Gartner’s report reveals that high-bandwidth memory (HBM) is gaining importance in AI technology, with revenues projected to reach $12.3 billion in 2024 and $21 billion in 2025. By 2026, over 40% of HBM chips are expected to support AI tasks, highlighting its growing role in advanced AI applications.
The global semiconductor market is on track for significant growth, with Gartner’s latest report predicting a 14% rise in revenue, reaching $717 billion by 2025. Published on October 28, 2024, the report indicates that after experiencing a recession in 2023, the industry has rebounded strongly, primarily driven by the rising demand for AI-related semiconductors and rise in electronics production. This year, semiconductor revenue is projected to hit $630 billion, reflecting an impressive 18.8% growth from the previous year. Despite this positive trend, the automotive and industrial sectors continue to struggle with weak demand, as noted by Rajeev Rajput, Senior Principal Analyst at Gartner.
In particular, the memory markets, especially DRAM and NAND are expected to see substantial growth that will significantly contribute to the overall revenue of the semiconductor sector. The memory segment is forecasted to expand by 20.5% in 2025, reaching $196.3 billion. The ongoing shortage in 2024 is anticipated to push NAND prices up by 60% this year, although a modest 3% price decline is expected in 2025 as supply stabilizes. Meanwhile, DRAM revenue is also set to increase due to rising demand for advanced technologies like double data rate 5 (DDR5) and high-bandwidth memory (HBM). This segment is expected to generate $115.6 billion in revenue in 2025, an increase from $90.1 billion in 2024.
Advancements in AI technologies, particularly with graphics processing units (GPUs), continue to reshape the semiconductor landscape. According to Gartner, revenue from GPUs used in AI model training is projected to grow to $51 billion, marking a 27% increase in 2025. The rising demand for HBM, vital for next-generation AI servers, has led to substantial investments in production. HBM revenue is expected to rise by 70% in 2025, totaling $21 billion. By 2026, more than 40% of HBM chips are anticipated to support AI inference workloads, highlighting the growing importance of semiconductors in facilitating advanced AI applications across various industries.
