
NVIDIA Forced by Shortages: Rubin Ultra Chip Considering Reducing HBM Configuration, 2027 DRAM Shortage Not Over Yet
Author: TrendForce
Compiled by: TechFlow
TechFlow Editor's Note: Continued tight memory supply is changing the design logic of AI chips. Nvidia began evaluating reduced HBM configurations for its next-generation Rubin Ultra chips starting Q3 2026, expanding from the original plan of 12-layer HBM4e to include various options such as 8-layer HBM4e, 12-layer HBM4, and 8-layer HBM4. This is not a technical choice, but a compromise forced by the supply chain—DRAM shortages will last until 2027, the mass production schedule for HBM4e remains uncertain, and AI chip manufacturers must trade off between performance and shipment volume.
According to TrendForce's latest memory industry research, the tight DRAM supply situation will continue until 2027, and the HBM4e validation timeline for memory suppliers remains uncertain. These limiting factors are prompting AI chip manufacturers to cut HBM configurations for next-generation products.
Since Q3 2026, Nvidia has begun expanding the scope of evaluation for Rubin Ultra chip HBM configurations, extending from the initial 12-layer HBM4e design to include alternative solutions such as 8-layer HBM4e, 12-layer HBM4, and 8-layer HBM4. Final specifications have not yet been determined. In addition to Nvidia, several cloud service providers are also evaluating reducing HBM capacity for their next-generation self-developed AI ASIC chips.
TrendForce pointed out that recent memory shortages have led to multiple adjustments in AI chip memory specifications. For example, cloud service providers and server OEMs reduced RDIMM capacity in server configurations in the first half of 2026. Recently, after judging that LPDDR5X supply constraints might last until 2027, Nvidia decided to halve the SOCAMM capacity of the next-generation Vera Rubin super chip module.
From 2025 to the first half of 2026, Nvidia consistently used 12-layer HBM4e as the baseline design for Rubin Ultra. However, starting from the beginning of Q3 2026, the company began evaluating lower-specification alternative solutions. This change is primarily driven by two supply-side limiting factors: first, the expected overall DRAM shortage in 2027 will limit the wafer capacity memory suppliers can allocate to HBM production. Second, the validation timeline and capacity ramp-up for 12-layer HBM4e remain uncertain.
TrendForce pointed out that Nvidia's primary goal for this generation of Rubin Ultra products is to increase I/O speed, while expanding GPU shipment volume remains a secondary priority. If Nvidia ultimately decides to cut HBM specifications, it is expected to be achieved by reducing the number of DRAM stack layers.
Ultimately, whether HBM4e can complete validation and enter mass production as planned will determine whether Rubin Ultra's I/O speed can increase from the previous generation Rubin's 8-11.7 Gbps to 14-16 Gbps, or if it can only reach 11-12 Gbps through HBM4 design optimization.
At the same time, within the same generation of products, the number of DRAM stack layers determines the trade-off between single GPU HBM capacity and the number of shippable GPUs.
TrendForce believes the final configuration will also depend on memory suppliers' wafer allocation decisions. From a supply and demand perspective, HBM bit shipments in 2027 are expected to increase by 50-60% year-over-year, but still not enough to keep up with demand growth.
TrendForce expects that under such supply-constrained conditions, HBM suppliers will maintain pricing power throughout 2027, and the industry generally expects HBM prices to rise significantly. Therefore, AI chip manufacturers will face the dual challenge of limited HBM supply and rising procurement costs, which increases the motivation to adopt lower capacity HBM configurations.















