Scaling Below 10 Å: CFETs and the Next Frontier of Semiconductor Logic
The semiconductor industry is looking beyond the transistor to sustain Moore’s Law. New technologies like Self-Aligned CFET and advanced packaging are becoming the new frontier for AI-driven compute.
The semiconductor industry is facing a crossroads as traditional transistor scaling reaches its physical limits. At SEMICON West 2026, the conversation has shifted "beyond the transistor," focusing on advanced packaging, photonics, and Self-Aligned CFET (Complementary Field-Effect Transistor) technology. These innovations are projected to extend CMOS scaling beyond the 10 Ångström node, providing the density and performance required for the next generation of AI data centers.
One of the most promising developments is the CFET, which stacks n-type and p-type transistors vertically. This architecture delivers higher performance through larger effective channel widths and improved electrostatics, all while occupying a smaller footprint. However, hardware design is only half the battle. Engineers are now focusing on PVTW (Process, Voltage, Temperature, and Workload) characterization. This recognizes that two identical chips may behave differently based on the specific AI workload they are running, necessitating smarter, workload-aware silicon.
Furthermore, the U.S. government is doubling down on domestic production, with CHIPS Act funding totaling $389 million aimed at powering chips, vertical GaN materials, and Micron's memory roadmap. As NVIDIA expands its supply chain capacity and Tenstorrent raises $1.5 billion, the semiconductor landscape is being reshaped to support the insatiable appetite for AI compute.
Source: Semiconductor Engineering