Semiconductors

The semiconductor industry designs and manufactures the chips that underpin essentially all modern electronics and computing, and it is in the middle of what may be its largest demand supercycle in history, driven overwhelmingly by the buildout of AI infrastructure. The industry's structure divides broadly into three roles: "fabless" companies that design chips but outsource manufacturing (Nvidia, AMD, Apple, Qualcomm), "foundries" that manufacture chips to other companies' designs (TSMC, and increasingly Samsung and Intel Foundry), and "integrated device manufacturers" that both design and manufacture their own chips (Intel, Samsung, and memory specialists like SK Hynix and Micron) — a structural split that shapes competitive dynamics across the entire sector.

The scale of the current growth is genuinely unusual even by an industry accustomed to cyclical booms: various forecasts put the global semiconductor market anywhere from roughly $975 billion to as much as $1.5 trillion for 2026, depending on methodology, with most estimates agreeing on year-over-year growth well above 50%, and some projections showing growth approaching 90%. AI-specific chip demand alone is estimated at around $500 billion for the year, and the memory segment — historically among the most volatile and commoditized parts of the industry — is projected to surge roughly 250% year over year to more than $800 billion, driven by soaring demand for the high-bandwidth memory used in AI accelerators. As one industry analysis put it, AI is no longer merely a demand catalyst for semiconductors — it has become the demand foundation the entire industry is now built around, with roughly half of total industry revenue in 2026 tied directly to AI data center chips.

That AI-driven boom has reshaped the industry's competitive geography in a few specific ways. TSMC has extended its lead in advanced logic manufacturing, now capturing a large majority of leading-edge chip production and commanding a premium for capacity that customers like Nvidia compete fiercely to secure. Samsung and SK Hynix have found a parallel, equally critical role supplying the advanced memory chips AI accelerators require, turning what was historically a lower-margin, highly cyclical memory business into one of the more strategically important segments of the entire chip supply chain. Intel, meanwhile, is attempting a multi-year turnaround of its own manufacturing capability (its 18A process) while competing simultaneously against AMD in CPUs and against Nvidia and AMD in AI accelerators — three separate competitive fights being waged at once from a company still working to prove its manufacturing can catch back up to TSMC's.

The industry's biggest structural risks are the flip side of its current strength: extreme geographic concentration (roughly 90% of the most advanced chips manufactured in Taiwan alone), a historical tendency toward brutal boom-bust cycles once capacity catches up with demand, and enormous capital intensity — building a single leading-edge fab now costs tens of billions of dollars, a bar so high that only a handful of companies worldwide can credibly compete at the frontier. Export controls on advanced chips and manufacturing equipment, particularly U.S. restrictions aimed at limiting China's access to cutting-edge AI chips, have also become a permanent feature of the industry's operating environment rather than a temporary disruption, forcing chip designers and manufacturers alike to navigate an increasingly fragmented, geopolitically shaped global market rather than a single unified one.