FOR FREE MAGAZINE

AI is powering multi-year growth for memory semiconductor vendors

Artificial intelligence has lit a fire under the DRAM market, helping to dull the painful memories of recent losses.

Memory semiconductors have never been so sweet and profitable. Shipments of DRAMs, flash and other memory products spiked in 2024, rising nearly 80 percent, and are set to post another double-digit climb in 2025. The party is expected to continue in 2026. By the end of the year, memory shipments will have jumped to $214.8 billion, from $166 billion, in 2024, making it the second-largest segment of the global semiconductor market after logic ICs, according to the World Semiconductor Trade Statistics (WSTS), an industry association. The surge in memory sales is part of a resurgence in global semiconductor sales, said John Neuffer, president and CEO of the US Semiconductor Industry Association, which represents more than 90 percent of global chip vendors.

Micron Technology, Hynix and Samsung are experiencing a renaissance of sorts in memory revenue, and executives believe this upcycle has strong enough legs that they see it extending for several years if they can avoid overproduction. One of the most significant factors driving the market’s growth is the exponential increase in data creation. With billions of devices connected to the internet, applications like machine learning, predictive analytics, and real-time processing require advanced memory solutions. This is particularly evident in industries like healthcare, where large datasets are used for diagnostics, and in financial services.

The rise of artificial intelligence has also had a profound impact on the semiconductor memory market. AI models, especially generative AI and machine learning systems, demand high-speed and high-capacity memory solutions to process massive datasets efficiently. High bandwidth memory (HBM) has emerged as a vital technology in this field due to its ability to deliver extraordinary bandwidth while minimizing power consumption. “High Bandwidth Memory is crucial for the advancement of AI, enabling faster and more efficient training of machine learning models,” said Sanjay Mehrotra, CEO of Micron. “Our focus on HBM aligns with the growing demand for intelligent systems.”

Micron’s focus on high-performance computing and HBM has positioned it as a leader in the industry. Recently, the company disclosed an ambitious capital expenditure plan that will see it shelling out about $200 billion over years on new fabs. The justification for this expenditure lies in the growing demand for memory solutions in AI, edge computing, and autonomous vehicles. “Our significant investment reinforces Micron’s leadership in developing cutting-edge semiconductor memory technologies that empower industries worldwide,” Mehrotra said. Analysts also suggest that Micron’s focus on sustainability through low-power memory designs is a strategic move to capture long-term market share.

The development of next-generation memory technologies such as 3D NAND, MRAM, and HBM is pushing the boundaries of speed, capacity, and reliability. These innovations are critical for supporting high-performance computing and reducing energy consumption, which is a growing concern in sustainable technology. This trend is expected to significantly boost the demand for semiconductor memory. The transition to cloud-based infrastructures and edge computing platforms has similarly created a surge in demand for memory. Industry experts emphasize the importance of vigilance and adaptability. As James Wang, a semiconductor analyst at ARK Invest, remarked: “The semiconductor memory market is incredibly dynamic. Companies must remain agile, balancing innovation with prudent capacity planning to weather potential downcycles effectively.”

Towards $1 trillion

The global semiconductor market is currently experiencing robust growth, driven by high demand across multiple industries. Analysts have predicted IC sales will top $1 trillion by the end of this decade, a fact that has been buttressed by the ongoing recovery in the market. The WSTS now expects global chip sales to rise 11 percent in 2025, after a 19.7 percent jump in 2024. It sees sales surging by 2026 to $760.7 billion by 2026, from $630.5 billion, in 2024. This will put the market on a steady path to the projected $1 trillion, according to industry sources. “The global market continues to notch year-to-year growth driven by increasing sales into the Americas and Asia Pacific,” said Neuffer. “A new WSTS industry forecast calls for solid global market growth in 2025 driven by demand for AI, cloud infrastructure, and advanced consumer electronics.”

Memory chips are contributing strongly to the market’s growth. DRAM remains crucial for high-speed computing applications, while NAND Flash dominates the storage segment, powering everything from smartphones and laptops to large-scale data centers. Emerging technologies such as 3D XPoint, Magnetoresistive Random Access Memory (MRAM), and High Bandwidth Memory (HBM) are adding diversity to the market, promising faster speeds and greater efficiency.

Even so, any shocks to the supply chain or problems in the production cycle could derail this rosy outlook. The semiconductor memory market has long been characterized by pronounced pricing cycles, often driven by supply-demand dynamics, technological advancements, and macroeconomic factors. Historically, the market has exhibited periods of rapid growth followed by significant downturns. The last major downturn in the semiconductor memory market occurred in 2019, triggered by a combination of trade tensions between major economies, declining smartphone sales, and excess inventory. Prices for both DRAM and NAND Flash fell steeply during this period, impacting profitability across the industry. Analysts noted that the downturn served as a reminder of the cyclical nature of the market and underscored the importance of strategic inventory and capacity management.

While the semiconductor memory market is currently in a growth phase, the risk of another sudden downcycle cannot be ruled out. Factors that could precipitate a downturn include macroeconomic instability, geopolitical disruptions, and fluctuations in demand from key industries such as consumer electronics and automotive. Additionally, rapid advancements in manufacturing technology could lead to oversupply if demand fails to keep pace. During upcycles, robust demand leads to tight supply and higher prices, prompting manufacturers to increase production. However, this often creates oversupply, triggering downcycles marked by plummeting prices and shrinking profits.