Gigabyte CEO says NVIDIA should prioritize certain GPUs

Will the Jensen Huangthe famous “leather jacket”, will he listen to the suggestion? Gigabyte CEO says NVIDIA should prioritize GPUs with the highest return per GB of memory.

For those on the outside, here’s some context: NVIDIA’s strategy for manufacturing GeForce RTX 50 graphics cards in 2026 is under analysis.

After the CES 2026which did not present new GPUs to the general publicinformation circulated about the interruption of manufacturing of some models in the series, placing them in the final phase of their cycle.

The company provided a position on the subject, complemented by an exclusive interview with the CEO da Gigabyte, Eddie Linduring the event. The statements help you understand the current supply scenario.

How GDDR7 memory shortages influence production

A NVIDIA confirmedon note, that demand for its GPUs remains high, but faces limitations in memory availability.

The company stated that continues to distribute all models in the GeForce line and is in contact with its suppliers to increase the volume of GDDR7 chips.

Although the message indicates continuity, the conversation with Eddie Lin from Gigabyte reveals that more specific decisions about manufacturing priorities are being taken behind the scenes.

The executive explained that, given memory restrictions, the manufacturer can adopt a calculation method to define which products receive the most resources.

The logic is focus on segments that deliver the highest financial return for each gigabyte of VRAM used, reducing the manufacturing ratio for those with the lowest contribution per GB. In other words, the allocation follows an economic rationality based on the scarce resource.

The math behind choosing models

Eddie Lin detailed this reasoning with practical examples. One 8GB GPU retails for $300 generates approximately $37.5 revenue per GB. A similar model of US$ 400 contributes to US$ 50 by GB.

A product of US$ 500 com 16GB of VRAMin turn, drops to around US$ 32 of revenue per GBmaking it less interesting under this metric.

He described that the company canr 5 product segments and choose to concentrate efforts on the segments 1, 3 e 5while reducing the percentage of manufacturing in the segments 2 e 4precisely where the revenue contribution per gigabyte of memory is lower.

As a result, a recent series like the RTX 5070 Ti would not be completely discontinued, as rumors indicated; however, your presence on the shelves would be more unusual.

The final decision, according to Linpasses through a precise calculation of the value that each category delivers per gigabyte of VRAM consumed in production.

Revenue per gigabyte comparison table

A practical application of this reasoning can be seen in the table below, which lists the modelsyour memory, suggested price ea gross income generated per gigabyte of VRAM.

Which graphics cards may be prioritized in 2026

Analyzing the data, it is clear which models become more or less strategic. A RTX 5060 of 8 GBwith revenue from $47.38 per GBhas a great chance of being prioritized when compared with the common RTX 5060even with its initial lukewarm reception.

The 16 GB version of the RTX 5060 Ti, on the other hand, becomes the most vulnerable in the series, with the lowest return per GB: just US$26.81.

In the middle of the line, the RTX 5070 ea 5070 If have very similar contributions per gigabyte. This could lead NVIDIA to favor the model that is cheapest to producethe RTX 5070, or deprioritize both in favor of the more profitable RTX 5060 Ti 8GB.

No mole, RTX 5080 e 5090 have almost identical values ​​per GB of VRAM. The choice can fall on the RTX 5080which uses a smaller silicon chip and a less complex board design, allowing better use of memory stock.

For the consumer who has not yet built a PC, All of this can be considered bad newsbecause it means that the Infamous 8GB VRAM GPUs Won’t Die Anytime Soonand that anyone who wants to have a moderate amount of video memory will need to shell out even more money for a much more expensive product.

Providing memory to partners

One additional point addressed inddia Lin concerns the provision of memory. The executive stated that the Gigabyte continues to receive GDDR7 chips packaged directly with the GPUs and NVIDIA.

This practice protects board manufacturers of having to purchase the component on the open market, where prices and availability are much more volatile.

read more

Based on the information above, it is possible project availability scenario to 2026. Models like RTX 5060 Ti 8GBa RTX 5070 ea RTX 5080 they should be relatively easier to find in stores.

On the other hand, products with enthusiastic appeallike RTX 5060 Ti 16GB ea Rtx 5070 of thetend to face more limited supply.

If this scenario materializes, the situation can be considered as a partial confirmation of rumors from recent daysinvolving the Rtx 5070 of the. In fact, as the NVIDIA informed, the product will continue to be produced; however, it will be much more difficult (and expensive) for the average consumer to acquire it, just as ASUS suggested.

Finally, we emphasize that the situation of the best GPU in the world, the RTX 5090is already being reflected in the market, with empty shelves and prices significantly higher than suggestedindicating its reduced prioritization in the current context of restrictions.

And there? What did you think of the situation? Share your point of view in this post and continue following Adrenaline!

Source: VideoCardZ