Gigabyte's CEO reveals Nvidia's strategy to allocate scarce GDDR7 memory to GPUs generating the highest revenue per gigabyte, potentially reshaping RTX 50-series availability.
Nvidia is implementing a calculated allocation strategy for its RTX 50-series GPUs amid severe GDDR7 memory shortages, prioritizing models that generate the highest gross revenue per gigabyte of memory according to insights shared by Gigabyte CEO Eddie Lin. This approach comes as Nvidia confirmed "memory supply is constrained" for GeForce GPUs while denying rumors about discontinued models.
Lin explained Nvidia's methodology during a CES interview: "They calculate how much revenue each segment contributes per gigabyte of memory." This metric determines production priorities across five GPU segments. Using this formula (MSRP ÷ VRAM capacity), entry-level cards face disproportionate challenges.
The revenue-per-GB calculation reveals significant disparities across Nvidia's product stack:
| Model | Memory (GB) | MSRP | Revenue/GB | GPU Die |
|---|---|---|---|---|
| RTX 5060 | 8 | $299 | $37.38 | GB206 |
| RTX 5060 Ti 8GB | 8 | $379 | $47.38 | GB206 |
| RTX 5060 Ti 16GB | 16 | $429 | $26.81 | GB206 |
| RTX 5070 | 12 | $549 | $45.75 | GB205 |
| RTX 5070 Ti | 16 | $749 | $46.81 | GB203 |
| RTX 5080 | 16 | $999 | $62.44 | GB203 |
| RTX 5090 | 32 | $1,999 | $62.47 | GB202 |
| RTX Pro 6000 Blackwell | 96 | $8,500 | $88.54 | GB202 |
Three key implications emerge from this data:
Mid-Range Shakeup: The RTX 5060 Ti 8GB ($47.38/GB) becomes unexpectedly prioritized over the standard RTX 5060 ($37.38/GB) despite similar memory configurations, while the 16GB variant ($26.81/GB) faces severe production constraints.
High-End Tradeoffs: Though the RTX 5090 and RTX 5080 generate nearly identical revenue per GB ($62.47 vs $62.44), Nvidia likely favors the 5080 due to its smaller GB203 die, simpler PCB design, and ability to produce two 5080s from the memory allocation of one 5090.
Professional Advantage: The RTX Pro 6000 Blackwell's $88.54/GB revenue explains why Nvidia skipped consumer Super refreshes. Using 3GB GDDR7 modules in clamshell configuration yields 41% higher revenue per GB than the RTX 5090, making professional GPUs more attractive amid shortages.
Market availability will reflect these priorities. The RTX 5060 Ti 8GB, RTX 5070, and RTX 5080 should see relatively stable supply, while the RTX 5060 Ti 16GB and RTX 5070 Ti face significant scarcity. RTX 5090 availability appears particularly threatened, with current market conditions showing depleted inventories and inflated third-party pricing.
This strategy represents Nvidia's effort to maximize profitability during an industry-wide memory crunch. As Lin noted, Nvidia continues bundling memory with partner shipments, contradicting rumors about direct memory purchases. The company's focus on revenue efficiency per memory unit may reshape GPU hierarchies throughout 2026, potentially accelerating the transition to next-generation memory technologies.
Jeffrey Kampman is Senior Analyst, Graphics at Tom's Hardware, covering GPU architectures and market dynamics.
Comments
Please log in or register to join the discussion