I speak with a lot of procurement managers working for original equipment manufacturers (OEMs) and the contract manufacturers who support them about managing their electronics supply chains.
The conversations revolve around obsolescence management, cost control, and navigating disruptive events. The subject of memory has come up repeatedly over the past year.
Demand for memory has surged along with the adoption of artificial intelligence (AI). AI accelerators use high bandwidth memory (HBM) to train models and limits the number of companies that can make them. High demand and limited capacity are a perfect recipe for price increases and that is what we’ve been seeing.
A lot of new customers come to us looking for help navigating the situation. I share strategies using a simple Past, Present, and Future structure.
The Past
“Those who cannot remember the past are condemned to repeat it.”
George Santayana, The Life of Reason, 1905
It’s important to know how long the situation has been going on. The memory market is historically cyclical. Advancements are made. New technology is adopted. Older technology is discontinued. By tracking the price, lead time, and inventory of a product family, you will you will be better able to estimate where it is in the cycle.
Double data rate 4 (DDR4) was supposed to give way to DDR5 much sooner, but a shifting market caused by the pandemic prevented it from happening. When DDR5 arrived in 2020, it required new supporting components whose short supply was made worse by COVID-related production disruptions. These conditions kept the price of DDR5 high, decreasing its adoption rate.
The Present
Today, we have a new market force impacting the memory market. The adoption of AI, with its monstrous appetite for data, has dramatically increased the demand for memory. Those consuming it, including Google, Microsoft, Amazon, and others, all have large budgets, which only intensifies the problem. Memory makers have focused on the higher margin AI products which has had a ripple effect on other areas of the market.
As we move from training to inference, agentic AI will increase the demand for NAND flash memory to manage the massive context, long-term memory, and state persistence required for autonomous, multi-step workflows. We are seeing increasing demand for NAND.
The Future
It’s difficult to predict the future with certainty, but we can certainly see where things are going. For me, it comes down to two things: innovation and money. Innovation is almost always tied to money, but it’s not a 1:1 correlation because not everything works out. Remember 3D televisions?
- AI is moving to the edge. Sensors from automobiles and robots will feed information that must be analyzed and acted upon locally.
- Enterprises will harness the power of analyzing data quickly to make business decisions.
- New manufacturing will spring up to address the need. China’s Yangtze Memory Technologies plans to build two fabs beyond the Phase 3 plant it’s due to complete in Wuhan this year, a move that would more than double its current wafer output.
- The demand from OEMs will go down as price elasticity is tested.
- Designs will be respun to incorporate the latest technology.
How SiliconExpert Can Help
SiliconExpert’s market-leading electronics data includes decades of price, lead time, and inventory history using direct feeds from suppliers. Our proprietary Years to End of Life algorithm feeds on that data, using machine learning to improve predictions year over year. Our Market Update Reports tell you what we are seeing and hearing in the marketplace for various product lines.
Our Supply Chain Risk Management tool will tell you when things change, which of your parts are potentially impacted, and which of your products contain those parts. We can identify potential risk vectors and possible alternatives that will keep production lines running and revenue accumulating.
Plan for the Memory Market Before Risk Reaches Production
The memory market is changing again. AI demand, technology transitions, pricing pressure, and capacity constraints are creating new planning challenges for OEMs and contract manufacturers. Join SiliconExpert for a practical discussion on what supply chain teams should watch and how to prepare for sustained memory-market pressure.
