As AI accelerators and high-bandwidth memory (HBM) demand ever-greater computing performance and bandwidth, the semiconductor industry is pursuing scaling along two complementary paths. At the transistor level, 3D gate-all-around (GAA) architectures and High-NA extreme ultraviolet (EUV) lithography are extending device scaling through finer patterning and improved imaging resolution. At the system level, as conventional transistor scaling becomes more complex and expensive, advanced 3D integration brings multiple compute and memory dies together to deliver greater performance and bandwidth within a single package.
This shift toward 3D integration is also driving significant changes in packaging technology. HBM roadmaps are advancing toward taller stacks, including 16-high configurations, increasing thermal, interconnect, and assembly-yield challenges. Hybrid bonding offers a pathway to denser integration, but places stringent demands on surface quality: copper-pad recess must be controlled within a few nanometres, dielectric roughness as low as 0.1–0.2nm RMS, and small particles or topographic defects can compromise bond integrity. Meeting these tolerances consistently, without compromising production speed, is becoming a critical challenge for manufacturers.
AI is not only driving demand for more advanced chips but is also increasingly being used to optimise semiconductor manufacturing processes. As fabs adopt AI-driven process control, trustworthy measurement data becomes essential, and the richer the data, the higher the potential for improved process control. Manufacturers need reliable, non-destructive measurements of surface height, feature depth, sidewall profiles and nanoscale roughness – all at speeds compatible with high-volume manufacturing. Dense, true-3D data of this kind can feed run-to-run and feed-forward control, support virtual metrology, and anchor hybrid metrology schemes in which optical or electron-beam measurements are calibrated against a 3D reference. In High-NA EUV, 3D resist profile and roughness data can be used to calibrate and validate stochastic resist models, linking measured line-edge and line-width roughness to patterning defectivity.
Traditional atomic force microscopy (AFM) delivers detailed 3D surface measurements but has historically been limited by throughput. This creates an opportunity for faster probe metrology, with Infinitesima developing technology to address this and bridge the metrology gap. Its Metron3D is a hybrid-optical AFM that enhances AFM capability through 3-axis interferometry and photothermal probe actuation, achieving measurement speeds 10-100x faster than conventional AFM. The fully automated 300mm system is designed to provide true 3D information, non-destructively and at speeds required in high-volume manufacturing. Such capabilities could help manufacturers generate the detailed measurement data needed to support increasingly sophisticated, AI-driven process control.
The growing importance of these enabling technologies also creates an opportunity for the UK to translate its research and engineering strengths into solutions that address some of the semiconductor industry’s most critical manufacturing challenges.
The UK is strengthening its semiconductor position through its National Semiconductor Strategy and 2026 AI Hardware Plan, focusing on strengths in chip design, compound semiconductors, photonics, advanced materials, and research. Initiatives include funding for AI hardware innovation, the UK Semiconductor Centre and support for research commercialisation.
However, as next-generation chips demand greater precision, speed, and yield, the opportunity extends beyond chip design to the technologies that enable advanced manufacturing. Investment in metrology, process control, advanced packaging, pilot manufacturing, and equipment innovation could help bridge the gap between research and high-volume production and strengthen the UK’s position in the overall semiconductor supply chain. Greater collaboration between government, researchers, equipment suppliers, and semiconductor manufacturers could also help accelerate the commercialisation and adoption of UK-developed technologies.
What lies ahead is a more complex and more three-dimensional industry. High-NA lithography, gate-all-around devices, hybrid bonding, and stacked memory will all need measurement that keeps pace with the process. Techniques that can see atomic-scale 3D detail quickly and economically will ramp faster and help fab yield. As metrology moves from a quality check to a driver of yield, that capability will matter more.
