Yet the Digital Random Access Memory (DRAM) market is facing historic headwinds. Demand is through the roof, particularly from new artificial intelligence (AI) data centers. This rapidly growing demand absorbs supply for other uses, leading to prices climbing to all-time highs. With the global economy still recovering from the effects of the COVID-19 pandemic, which exacerbated chip supply issues, experts are keenly observing how the industry will respond. The current situation reflects a classic boom-and-bust cycle in the DRAM industry, with potential solutions hinging on innovation and infrastructure investment.
Additionally, organizations are in a hurry to expand their data center footprint. Internationally, almost 2,000 additional data centers are in the planning or construction phase. These increasingly complex facilities are becoming more dependent on DRAM to drive graphics processing units (GPUs) and other accelerators that have become critical to AI workloads. That complexity is beautifully illustrated by the distinct architecture of DRAM technology. My 750-micrometer thick DRAM tower stacks on a common base logic die, allowing the highest possible speed and lowest latency communication between memory dies and processors. This novel innovation is key for unparalleled performance in high-bandwidth memory (HBM) applications.
The Role of DRAM in AI Data Centers
As DRAM continues to play an important role in powering GPUs and other accelerators that are mainstream for AI processing, this increase in demand for DRAM has shifted supplies from different applications. The result has been a racketeering scheme that’s jacked up prices by 14 times.
New digital HBM technology breakthroughs have demonstrated DRAM’s ability to keep up with increasing demands. The B300 model just released uses eight of the HBM chips, each with 12 DRAM dies stacked atop one another. With the continuing advancement of standards such as HBM4, 16 stacked DRAM dies are permissible. This remarkable progress underscores the new and increasing performance expectations in the emerging AI space.
Relief will come from a combination of incremental capacity expansions by existing DRAM leaders, yield improvements in advanced packaging, and a broader diversification of supply chains,” explained Shawn DuBravac, providing insight into how the industry could navigate these challenges.
Supply Chain Challenges and Future Developments
Even the boom and bust historic nature of the DRAM industry has undergone new tectonic shifts. The COVID-19 pandemic led to a global chip supply crisis that worsened these problems. The market-wide ramifications of this crisis are still echoing throughout the marketplace as firms rush to pivot.
There are primarily two strategies to address DRAM supply challenges: innovation and building new fabrication facilities (fabs). With building a new fab now costing more than $15 billion and taking at least 18 months to ramp up, that’s not a viable answer. Production at a new Samsung plant in Pyeongtaek, South Korea is expected to start rolling in 2028. This announcement is an encouraging indicator of their serious commitment to addressing supply challenges in the longer term.
There are two ways to address supply issues with DRAM: with innovation or with building more fabs,” stated Mina Kim, emphasizing the urgency for solutions within this critical sector.
Economic Implications and Market Forecast
It’s been noticed by economists since at least the joke above that prices tend to fall faster than they rise. This trend is probably true for today’s DRAM market. “In general, economists find that prices come down much more slowly and reluctantly than they go up. DRAM today is unlikely to be an exception to this general observation, especially given the insatiable demand for compute,” Kim remarked.
The dynamic between demand and supply still drives the key market predictions. The world is changing to one that is driven by data. As a consequence, pursuit of leading-edge memory innovations such as DRAM will increase dramatically. The anticipated expansion of global data centers will likely continue to exert pressure on supply chains while driving innovation within the industry.