"Memory Semiconductor Revenue to Reach 805 Trillion Won This Year...Expected to More Than Double Foundries"

TrendForce Survey
Memory Revenue Up 134% Year-on-Year

Both the memory semiconductor and foundry (contract semiconductor manufacturing) markets are projected to post all-time high sales this year. In particular, the memory market is expected to scale up to more than twice the size of the foundry market.

According to market research firm TrendForce on February 10, this year’s global memory market revenue is estimated at 551.6 billion dollars (about 805 trillion won), up 134% from a year earlier. The analysis is that the market will peak in size as prices of memory semiconductors, whose supply has tightened amid the boom in artificial intelligence (AI) semiconductors, rise sharply.

Over the same period, foundry market revenue is also expected to grow 25% to 218.7 billion dollars (about 319 trillion won), hitting a record high, but in terms of scale it will fall short of half the memory market. Foundries, by their nature, rely on long-term contracts, so their price volatility is lower than in the memory market.

TrendForce noted that the current AI semiconductor super cycle is showing far stronger demand recovery and pricing power than in 2017. In the past, finished-product manufacturers such as home appliance and smartphone makers led the market, but now cloud service providers (CSPs), represented by big tech (large information technology companies), have taken the wheel. They have a strong tendency to secure volume even at high prices, which is driving a much steeper price increase than in past boom cycles.

As reasons why revenue growth in the foundry market is more moderate than in memory, the firm cited high technological barriers and the limited capacity expansion of suppliers. TrendForce explained, “While legacy node processes account for about 70–80% of total foundry capacity, advanced node processes make up only 20–30%,” adding, “This means that despite their high prices, advanced processes contribute a relatively small share of total revenue.” In addition, the fact that standardized memory products are more flexible than foundries when it comes to expanding production capacity (CAPA) is also cited as a cause of the revenue gap.

This content was produced with the assistance of AI translation services.