According to the latest data, the outlook for the foundry market is positive.TrendForce expects the foundry market to recover in 2025, with a projected annual growth rate of 20%, up from 16% in 2024.
This positive outlook comes despite weak end-market demand for consumer products, which has led component manufacturers to adopt conservative stocking strategies, causing average foundry capacity utilization to fall below 80% in 2024.
Only advanced processes such as 5/4/3nm nodes for HPC products and flagship smartphones are able to maintain full capacity, and this is expected to continue through 2025. However, visibility into the consumer end market in 2025 remains low. These advanced technologies are not only enhancing efficiency but also setting new trends in the industry.
The automotive and industrial control supply chain, on the other hand, began to recover from inventory adjustments in the second half of 2024, and is expected to gradually resume replenishment in 2025. Coupled with the increase in per-wafer consumption driven by edge AI, and the continued expansion of cloud AI infrastructure, it is estimated that these factors will drive the foundry market output value to grow at an annualized rate of 20% in 2025.
TrendForce forecasts that advanced processes and packaging will drive TSMC’s 2025 revenue growth beyond the industry average. Although the growth momentum of non-TSMC foundries is still constrained by consumer end-use demand, revenue growth is expected to reach nearly 12 percent in 2025, outpacing the previous year. This is driven by ample parts inventory across industries for IDM and fabless customers, power demand driven by cloud/edge AI, and a lower base in 2024.
According to TrendForce:
- Over the past two years, 3nm process capacity has entered a scale-up phase and is expected to become the mainstream for flagship PC CPUs and mobile APs by 2025;
- Mid-to-high-end smartphone chips, AI GPUs, and ASICs are still at the 5/4nm node, so the utilization rate of these processes is expected to remain high;
- While demand for 7/6nm processes has been weak over the past two years, new demand is expected to emerge between H2 2025 and 2026, driven by smartphone RF/WiFi process transition programs.
TrendForce predicts that by 2025, 7/6nm, 5/4nm and 3nm processes will contribute 45% of global revenue for foundries.
In addition, driven by the strong demand for AI chips, the supply of 2.5D advanced packaging in 2023 and 2024 is significantly limited, TSMC, Samsung, Intel and other major vendors to provide the integration of front-end process and back-end packaging solutions are actively expanding production capacity, TrendForce estimates that in 2025 the revenue growth rate of foundry 2.5D packaging solutions will reach more than 120%, accounting for the wafer foundry Although the proportion of total revenue is still less than 5%, but the importance continues to increase.
Mature process utilization rate is expected to increase by 10%
TrendForce pointed out that due to the unpredictability of demand for consumer products, supply chain participants will tend to be conservative in their attitude toward inventory building, and foundry orders in 2025 are expected to be similar to those in 2024, maintaining the ad-hoc (temporary flexible) model.
However, as inventories of automotive, industrial control, and general-purpose server components gradually replenish to a healthy level in 2024, replenishment is expected to resume in 2025, and the capacity utilization rate of mature processes will increase by 10 percentage points to exceed the 70% mark.
After postponing capacity expansion plans for two consecutive years, foundries are also expected to start introducing previously postponed new capacity in 2025, especially in 28nm, 40nm and 55nm modes. Low demand visibility and the influx of new capacity could put additional downward pressure on mature process prices.
While foundry revenues are expected to grow 20% in 2025, thanks to growing AI and bottoming out of application component inventories, foundries will still face several challenges. These challenges include uncertainty in end-market demand due to macroeconomic factors, the potential impact of high costs on the intensity of AI deployments, and increased capital expenditures due to capacity expansion plans.