New frontiers of semiconductor and AI innovation

The semiconductor industry stands at a fascinating juncture. For decades, its trajectory was famously charted by Intel co-founder Gordon Moore's 1965 prediction that the number of transistors on an integrated circuit (IC) would double approximately every two years. Moore's Law guided the industry to deliver ever smaller, faster, and cheaper chips, transforming our infrastructure, our homes, and even the mobile devices in our hands.

However, as the industry continues its quest to shrink transistors beyond the sub one nanometer (nm) range, it faces unprecedented challenges. Hurdles such as skyrocketing manufacturing costs are reshaping the roadmap for chipmakers, leading to deviations from Moore’s Law.

One of the most transformative developments in recent semiconductor history has been the advancement of ASML’s Extreme ultraviolet (EUV) lithography technology. By using light with a wavelength of 13.5nm, EUV machines have significantly advanced the field by enabling precise patterning of features at sub-7nm scales.

Looking forward, the semiconductor industry is addressing challenges in miniaturisation by adopting new technologies and strategies. These approaches involve examining different architectures and creating new testing methods to enhance yields (increase the number of functional chips produced from each wafer), addressing limitations imposed by physics and economics.

Chip manufacturing is increasingly complex

(Number of process steps for each node)

While much of the spotlight in semiconductor innovation has focused on manufacturing breakthroughs and high-performance computing, a growing wave of advancement is also taking place at the application layer – particularly in how and where AI is deployed. Increasingly, innovation is being driven not just by how chips are made, but by how they are used, with so-called ‘edge’ AI emerging as a compelling frontier. This shift reflects a broader trend: as devices become more intelligent and autonomous, the demand for localised, real-time processing is accelerating, opening up new opportunities for performance, efficiency, and user experience beyond the cloud.

One of the key challenges facing edge AI today is the lack of a definitive “killer app.” While smartphone makers have introduced useful features — such as Samsung’s Circle to Search and live translation – the market has yet to see a standout application that drives widespread adoption.

But given the fact that the transition from 3G to 4G took six years, while the shift from 4G to 5G occurred within four years, the direction of travel for edge AI looks clear.