The silicon wafer industry entered 2026 with a striking contrast between its two core markets. While demand for high-performance wafers surged—fueled by AI and high-bandwidth memory (HBM) applications—legacy semiconductor segments remained stuck in a cautious recovery phase. According to the latest annual report from the SEMI Silicon Manufacturers Group, worldwide wafer shipments grew by 5.8% in 2025, reaching 12,973 million square inches. Yet revenue for these critical substrates fell by 1.2% to $11.4 billion, reflecting deeper structural shifts in how the semiconductor supply chain is evolving.
At the heart of the divergence is the relentless push into advanced logic and memory fabrication. Epitaxial wafers—essential for sub-3 nm processes—and polished wafers for HBM stacks saw robust demand, as data centers and AI workloads prioritize performance over cost. The adoption of 300 mm wafers in these segments remains particularly strong, with manufacturers emphasizing tighter tolerances for wafer quality to support next-generation chips.
Key specs and trends
- Shipment growth: 5.8% YoY (12,973 MSI in 2025)
- Revenue decline: 1.2% YoY ($11.4 billion)
- Drivers: AI-driven demand for sub-3 nm logic and HBM4 wafers
- Legacy lag: Automotive, industrial, and consumer electronics wafers show gradual normalization but slower recovery
- Material focus: 300 mm wafers dominate advanced applications; legacy nodes still rely on older diameters
The market’s two-speed dynamic underscores a broader industry transition. While AI and high-bandwidth memory applications are accelerating investments in wafer innovation—pushing for ever-thinner, more precise substrates—traditional semiconductor sectors are still digesting years of overproduction. Automotive, industrial, and consumer electronics, which rely on mature-node processes, have only begun to stabilize after inventory corrections, and their recovery remains sensitive to economic conditions.
For wafer manufacturers, the outlook hinges on balancing these competing priorities. Advanced-node customers, particularly those in AI and data center infrastructure, are willing to pay a premium for wafers that meet exacting specifications. Meanwhile, legacy segments continue to demand cost efficiency, creating a fragmented pricing environment. The result is a market where high-margin, high-tech wafers are outpacing lower-margin, volume-driven substrates in terms of growth.
Looking ahead, the trajectory of wafer demand will likely depend on how quickly AI and memory-intensive applications scale—while legacy sectors may take years to return to pre-pandemic levels. The split between cutting-edge and traditional wafer needs is not just a statistical footnote; it’s reshaping supply chains, investment strategies, and even the geographic distribution of semiconductor production.