The wafer foundry market is poised for a recovery in 2025, with TrendForce forecasting an annual growth rate of 20%, up from 16% in 2024.
This optimistic outlook comes despite ongoing weak demand in the consumer electronics sector, which has prompted component manufacturers to adopt a cautious stocking approach. As a result, the average capacity utilisation rate for wafer foundries has dropped below 80% in 2024. Only advanced processes, such as 5, 4, and 3nm nodes used in high-performance computing (HPC) products and premium smartphones, have been able to maintain full capacity utilisation – a trend expected to continue into 2025. However, visibility in the consumer market remains unclear for the coming year.
Conversely, the automotive and industrial control sectors have begun to show signs of recovery from inventory corrections, starting in the second half of 2024. Restocking in these sectors is anticipated to gain momentum in 2025. The increase in wafer consumption per unit, driven by developments in edge AI and the ongoing expansion of Cloud AI infrastructure, is expected to further support the wafer foundry market’s projected 20% growth in 2025.
TSMC positioned for above-average revenue growth
According to TrendForce, TSMC’s revenue growth rate in 2025 is expected to outpace the industry average, driven by advanced processes and packaging innovations. While other foundries remain constrained by weak consumer demand, non-TSMC foundries are projected to see revenue growth of nearly 12% in 2025, largely due to healthier component inventory levels among IDM and fabless clients across various industries, alongside strong demand from Cloud and Edge AI applications. A lower base in 2024 is also contributing to these growth expectations.
Advanced processes and packaging shaping the future
The 3nm process has been scaling up over the past two years and is expected to become mainstream for flagship PC CPUs and mobile application processors by 2025. Meanwhile, mid-range and high-end smartphone chips, AI GPUs, and ASICs are expected to remain on the 5nm and 4nm nodes, maintaining high utilisation rates for these processes. While demand for 7nm and 6nm processes has been sluggish in recent years, TrendForce anticipates renewed demand between the second half of 2025 and 2026, driven by the shift in RF/WiFi processes for smartphones. Collectively, the 7, 6, 5, 4, and 3nm processes are forecast to account for 45% of global wafer foundry revenues in 2025.
Demand for AI chips has also driven significant constraints in the supply of 2.5D advanced packaging in 2023 and 2024. Major manufacturers like TSMC, Samsung, and Intel, who offer both front-end manufacturing and back-end packaging solutions, have been actively expanding their capacities. TrendForce projects that revenues from 2.5D packaging solutions will surge by over 120% in 2025. Although this segment’s share of total wafer foundry revenue will remain below 5%, its strategic importance continues to grow.
Utilisation rates for mature processes to rise, but price pressure looms
Despite the positive outlook, low visibility in consumer product demand has led supply chain participants to maintain a conservative stance on inventory building. Wafer foundry orders are expected to remain on an ad-hoc basis in 2025, similar to 2024. However, with component inventories for automotive, industrial control, and general-purpose servers gradually returning to healthy levels, restocking is likely to pick up in 2025. This is expected to push utilisation rates for mature processes up by 10 percentage points, breaking the 70% threshold.
Wafer foundries are also expected to introduce previously delayed capacity expansions in 2025, particularly for 28, 40, and 55nm nodes. The combination of low demand visibility and the introduction of new capacity could result in downward pressure on prices for mature processes.
Challenges ahead despite growth projections
Although the wafer foundry market is expected to grow by 20% in 2025, buoyed by advancements in AI and the recovery of component inventories, several challenges remain. These include uncertainties around end-market demand due to broader macroeconomic conditions, potential cost-related barriers to AI deployment, and increased capital expenditures resulting from capacity expansion plans.
