webinar promises practical guidance for semiconductor professionals navigating the shift towards autonomous, AI-augmented design processes.
- Webinar on Dec 4, 2025, addresses Agentic AI’s impact on EDA.
- Experts from Cadence, NetApp, and AMD will discuss current state, future, and infrastructure needs.
- Aims to provide practical guidance for adopting AI-driven design automation.
Stay tuned for more cutting-edge developments shaping the future of the semiconductor industry!