Micron Bets on GDDR Stacking Tech First, Quietly Igniting a New Race in the AI Memory Market

Micron Technology is seeking to carve out a new track in AI memory— by vertically stacking Graphics Double Data Rate (GDDR) memory like High Bandwidth Memory (HBM) to fill the market gap between HBM and standard GDDR.

According to Korean media outlet ETNews, Micron has initiated the development of new vertically stacked GDDR products and plans to deploy related equipment and enter the process testing phase in the second half of this year.

The initial plan is to achieve approximately four layers of GDDR stacking, with samples potentially launching as early as next year. This strategic move precedes Samsung Electronics and SK Hynix, positioning Micron to seize an early advantage in this emerging niche market.

The backdrop for this initiative is the continuous expansion of AI application scenarios driving accelerated differentiation in memory demand. GDDR, with its cost advantages, is increasingly being adopted in AI inference accelerators. However, its bandwidth bottleneck limits further penetration. Enhancing GDDR performance through stacking technology could open new growth avenues for this market.

GDDR Stacking: Positioning in the Price Range Between HBM and Standard Memory

GDDR is memory optimized for video processing and 3D graphics rendering, primarily used in graphics cards and gaming devices. Compared to HBM, GDDR has lower bandwidth but is more cost-competitive. In recent years, it has begun to be incorporated into some AI accelerators.

The stacked GDDR product being developed by Micron aims to create a new product with performance between HBM and standard GDDR—offering superior bandwidth to existing GDDR but at a lower cost than HBM. The product specifications are currently being discussed in alignment with the needs of customers such as AI accelerator manufacturers, clearly targeting custom demand.

Analysts believe that stacked GDDR will occupy the intermediate market between HBM and non-stacked GDDR, and is also expected to see considerable demand in the continuously growing high-performance gaming graphics card sector. The expanding scale of the AI market is driving demand for diversified products—Nvidia's use of SRAM in its inference-specific chips is a prime example of this trend.

Technical Barriers and Cost Control are Key to Mass Production

Despite market interest, GDDR stacking technology is still in its early stages, having previously only appeared in academic papers and cutting-edge technical research, with no prior mass production examples.

The industry points out that Micron faces several technical challenges, including the stacking interconnection methods between GDDR chips, power management, and heat dissipation control. On the cost front, the additional manufacturing expenses introduced by the stacking process are also significant—if a sufficient cost-performance advantage over HBM cannot be maintained, the product's market competitiveness will be greatly diminished.

An industry insider commented:

"In the past, besides technical issues, the ambiguous market positioning of GDDR stacking was a major reason for its delayed commercialization. With the evolution of the AI memory market landscape, the necessity of GDDR stacking has begun to become apparent. Micron's attempt this time is expected to be the starting point of a new round of memory stacking competition."

Outpacing Samsung and SK Hynix, Strategizing for an Emerging Niche

Micron's entry into GDDR stacking has a clear first-mover strategic intent. Samsung Electronics and SK Hynix have not yet publicly announced similar plans. If Micron can achieve technological breakthroughs and commercialization first, it will establish a competitive barrier in this niche segment.

In terms of market size, stacked GDDR currently remains a niche market. However, with the continuous popularization of AI inference applications and the growth of demand for diverse accelerator hardware, its potential cannot be underestimated.

Micron has clearly factored this judgment into its product strategy, choosing to make its move early while the market landscape is still uncertain.