The Chiplet Revolution: Unifying Silicon Connectivity with AMBA C2C
The semiconductor industry is moving toward converged chiplet connectivity with the AMBA C2C foundation. Scalable protocols and flexible transport support are essential for the next generation of 3D-stacked AI chips.
The era of the monolithic system-on-chip (SoC) is giving way to the chiplet revolution, and the semiconductor industry is unifying around a new standard to manage the complexity. The AMBA C2C (Chip-to-Chip) foundation is emerging as the converged framework for interconnecting disparate silicon dies, extending proven protocols like AXI and CHI to the chiplet ecosystem. This standardization is critical for building the high-performance processors required for AI and autonomous systems.
As AI models grow in complexity, the physical limits of a single die become a bottleneck. By breaking the processor into specialized chiplets—one for compute, one for memory, one for I/O—designers can optimize each component for its specific function. However, these chiplets must communicate with extremely low latency and high bandwidth to function as a single unit. AMBA C2C provides the scalable protocol subsets and flexible transport support needed to ensure interoperability across different vendors and manufacturing nodes.
This move toward an open, standardized chiplet architecture allows for greater customization. A car manufacturer, for instance, could combine an off-the-shelf AI accelerator chiplet with a custom-designed safety controller chiplet, all connected via a unified AMBA fabric. This modularity will accelerate the development of specialized silicon for the edge, making high-performance compute more accessible and cost-effective across the industry.
Source: Semiconductor Engineering
